Images

Classifications

• • F24F11/0034—
• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
  • G06N7/00—Computing arrangements based on specific mathematical models
  • G06N7/02—Computing arrangements based on specific mathematical models using fuzzy logic
  • G06N7/06—Simulation on general purpose computers
• • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
  • F24—HEATING; RANGES; VENTILATING
  • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
  • F24F11/00—Control or safety arrangements
  • F24F11/30—Control or safety arrangements for purposes related to the operation of the system, e.g. for safety or monitoring
• • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
  • F24—HEATING; RANGES; VENTILATING
  • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
  • F24F11/00—Control or safety arrangements
  • F24F11/62—Control or safety arrangements characterised by the type of control or by internal processing, e.g. using fuzzy logic, adaptive control or estimation of values
• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F7/00—Methods or arrangements for processing data by operating upon the order or content of the data handled
  • G06F7/02—Comparing digital values
  • G06F7/023—Comparing digital values adaptive, e.g. self learning
• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
  • G06N20/00—Machine learning
• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
  • G06N5/00—Computing arrangements using knowledge-based models
  • G06N5/04—Inference or reasoning models
  • G06N5/048—Fuzzy inferencing
• • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
  • F24—HEATING; RANGES; VENTILATING
  • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
  • F24F2120/00—Control inputs relating to users or occupants
  • F24F2120/10—Occupancy
• • G—PHYSICS
  • G05—CONTROLLING; REGULATING
  • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
  • G05B2219/00—Program-control systems
  • G05B2219/20—Pc systems
  • G05B2219/26—Pc applications
  • G05B2219/2642—Domotique, domestic, home control, automation, smart house

Definitions

• the present invention relates to the field of programmable sequencing devices, or, more particularly, the field of remote controls for consumer electronic devices.
• the present invention provides an enhanced interface for facilitating human input of a desired control sequence in a programmable device by employing specialized visual feedback.
• Some of the most popular programmable products include VCRs, answering machines, microwave ovens, alarm clocks, thermostats, cameras, home security systems, lighting systems, and automobiles.
• VCRs video cassette recorders
• VCRs employ a bar-code reader in order to allow entry of programming steps from a menu of functions, or from an encoded description of an event to be programmed.
• this method suffers from the limitation that the channel, time and duration must be available in encoded form, otherwise the use of the device will not simplify the use or programming of the VCR.
• These machines come with a laminated sheet of bar codes.
• the user In order to program the VCR, the user must press a button on a wand, which lights its tip, and then run or pass the tip over a bar-code, to set each step separately. Finally, when all the information has been scanned in, the user must press the “Transmit” button.
• VCRplus+ is a device which allows the entry of a code representing a channel, time, date and duration of a program to be recorded, which when entered into the remote control device, is translated into commands for programming the VCR, and transmitted through an infrared link to the VCR, thus programming the VCR.
• This system has the limitations that the published codes must be available, and manually entered, which may be thus be erroneously entered, and the system does not allow for rescheduled programs, so that any variation in schedule will result in a defective recording.
• the time and date in the VCR device must also be set accurately for this system to operate.
• On-screen programming systems exist; however, these generally require the user to scroll through menus and option choices without allowing direct entry of programming information.
• Direct-entry systems are available with, for example, programmable controllers with keypad entry. However, these do not generally have full information visual displays, meaning that all vital information is not or cannot be simultaneously displayed, and must be “multiplexed”, meaning that data must share display space with other data, displayed at different times.
• VCR with on-screen programming all programming information feedback is displayed on the television screen, and prompts are provided to guide the user through the necessary steps.
• Some VCRs have numeric keypads to enter the information, while others allow choices to be entered by the selection method, which depends on the use of “up” and “down” arrow keys to select a desired option.
• information may be entered on the display panel using the selection method, with either the “up” key or both “up” and “down” keys, or by direct entry in devices that support such a system.
• the remote control device of a VCR is often the primary input device, and it sometimes has control functions not accessible from a keypad input present on the VCR itself.
• Remote controls often contain many buttons, which may be found overwhelming and confusing by the user. This results in under-utilization of the various actuators or buttons, and consequently, various useful features are unused or inaccessible, or the programming operation is inefficient.
• the extra clutter results in a greater “search time”, the time needed to locate and execute a desired function, and thus it takes longer to program the VCR.
• the general structure of the search time in programming a VCR is shown diagrammatically in FIG. 1 . Other problems arise from the layout and coding of the buttons.
• VCRs Many techniques have been used to facilitate the programming of devices such as VCRs, including:
• the present invention is directed toward improving the interface to allow all segments of the population to effectively interface with these programmable devices.
• the present method and apparatus allow a manufacturer to produce a single device, without regard to the ability of the user to learn the programming steps.
• U.S. Pat. No. 5,068,664 incorporated herein by reference, discloses a method and device for recognizing a target among a plurality of known targets, by using a probability based recognition system.
• This patent document cites a number of other references, each incorporated herein by reference, which are relevant to the problem of image recognition: Vannicola et al, “Applications of Knowledge based Systems to Surveillance”, Proceedings of the 1988 IEEE National Radar Conference, 20-21 Apr. 1988, pp. 157-164; Ksienski et al., “Low Frequency Approach to Target Identification”, Proc.
• the image or object recognition feature of the present invention may be implemented in the manner of U.S. Pat. No. 5,068,664. Further, it is clear that this recognition feature may form an integral part of certain embodiments of the present invention. It is also clear that the various features of the present invention would be applicable as an adjunct to the various elements of the system disclosed in U.S. Pat. No. 5,068,664.
• U.S. Pat. No. 5,065,447 cites a number of references, all incorporated herein by reference, relevant to the use of fractals in image processing: U.S. Pat. No. 4,831,659; Barnsley et al., “Hidden Variable Fractal Interpolation Functions”, School of Mathematics, Georgia Institute of Technology, Atlanta, Ga. 30332, July, 1986; Barnsley, M. F., and Demko, S., “Iterated Function Systems and The Global Construction of Fractals”, Proc. R. Soc.
• U.S. Pat. No. 5,063,603, incorporated herein by reference, relates to a dynamic method for recognizing objects and image processing system therefor.
• This reference discloses a method of distinguishing between different members of a class of images, such as human beings.
• a time series of successive relatively high-resolution frames of image data any frame of which may or may not include a graphical representation of one or more predetermined specific members (e.g., particular known persons) of a given generic class (e.g. human beings), is examined in order to recognize the identity of a specific member; if that member's image is included in the time series.
• the frames of image data may be examined in real time at various resolutions, starting with a relatively low resolution, to detect whether some earlier-occurring frame includes any of a group of image features possessed by an image of a member of the given class.
• the image location of a detected image feature is stored and then used in a later-occurring, higher resolution frame to direct the examination only to the image region of the stored location in order to (1) verify the detection of the aforesaid image feature, and (2) detect one or more other of the group of image features, if any is present in that image region of the frame being examined.
• the accumulated detected features can first reliably recognize the detected image region to be an image of a generic object of the given class, and later can reliably recognize the detected image region to be an image of a certain specific member of the given class.
• the personae recognition feature of the present invention may be implemented in this manner. Further, it is clear that this recognition feature may form an integral part of certain embodiments of the present invention. It is also clear that the various features of the present invention would be applicable as an adjunct to the various elements of the system disclosed in U.S. Pat. No. 5,063,603.
• U.S. Pat. No. 5,055,658, incorporated herein by reference, relates to a security system employing digitized personal characteristics, such as voice.
• the following cited references are incorporated herein by reference: Naik et al., “High Performance Speaker Verification . . . ”, ICASSP 86, Tokyo, CH2243-4/86/0000-0881, IEEE 1986, pp. 881-884; “Voice Recognition and Speech Processing”, Elektor Electronics, September 1985, pp. 56-57; Shinan et al., “The Effects of Voice Disguise . . . ”, ICASSP 86, Tokyo, CH2243-4/86/0000-0885, IEEE 1986, pp. 885-888.
• Parts of this system relating to speaker recognition may be used to implement a voice recognition system of the present invention for determining an actor or performer in a broadcast.
• U.S. Pat. No. 5,067,164 incorporated herein by reference, relates to a hierarchical constrained automatic learning neural network for character recognition, and thus represents an example of a trainable neural network for pattern recognition, which discloses methods which are useful for the present invention.
• This patent cites various references of interest, which are incorporated herein by reference: U.S. Pat. Nos. 4,760,604, 4,774,677 and 4,897,811; Rumelhart, D. E., et al., Parallel Distr. Proc.: Explorations in Microstructure of Cognition, vol. 1, 1986, “Learning Internal Representations by Error Propagation”, pp. 318-362; Lippmann, R.
• U.S. Pat. No. 5,058,180 incorporated herein by reference, relates to neural network apparatus and method for pattern recognition, and is thus relevant to the intelligent pattern recognition functions of the present invention.
• This patent cites the following documents of interest, which are incorporated herein by reference: U.S. Pat. Nos. 4,876,731 and 4,914,708; Computer Visions, Graphics, and Image Processing 1987, 37:54-115; Jackel, L. D., H. P. Graf, J. S. Denker, D. Henderson and I. Guyon, “An Application of Neural Net Chips: Handwritten Digit Recognition,” ICNN Proceeding, 1988, pp. II-107-15; Carpenter, G. A., S.
• U.S. Pat. No. 5,067,161 incorporated herein by reference, relates to a video image pattern recognition system, which recognizes objects in near real time.
• U.S. Pat. Nos. 4,817,176 and 4,802,230 both incorporated herein by reference, relate to harmonic transform methods of pattern matching of an undetermined pattern to known patterns, and are useful in the pattern recognition method of the present invention.
• U.S. Pat. No. 4,998,286, incorporated herein by reference relates to a harmonic transform method for comparing multidimensional images, such as color images, and is useful in the present pattern recognition methods.
• U.S. Pat. No. 5,060,282 incorporated herein by reference, relates to an optical pattern recognition architecture implementing the mean-square error correlation algorithm. This method allows an optical computing function to perform pattern recognition functions.
• U.S. Pat. No. 5,060,282 cites the following references, incorporated herein by reference, which are relevant to optical pattern recognition: is Psaltis, D., “Incoherent Electro-Optic Image Correlator”, Optical Engineering, 23(1):12-15 (January/February 1984); Kellman, P., “Time Integrating Optical Signal Processing”, Ph. D. Dissertation, Stanford University, 1979, pp.
• U.S. Pat. No. 5,067,160 incorporated herein by reference, relates to a motion-pattern recognition apparatus.
• the apparatus recognizes a motion of an object which is moving and is hidden in an image signal, and discriminates the object from the background within the signal.
• the apparatus has an image forming unit comprising non-linear oscillators, which forms an image of the motion of the object in accordance with an adjacent-mutual-interference-rule, on the basis of the image signal.
• a memory unit comprising non-linear oscillators, stores conceptualized meanings of several motions.
• a retrieval unit retrieves a conceptualized meaning close to the motion image of the object.
• An altering unit alters the rule, on the basis of the conceptualized meaning.
• the image forming unit, memory unit, retrieval unit and altering unit form a holonic-loop. Successive alterations of the rules by the altering unit within the holonic loop change an ambiguous image formed in the image forming unit into a distinct image.
• U.S. Pat. No. 5,067,160 cites the following references, incorporated herein by reference, which are relevant to the task of discriminating a moving object in a background: U.S. Pat. No.
• U.S. Pat. No. 5,067,160 discloses an adaptive pattern recognition system that may be useful in various embodiments of the present invention. It is also clear that the interface and control systems of the present invention provide useful adjuncts to the elements disclosed in U.S. Pat. No. 5,067,160.
• U.S. Pat. No. 5,065,440 incorporated herein by reference, relates to a pattern recognition apparatus, which compensates for, and is thus insensitive to pattern shifting, thus being useful for decomposing an image into its structural features and recognizing the features.
• U.S. Pat. No. 5,065,440 cites the following references, incorporated herein by reference, which are also relevant to the present invention: U.S. Pat. Nos. 4,543,660, 4,630,308, 4,677,680, 4,809,341, 4,864,629, 4,872,024 and 4,905,296.
• U.S. Pat. No. 5,067,166 incorporated herein by reference, relates to a pattern recognition system, in which a local optimum match between subsets of candidate reference label sequences and candidate templates. It is clear that this method is useful in the pattern recognition aspects of the present invention. It is also clear that the interface and control system of the present invention are useful adjuncts to the method disclosed in U.S. Pat. No. 5,067,166.
• U.S. Pat. No. 5,048,095 incorporated herein by reference, relates to the use of a genetic learning algorithm to adaptively segment images, which is an initial stage in image recognition.
• This patent has a software listing for this method. It is clear that this method is useful in the pattern recognition aspects of the present invention. It is also clear that the interface and control system of the present invention are useful adjuncts to the method disclosed in U.S. Pat. No. 5,048,095.
• Neural Networks 1:I525-I532 (July 1988), San Diego, Calif.; Wald, Sequential Analysis, Dover Publications Inc., 1947, pp. 34-43; Rosenfeld, Azriel and Avinash C. Kak, Digital Picture Processing, Second Edition, Volume 2, Academic Press, 1982; Mori, “Towards the construction of a large-scale neural network”, Electronics Information Communications Association Bulletin PRU 88-59, pp. 87-94; Yamada et. al., “Character recognition system using a neural network”, Electronics Information Communications Association Bulletin PRU 88-58, pp.
• the interface of the present invention serves to minimize the learning and searching times, better reflect users' expectations, provide better matching to human memory limits, be usable by both novices and experienced users, reduce intimidation of novice users by the device, and simplify the entering of programming data.
• the present invention optimizes the input scheme for programming an event-driven device, and can also be applied to many types of programmable devices. Thus, certain human factors design concepts, heretofore unexploited in the design of consumer electronics devices and industrial controls, have been incorporated.
• One aspect of the present invention relates to a programmable device that comprises a menu-driven interface in which the user enters information using a direct manipulation input device.
• a type of interface scheme is disclosed in Verplank, William L., “Graphics in Human-Computer Communication: Principles of Graphical User-Interface Design”, Xerox Office Systems, which is incorporated herein by reference; the references cited therein: Foley, J. D., Wallace, V. L., Chan, P., “The Human Factor of Computer Graphics Interaction Techniques”, IEEE CG&A, November 1984, pp. 13-48; Koch, H., “Ergonomische Wegung von Schreibtastaturen”, Humane Production, 1, pp. 12-15 (1985); Norman, D.
• the apparatus typically involves a remote control entry device, and the interface of the present invention contains a displayed graphical interface for programming programmable devices.
• the present invention seeks more accurate programming through the use of program verification to ensure that the input program is both valid and executable. Thus, it has a mechanism to store and check to verify that there are no conflicting programs.
• An apparatus according to the present invention can be connected, for example, to any infrared programmable device in order to simplify the programming process.
• an improved video cassette recorder (VCR) interface forms the basis of a disclosed is example.
• any programmable controller i.e., any device which monitors an event or sensor and causes an event when certain conditions or parameters are met, and may also be used in other programming environments, which are not event driven.
• a directional sensor based infrared remote control is disclosed in Zeisel, Tomas, Tomaszewski, “An Interactive Menu-Driven Remote Control Unit for TV-Receivers and VC-Recorders”, IEEE Transactions on Consumer Electronics, Vol. 34, No. 3, 814-818, incorporated herein by reference, which relates to a control for programming with the West German Videotext system.
• This is a different implementation of the Videotext programming system than described in Bensch, U., “VPV—VIDEOTEXT PROGRAMS VIDEORECORDER”, IEEE Transactions on Consumer Electronics, Vol. 34, No.
• the present invention incorporates an intelligent program recognition and characterization system, making use of any of the available cues, which allows an intelligent determination of the true nature of the broadcast and therefore is able to make a determination of whether parameters should be deemed met even with an inexact match to the specified parameters.
• VPV also does not incorporate the interface of the present invention, and is much more like the “VCRPlus+” device.
• the videotext signal of the prior art includes a digitally encoded text message which may be displayed in conjunction with the displayed image, similar to the closed caption system.
• the West German system demonstrates how the signal may be received by a computer and interpreted.
• the prior art does not disclose how this signal may be used to index and catalog the contents of a tape, nor does it disclose how this signal may be used to classify or interpret the character of the broadcast.
• the videotext or closed caption signal is not only interpreted as a literal label, as in the prior art, but is also further processed and analyzed to yield data about the content of the broadcast, other than merely the simultaneously broadcast information.
• characterizations may, as stated above, make use of unused available spectrum bandwidth within the NTSC channel space, or other broadcast system channel space, or may be “simulcast” on a separate channel, such as an FM sideband or separate transmission channel.
• a separate channel such as an FM sideband or separate transmission channel.
• Use of a separate channel would allow a separate organization, other than the network broadcasters, to provide the characterization data for distribution to users of devices that make use of the present intelligent system for controlling a VCR.
• the characterization generating means need not be directly linked to the local user machine in order to fall within the scope of the present invention.
• a menu based remote control-contained display device is disclosed in Platte, Oberjatzas, and Voessing, “A New Intelligent Remote Control Unit for Consumer Electronic Device”, IEEE Transactions on Consumer Electronics, Vol. CE-31, No. 1, February 1985, 59-68, incorporated herein by reference. This system does not incorporate on-screen programming, nor various aspects of the display of the present invention.
• VCR video cassette recorder
• the present invention incorporates adaptive features which optimize the programming environment for both the level of the user and the task to be programmed.
• the input device In optimizing the interface, four elements are most important: the input device, the display format, the sequence of the programming operation, and the ability of the device to properly interpret the input as the desired program sequence.
• the present invention proceeds from an understanding that an absence of user frustration with respect to a programmable consumer or industrial device or interface, may be as important as the potential functionality thereof.
• the interface must be designed to minimize the user's frustration level. This can be accomplished by clearly furnishing the possible choices, presenting the data in a logical sequence, and leading the user through the steps necessary to program the device.
• the control means may further comprise a calendar.
• control means provides an option, selectable by said input means in conjunction with said display means, for changing an input program instruction prior to execution by said control means, so that said apparatus enters a state wherein a new program instruction may be input to substitute for said changed input step, wherein said control means verifies said program instructions so that said program instructions are executable by said control means.
• control means further causes said display means to display a confirmation screen after said program instructions are input, so that the programmer may confirm said program instructions.
• Another object of the present invention provides a programmable information storage apparatus having a data input, for receiving data to be stored, said apparatus receiving instructions from a programmer and causing an action to occur on the receipt of data indicating an event, comprising: means for storing data from said data input; an input device, producing an input instruction signal; a control means for receiving said input instruction signal, and storing a program instruction associated with said input instruction signal, said control means storing sufficient program instructions to perform an action on the receipt of data from said data input indicating an event, said control means monitoring the data input to determine the occurrence of various events, comparing the determined events with the program instructions, and performing for storing the data said action on the occurrence of said event; wherein said control means receives identifying data from at least one of said input device and the data input, said identifying data being stored separately from said input data on a storage medium.
• the programmable information storage apparatus may also include means for reading said identifying data stored separately on said storage medium, and may also receive as an input said identifying data.
• Another object of the present invention is to provide a programmable information storage apparatus, wherein said control means further comprises means for recognizing character data present in a data stream of said input data, said identifying data comprising said recognized character data.
• Another object of the present invention is to provide a recording device for recording an analog signal sequentially on a recording medium, comprising means for characterizing the analog signal, wherein data representing said characterization and a location of the analog signal on the recording medium are stored in a directory location on the recording medium separately from the analog signal.
• Another object of the present invention provides a system for presenting a program to a viewer, comprising: a source of program material; means for determining a viewer preference; means for receiving the program material from said source; means for characterizing the program material based on its content; means for correlating said characterized content of the program material with said determined viewer preference to produce a correlation index; and means for presenting the program material to the viewer, if said correlation index indicates a probable high correlation between said characterization of the program material and said viewer preference.
• Still another object of the present invention is to provide a system, wherein said program material is encrypted, further comprising: means for decrypting the program material to produce a decryption event; and means for charging an account of the viewer based on the occurrence of a decryption event.
• Another object of the present invention is to allow said means for characterizing the program material to operate without causing a decryption event.
• the system may comprise a memory for storing the program material while said characterizing means produces characterized content and said correlating means produces said correlation index.
• the characterizing means may also characterize the program material stored in memory, and the program material stored in memory may be compressed.
• Another object of the present invention is to provide a system for presenting a program to a viewer, comprising: a source of program material; means for determining a viewer preference; means for receiving the program material from said source; means for storing the program material; means for preprocessing the program material to produce a reduced data flow information signal retaining information relating to a character of the program material and eliminating data not necessary to characterize the program material; means for characterizing said information signal based on its content; means for correlating said characterized content of said information signal with said determined viewer preference to produce a correlation index; and means for presenting said stored program material to the viewer, if said correlation index indicates a probable high correlation between said characterization of said information signal and said viewer preference.
• the system may also include a means for storing said information signal, wherein said characterizing means characterizes said stored information signal, and also a memory for storing the program material while said characterizing means produces characterized content and said correlating means produces said correlation index.
• Another object of the present invention is to provide a controller for controlling a plant, having a sensor for sensing an external event and producing a sensor signal, an actuator, responsive to an actuator signal, for influencing said external event, and a control means for receiving said sensor signal and producing an actuator signal, comprising: means for inputting a program; means for storing said program; means for characterizing said sensor signal to produce a characterized signal; and means for comparing said characterized signal with a pattern stored in a memory to produce a comparison index, wherein said actuator signal is produced on the basis of said comparison index and said program, wherein said characterization comprises an Affine transformation of said sensor signal.
• the characterization may comprise both an Affine transformation and a Fourier transformation.
• Another object of the present invention provides a method wherein the step of selecting the most closely corresponding one of mapped ranges includes the step of selecting, for each domain, the mapped range with the minimum modified Hausdorff distance calculated as D[db,mrb]+D[1 ⁇ db,1 ⁇ mrb], where D is a distance calculated between a pair of sets of data each representative of an image, db is a domain, mrb is a mapped range, 1 ⁇ db is the inverse of a domain, and 1 ⁇ mrb is an inverse of a mapped range.
• Another object of the present invention provides a method wherein the digital image data consists of a plurality of pixels each having one of a plurality of associated color map values, further comprising the steps of: optionally transforming the color map values of the pixels of each domain by a function including at least one scaling function for each axis of the color map, each of which may be the same or different, and selected to maximize the correspondence between the domains and ranges to which they are to be matched; selecting, for each of the domains, the one of the mapped ranges having color map pixel values which most closely correspond to the color map pixel values of the domain according to a predetermined criteria, wherein the step of representing the image color map information includes the substep of representing the image color map information as a set of values each including an identifier of the selected mapped range and the scaling functions; and selecting a most closely corresponding stored template, based on the identifier of the color map mapped range, the scaling functions and the set of identifiers representing the image information.
• the first criteria may comprise minimizing the
• Another object of the present invention is to provide a method further comprising the steps of: storing delayed image data, which represents an image of a moving object differing in time from the image data in the data processor; generating a plurality of addressable further domains from the stored delayed image data, each of the further domains representing a portion of the delayed image information, and corresponding to a domain; creating, from the stored delayed image data, a plurality of addressable mapped ranges corresponding to different subsets of the stored delayed image data; matching the further domain and the domain by subjecting a further domain to one or both of a corresponding transform selected from the group consisting of a null transform, a predetermined rotation, an inversion, a predetermined scaling, and a predetermined frequency domain preprocessing, which corresponds to a transform applied to a corresponding domain, and a noncorresponding transform selected from the group consisting of a predetermined rotation, an inversion, a predetermined scaling, a translation and a predetermined frequency domain preprocessing, which does not correspond to a transform
• Another object of the present invention provides an apparatus for automatically recognizing digital image data consisting of image information, comprising: means for storing template data; means for storing the image data; means for generating a plurality of addressable domains from the stored image data, each of the domains representing a different portion of the image information; means for creating, from the stored image data, a plurality of addressable mapped ranges corresponding to different subsets of the stored image data, the creating means including means for executing, for each of the mapped ranges, a procedure upon the one of the subsets of the stored image data which corresponds to the mapped range; means for assigning identifiers to corresponding ones of the mapped ranges, each of the identifiers specifying for the corresponding mapped range an address of the corresponding subset of stored image data; means for selecting, for each of the domains, the one of the mapped ranges which most closely corresponds according to predetermined criteria; means for representing the image information as a set of the identifiers of the selected mapped ranges; and
• FIG. 1 is a flow chart of the steps required to set a VCR
• FIG. 2 shows a graphical comparison of required and extra keypresses for the prior art and the interface of the present invention
• FIG. 3 graphically shows the differences in seconds between total time for the prior art for each subject
• FIG. 4 graphically shows the differences in seconds between total time for the interface of the present invention for each subject
• FIG. 5 graphically shows the programming steps for the comparison of the prior art and the interface of the present invention
• FIG. 6 graphically shows comparative statistics by subject comparing the prior art and the interface of the present invention
• FIGS. 7 and 8 graphically show the critical steps in programming the prior art and the interface of the present invention
• FIG. 9 graphically shows the number of keypresses made by test participants comparing the prior art and the interface of the present invention.
• FIG. 10 graphically shows the comparison of the actual and theoretical number of keypresses necessary for programming the prior art and the interface of the present invention
• FIG. 11 graphically compares the actual and theoretical time necessary for programming the prior art and the interface of the present invention
• FIGS. 12 a and 12 b graphically compares the actual and theoretical time necessary for setting the programs in the prior art and the interface of the present invention
• FIGS. 13 and 14 graphically show the percentage time for the critical steps in programming the prior art and the interface of the present invention
• FIG. 15 is a flow diagram of a predictive user interface of the present invention.
• FIG. 16 is a flow diagram of the program input verification system of the present invention.
• FIG. 17 is a flow diagram of a predictive user preference aware interface of the present invention.
• FIG. 18 is a block diagram of a non-program information feature extraction circuit of the present invention.
• FIG. 19 is a diagram of a block of information for a catalog entry of the present invention.
• FIG. 20 is a block diagram of a digital information and analog signal reading/recording apparatus
• FIG. 21 is a block diagram of a user level determining system of the present invention.
• FIG. 22 is a block diagram of a template-based pattern recognition system of the present invention.
• FIG. 23 is a block diagram of a control system of the present invention incorporating a pattern recognition element and an interface
• FIG. 24 is a block diagram of a control system for characterizing and correlating a signal pattern with a stored user preference of the present invention.
• FIG. 25 is a block diagram of a multiple video signal input apparatus, with pattern recognition, data compression, data encryption, and a user interface of the present invention
• FIG. 26 is a block diagram of a control system for matching a template with a sensor input, of the present invention.
• FIGS. 27 , 28 and 29 are flow diagrams of an iterated function system method for recognizing a pattern according to the present invention.
• FIG. 30 is a semi-cartoon flow diagram of the object decomposition and recognition method of the present invention.
• the interface should preferably employ only minimal amounts of abbreviations and the use of complete words is especially preferred.
• standard abbreviations and symbols are acceptable, and displayed character strings may be shortened or truncated in order to reduce the amount of information that is to be displayed, where necessary or desirable.
• An option may be provided to the user to allow full words, which may decrease the information which may be conveyed on each screen and increase the number of screens that must be displayed, or abbreviations and symbols, which may minimize the number of displayed screens of information, thus allowing the user to make the compromise.
• This aspect of the system may also be linked to the adaptive user level function of the present invention, wherein abstract symbols and abbreviations are presented to advanced users, while novices are presented with full words, based on an implicit indication of user level.
• These abstract symbols and abbreviations may be standard elements of the system, or user designated icons. Of course, the user could explicitly indicate his preference for the display type, thus deactivating the automatic adaptive user level function.
• Some display systems have a higher available resolution than others, and the interface is preferably arranged to optimize the intended display for the resolution limits and display format of the intended or available display device. Further, even with sufficient resolution, certain displays are of small size, and thus the visibility of the information may also be optimized by taking into consideration the size, resolution, contrast, brightness of the display, ambient conditions, characteristics of the human visual system, factors specific for a known user, and the available options of the apparatus.
• the interface may employ a number of methods to optimize the visibility of the information for a variety of display devices, storage formats and transmission standards, which may include: NTSC, PAL, SECAM, CCIR-601, HDTV, MUSE, IDTV, VHS, S-VHS, Beta, SuperBeta, Hi-8 mm, videotel or picturephone (Px64), computer display standards (CGA, HGC, EGA, VGA, SVGA, XGA, MacintoshTM, 8514, Private EyeTM, LCD, etc.), etc., over a number of size ranges, e.g.
• a preferred embodiment of the interface of the present invention by automatic sequencing of steps, leads the user through the correct sequence of actions to set a program on the screen, so that no necessary steps are omitted, and no optional steps are accidentally omitted. These steps are shown diagrammatically in FIG. 15 of the present invention.
• such a system does not burden the user with the necessity of inputting superfluous information, nor overwhelm the user with the display of unnecessary data.
• a built-in calendar menu screen is employed so that the user cannot set the device with a program step that relies on a non-existent date.
• Technology that will help eliminate the human problem of setting the wrong (yet existing) date may also be employed. Such technology might include accessing an on-line or other type of database containing media programming information, and prompting the user regarding the selected choice. In situations where it is applicable, the interface should prompt the user as to how many characters the interface is expecting, such as when entering the year.
• a preferred embodiment of the present invention has an easily accessible CHANGE or CANCEL feature, which facilitates backtracking or reprogramming the immediately previously entered information rather than forcing the user to repeat all or a substantial portion of the programming steps.
• a method of the is type described is shown in FIG. 16 of the present invention.
• User input is also facilitated by the provision of frequently used settings as explicit choices, such as “Record today”, “Record tomorrow”, “Noon”, and “Midnight”, so that the user does not have to specify a date in these cases. This will eliminate extra keypresses, and reduce the programming time. In addition, this could eliminate user errors.
• Frequently used choices for program selections are also provided to the user to reduce the number of programming steps necessary and provide the user with all the frequently used selections.
• the especially preferred choices are “Once On . . . ”, “Once a Week on . . . ”, “Monday-Friday at . . . ”, “Everyday at . . . ”. These redundant, complex instructions reduce the number of keystrokes required for data entry, and reduce the amount of programming time required.
• a preferred embodiment of the present invention provides, in the event that a color screen is available, conservatively used color coding, which allows the user to effectively and quickly acknowledge the function of each aspect of the screen.
• the preferred colors are royal blue for “help,” red for mistakes, light blue for information previously entered, and yellow for current information being entered. Of course, other colors could be used, according to the user's or designer's preference, cultural differences, and display parameters.
• a preferred embodiment of the interface contains a confirmation screen which displays to the user all of the categories and selections previously explicitly entered or otherwise inferred, and should be easily understandable. This is shown in FIG. 15 of the present invention. All of the necessary information is displayed on this screen, in addition to the change and cancel options, if possible.
• each screen is preferably consistent throughout the program. All of the screens preferably have similar layouts. “Buttons” or screen locations which are keyed to a particular function, which appear on multiple screens, should appear in approximately the same location on all screens. However, in certain cases, relatively more important information on a given screen may be displayed more prominently, and possibly in a different screen location, in order to reduce the search time. Further, when other factors dictate, each screen may be independently optimized for the prescribed function. For example, a representation of an analog clock dial may be used to set time information. However, even if the format does change, a standard scheme should be maintained, such as the use of a particular color to indicate that a particular program aspect has been changed.
• the interface should display data consistent with standards and conventions familiar to users. For, e.g., when entering dates, users are most familiar with calendars. However, this type of presentation of choices does not eliminate the human problem of entering incorrect information, e.g., setting a wrong, but existing, date.
• the problem of ensuring the accuracy of user input may be addressed by an intelligent interface which stores data concerning programming, user preferences, and by means of some logical method, such as Boolean logic, fuzzy logic, neural network theory, or any other predictive system, determines if an entry is likely in error. Of course, these predictive systems would also provide an initial default entry, so that the a priori most probably action or actions are initially presented to the user.
• the interface of the present invention may also provide emulations of other user interfaces of which a particular user may be familiar, even if these are not optimized according to the presently preferred embodiments of the present invention, or not otherwise well known. These emulations need not be of the same type of device, so that a broad based standard for entry of information into programmable controls, regardless of their type, may be implemented. By allowing emulation, the interface of the present invention could provide compatibility with a standard or proprietary interface, with enhanced functionality provided by the features of the present interface.
• controllers may be implemented by means of software programming of a simple microcomputer, or by use of more specialized processors, such as a Fuzzy Set Processor (FSP) or Neural Network Processor.
• FSP Fuzzy Set Processor
• Neural Network Processor Neural Network Processor
• FSP's are preferred because they have the advantage of being easier to program through the use of presumptions or rules for making the fuzzy inferences, while Neural Networks are less easily programmed and their network weighing values are not easily understood in the abstract.
• neural networks tend to require extensive “training”, while Fuzzy Set Processors may be explicitly programmed without the need of duplicating or simulating actual operating conditions.
• the most frequently used choices preferably should be displayed as the default setting.
• the screen cursor preferably appears at the “accept” screen button, when the screen is displayed.
• This default can either be set in advance, or acquired by the system. In the case of acquired defaults, these may be explicitly set by the user or adaptively acquired by the system through use.
• the interface of the present invention may be taught, in a “teach” mode, the preferences of the user, or may also acquire this information by analyzing the actual choices made by the user during operation of the interface and associated controller. This type of operation is shown schematically in FIG. 15 of the present invention.
• the options of “Midnight” (12:00 AM) and “Noon” (12:00 PM) should preferably be present, as some people often become confused when distinguishing between them.
• Icons such as those indicative of the “sun” and the “moon”, may also be used to facilitate data entry for AM and PM.
• the interface should preferably utilize an internal clock and calendar so that the user cannot set the time or program to record on a nonexistent date. Such a system could also compensate for daylight-savings time seasonal adjustments.
• the cursor is preferably distinctive and readily distinguished from other parts of the screen. This may be by color, attribute (i.e. blinking), size, font change of underlying text, or by other means.
• the user can preferably exit the programming sequence at any time by selecting a “Main Menu” button which may be on the lower left-hand corner of every screen.
• the user is preferably provided with an adequate amount of feedback, and error messages should be directive in nature.
• An acknowledgement is preferably displayed after each entry.
• the user should preferably not be able to go to the next programming step until the current step has been completed.
• a message to convey why the user can not continue should appear when an attempt to prematurely continue is recognized.
• the “help” function is available for when the subject does not know what to do.
• the “help” screen(s) preferably explains the functions of each of the available buttons or functions, but may also be limited to those that are ambiguous.
• the “help” screen may also be used to indicate a current status of the interface and the controller.
• the “help” function may also provide access to various other functions, such as advanced options and configurations, and thus need not be limited to merely providing information on the display.
• the help system may incorporate a hypertext-type system, wherein text or information relating to concepts that are conceptually linked may be easily accessed from one another, and in a logical sequence.
• help should be locked to a choice which returns the user to where they left off in the programming sequence, and this choice should be highlighted.
• the “help” function may also comprise “balloon help” similar to the system adopted by Apple Computer, Inc. in Macintosh Operating System 7.0 and later versions.
• the interface preferably initiates the programming sequence where the user wants to be, so that the interface has so-called “smart screens”. For example, when a VCR is first powered up, and the time and date are not stored in the machine, the “set date” and “set time” screens should appear.
• the sequence of screens may also vary depending on the system predicted requirements of the user and various aspects of the improved interface of the present invention. This is shown schematically in FIG. 17 of the present invention.
• the preferable input device for the interface of the present invention provides as few buttons as possible to achieve the required functionality, thus reducing potential user intimidation, focusing the user's attention on the interactive display screen, where the available choices are minimized to that number necessary to efficiently allow the user to program the discrete task presented.
• a computer mouse with 1 to 3 buttons is the preferred input device, for use with a general purpose computer as a controller, while a trackball on a remote control device is especially preferred for limited purpose controllers because it does not require a flat surface for operation.
• Other stationary or movement sensitive input devices may, of course be used, such as joysticks, gyroscopes, sonic echo-location, magnetic or electrostatic location devices, RF phase location devices, etc.
• the present interface minimizes the number of necessary keys present on an input screen, while maintaining the functionality of the interface. It is noted that a strict minimization without consideration of functionality, might lead to inefficiency. For example, if the user wants to record a program which airs Monday-Friday, he would have to set five separate programs, rather than one program if a “weeknights” choice is made available.
• the interface preferably should be easy to learn and should not require that a user have prior knowledge of the interface in order to use it.
• Research has shown that people do not program their VCRs often, and they often forget the sequence of steps between recording sessions.
• the present invention incorporates an adaptive user level interface, wherein a novice user is presented with a simpler interface with fewer advanced features initially available, so that there is less searching for the basic functions. A more advanced user is presented with more advanced choices and functions that are available initially.
• the user identifies himself to the controller in block 1701 .
• a directory or a catalog is recorded, preferably digitally, containing the programming information, as well as additional information about the recorded programs, in a header, i.e., at the beginning of the tape, or in other locations on the tape.
• the device may also catalog the tape contents separately, and based on an identification of the tape, use a separately stored catalog.
• a format for storing information is shown in FIG. 19 .
• the tape transport apparatus In a sequential access medium, the tape transport apparatus must still shuttle to the location of the desired material, but it may do so at increased speeds, because there is no need to read the tape once the location is determined; after the tape transport nears the desired spot, the tape may be slowed or precisely controlled to reach the exact location.
• the tape read and drive system is shown schematically in FIG. 20 .
• the algorithm used in the final stage of approach may use fuzzy logic, mathematical formulae modeling the system (differential equations) to implement a Proportional-Differential-Integral (PID) controller, or a controller of higher order, or other known control methods. If a selection is to be recorded over, the start and stop locations would be automatically determined from the locations already indicated on the tape.
• PID Proportional-Differential-Integral
• this information could be stored in memory device (which reads a catalog or index of the tape when a new tape is loaded) or non-volatile memory device (which stores information relating to known tapes within the device) or both types of memory in the VCR, so that an index function may be implemented in the VCR itself, without the need to read an entire tape.
• a printer such as a thermal label printer (available from, e.g. Seiko Instruments, Inc.), attached to the device, could be available to produce labels for the tapes, showing the index, so that the contents of a tape may be easily indicated. These contents may be derived from published data or database, transmitted data, and/or data determined by the control itself.
• the present invention also allows encryption and decryption of material, much as the Videocipher series systems from General Instruments, and the fractal enciphering methods of EMC 2 and Iterated Systems, Inc.
• the present invention is not limited to broadcasts, and instead could implement a system for both broadcasts and prerecorded materials. In the case of copying from one tape to another, such a system could not only provide the herein mentioned library functions of the present invention, it could also be used to aid in copy protection, serial copy management, and a pay-per-view royalty collection system.
• Such a system could be implemented by way of a telecommunication function incorporated in the device, shown as block 1808 of FIG. 18 , or an electronic tag which records user activity relating to a tape or the like.
• a royalty fee, etc. could automatically be registered to the machine either by telecommunication or registry with the electronic tag, allowing new viewer options to be provided as is compared with present VCR's.
• an encrypted tape or other source material (so that special playback equipment need be used, and a usage registered), used with this device, could be decrypted by a decryption key available by telecommunication with a communication center, remote from the user, in a decryption unit, shown schematically as the decrypt unit 1806 a of FIG. 18 .
• a VCR device of an embodiment of the present invention would indicate its identity, and an account is charged a fee for such use.
• Such a system could also be used for controlled access software, for example for a computer, wherein a remote account is charged for use of the software.
• a system differs from the normal “key” or “dongle” because it requires on-line access for an encryption key, which may offer different levels of use. It also differs from a call-in registration, because of the automatic nature of the telecommunication. This presently described system differs from normal pay-per-view techniques because it allows, in certain instances, the user to schedule the viewing.
• the device allows a user to create and distribute custom “software” or program material.
• the present controller could then act as the “telecommunication center” and authorize decryption of the material.
• the present invention is advantageous in this application because it provides an advanced user interface for creating a program (i.e. a sequence of instructions), and it assists the user in selecting from the available programs, without having presented the user with a detailed description of the programs, i.e., the user may select the choice based on characteristics rather than literal description.
• a program i.e. a sequence of instructions
• the user may make a viewing decision based on the recommendation of the interface system, or may review the decision based on the title or description of the program.
• the encryption may be of any type, but for sensitive material, i.e. where mere distortion of the material (e.g., loss of synchronization information and phase distortion) would be insufficient, an analog multiple subband transform, with spread spectrum band hopping and digital encryption of various control signals, would be particularly difficult for the user to view without authorization, and could be effectively implemented with conventionally available technology.
• the fractal compression and encryption of the EMC 2 and Iterated Systems, Inc. system is also particularly preferred.
• a digital storage format if a digital storage format is employed, a strict digital encryption system may be used.
• the implementation of these encryption systems is known to those skilled in the art. These may include the NBS, VSS and NSA encryption standards, as well as various proprietary standards.
• Menu options are preferably displayed in logical order or in their expected frequencies. Research has shown that a menu-driven interface is best for applications involving new users and does not substantially hinder experienced users. Menu selection is preferably used for tasks which involve limited choices. They are most helpful for users with little or no training. Each menu should preferably allow only one selection at a time. Most of the information is preferably entered using a numeric keypad (entry method), rather than using up and down arrow keys (selection method). If there is more than one keystroke required, the user must then select an “OK” button to continue in the programming sequence. However, if the selection method is used, all of the choices are displayed on the screen at once. In addition, no leading zeros are required.
• the number of steps required to complete the task through a sequence of menus should be minimized.
• the choice of words used to convey information should not be specific computer terms, but rather normal, everyday terms which are easy to understand. In addition, very few abbreviations should be used. All necessary information which the user needs should preferably be displayed at once. A user preferably should not have to rely on his memory or his previous experience, in order to find the correct choice, at least at the lower user levels. If all selections cannot be displayed at once, a hierarchical sequence is preferably used.
• a main menu should preferably provide a top level to which the user can always return and start over.
• VCRs Users of VCRs are concerned with the layouts of both the control panel on the VCR device and the remote control.
• the vast majority prefer on-screen programming, which utilizes the remote control rather than the control panel, and express a preference for entering the numbers over pressing the “up” and “down” arrow keys for selecting the time and channel. Some favor choosing the “start” and “stop” times over choosing the “start” time and duration.
• users When using existing VCRs, users generally want more feedback, and they want to know when the VCR is ready to program. Subjective data indicates that it is preferable to reduce the amount of time required to set the clock and two programs on a VCR to a maximum of 7 minutes, wherein the reduction should focus on lessening the search time, which is the amount of time consumed because users do not know what to do next.
• Searching and learning times should be kept to a minimum in order to obtain a subjectively better interface.
• the system's logic should reflect the users' expectations, offer visual clues and feedback, and stay within human memory limits.
• the VCR should turn on not only with the “Power” button, but also by inserting a tape into the device.
• the sequence of steps for setting the machine to is record, if the user does not indicate implicitly or explicitly that he knows how to use the device, should assume that the user is a novice. Nothing should be taken for granted.
• an improved interface an attempt is made to: Reduce the searching time; Reduce the learning time; Simplify the entering of data; and, Reduce the intimidation experienced by certain persons when using electronic devices.
• the apparatus comprises a program entry device for a VCR.
• the human interface element has an infrared device to allow wireless communication between the human interface device and the VCR apparatus proper.
• the human interface device also includes a direct-manipulation type input device, such as a trackball or joystick.
• a direct-manipulation type input device such as a trackball or joystick.
• J-cursor or “mousekey” which embeds a two (x,y) or three (x,y,p) axis pressure sensor in a button conformed to a finger, present in a general purpose keyboard
• a keyboard joystick of the type described in Electronic Engineering Times, Oct. 28, 1991, p.
• a new device which may also be suitable is the GyroPoint, available from Gyration Inc., which provides 2-D or 3-D input information in up to six axes of motion: height, length, depth, roll, pitch and yaw. While such a device is generally considered too complex and costly for use with a VCR, the many degrees of freedom available may provide suitable input for other types of controllers, such as those based on “Artificial Reality” or which track a moving object, where many degrees of freedom and a high degree of input accuracy is required.
• These input devices may be broken down into a number of categories: direct inputs, i.e. touch-screen and light pen; indirect inputs, i.e. trackball, joystick, mouse, touch-tablet, bar code scanner (see, e.g., Atkinson, Terry, “VCR Programming: Making Life Easier Using Bar Codes”), keyboard, and multi-function keys; and interactive input, i.e. Voice activation/instructions (see, e.g., Rosch, Winn L., “Voice Recognition: Understanding the Master's Voice”, PC Magazine, Oct. 27, 1987, 261-308); and eye tracker and data suit/data glove (see, e.g. Tello, Ernest R., “Between Man And Machine”, Byte, September 1988, 288-293; products of EXOS, Inc; Data Glove).
• direct inputs i.e. touch-screen and light pen
• indirect inputs i.e. trackball, joystick, mouse, touch-tablet, bar
• GUIs Graphical User Interfaces
• a device such as a touch-screen, with a more natural selection technique, is technically preferable to the direct manipulation method.
• its low accuracy and high cost make other inputs more commercially practical.
• the user must be within arms' length of the touch-screen display.
• EET Electronic Engineering Times
• Oct. 28, 1991, p. 62 incorporated herein by reference
• EET Electronic Engineering Times
• This technique is directed at a different aspect of user interaction with a programmable device than the preferred embodiment of the present invention, in that the input device does not have a minimal number of keys.
• the device disclosed in EET is intended for use in a full function keyboard
• the preferred embodiment of the present invention is directed towards the minimization of the number of keys and avoidance of superfluous keys by provision of a pointing device.
• the present invention could be used with a full function input device, where appropriate, and the joystick of EET (Oct. 28, 1991, p. 62) would be suitable in this case.
• Search time the time necessary in order to locate and execute the next desired function or instruction. Search time may be minimized by the inclusion of up to a maximum of 4-8 choices per screen and by use of consistent wording and placement of items on the display.
• the present invention proceeds from the understanding that there are a number of aspects of a programmable interface that are desirable:
• the system should provide on-screen understandable information, with adequate visual feedback.
• the displays should be consistent.
• Color coding should be employed, where applicable, using, e.g. blue—new input; red—error condition; yellow—static, unchanged value.
• Layouts should be logical, and follow a predictable pattern. There should be a maximum of 4-8 choices per screen to minimize searching time. Keys should be labelled with text rather than with ambiguous graphics. However, a combination of both may be preferable in some cases.
• Tasks should be simple, require a short amount of time and not create user frustration.
• the system should guide the user along a decision path, providing automatic sequencing of steps.
• the most frequently used choices should be provided as defaults, and smart screens may be employed.
• the learning curve should be minimized through the use of easily understandable choices. As a user becomes more sophisticated, the interface may present more advanced choices.
• the VCR should be programmable from both the remote device and the control panel.
• Each operation should require only one keypress, if possible, or otherwise reduce the number of keypresses required.
• the number of buttons on the remote control should be minimized so that as few buttons as are required are provided.
• the input device should provide for the direct manipulation of screen elements.
• a menu driven interface should be provided.
• the interface of the present invention provides an automatic sequencing of steps which does not normally let the user continue until the previous step is complete. This is shown schematically in FIG. 16 . In this manner, important steps will not be inadvertently omitted.
• the interface Upon entering the programming sequence, if the current date or time is not set, the interface will prompt the user to enter this information. Thereafter, the interface will normally default to the main menu, the most frequently used first screen.
• the interface of the present invention is adaptive, in that its actions depend on the current state of the device, including prior programming or use of the device by the user. It can be appreciated that this adaptive behavior can be extended to include extended “intelligence”.
• the apparatus could provide multiple levels of user interface, e.g. beginner, intermediate, and advanced, which may differ for various functions, based on the behavior of the user.
• This user interface level determining feature extraction system is shown diagrammatically in FIG. 18 .
• prior art interfaces that have different user interface levels, allow the user to explicitly choose the interface level, which will then be used throughout the system until reset.
• the interface of the present invention would study the initial behavior of the user to determine the expected user level of that user.
• An apparatus for performing this function is shown schematically in FIG. 18 , and in more detail in FIG. 21 .
• the output could be adaptively filtered to increase the reliability, a function of the main control 1806 , without unnecessarily limiting an advanced user who wishes to move the cursor quickly.
• an adaptive user interface level is a user who repeatedly requests “help” or user instructions, through the explicit help request detector 2115 , which causes an output from the current help level output 2102 ; such a user may benefit from an automatic context-sensitive help system, however such a system may interfere with an advanced user, and is unnecessary in that case and should be avoided.
• This adaptive user interface level concept is not limited to a particular embodiment of the present invention, such as a VCR, and in fact, may be broadly used wherever a system may be used by both experienced and inexperienced users.
• the interface could maintain a history of feature usage for each user, as in the past user history block 2107 , and provide a lower user interface level for those features which are rarely used, and therefore less familiar to the user, through the current user level output 2101 .
• the intelligence of the device of the present invention is not limited by the foregoing examples; the user could also input characteristics of the program material that are desired, and characteristics of that program material which is not desired.
• the device would then, over time, monitor various broadcast choices, and determine which most closely match the criterion, and thus be selected. For example, if the user prefers “talk-shows”, and indicates a dislike for “situation comedies” (“sitcoms”), then the device could scan the various available choices for characteristics indicative of one or the other type of programming, and perform a correlation to determine the most appropriate choice(s).
• a sitcom for example, usually has a “laugh track” during a pause in normal dialogue.
• the background of a sitcom is often a confined space, from different angles, which has a large number of props.
• a talk-show on the other hand, more often relies on actual audience reaction (possibly in response to an “applause” sign), and not prerecorded or synthesized sounds.
• the set is simple, and the broadcast often shows a head and neck, or full body shot with a bland background.
• a signal processing computer programmed for audio and/or video recognition, could differentiate between at least the two types with some degree of efficiency, and with a possibly extended sampling time, have excellent recognition accuracy. Further, with the aid of feedback, the search criterion would be improved.
• the presently described recognition algorithms may be adaptive and learning, and need not apply a finite set of predetermined rules in operation.
• a neural network processor may be implemented, as known in the art.
• the feature extraction and correlation system of the present invention is shown in FIG. 22 .
• the video input including the audio signal and all other available data, are input in the video input 2201 .
• This is transferred to a frame buffer 2202 , which temporarily stores all of the information.
• This frame buffer 2202 may have an integral or separate prefiltering component 2203 .
• the filtered signal(s) are then passed to a feature extractor 2204 , which divides the video frame into a number of features, including sound, movement, objects, correlated sound and object, background, etc.
• These features are then passed to a transform engine or multiple engines in parallel, 2205 .
• These transform engines 2205 serve to match the extracted features with the standard form of the templates in the template database 2206 .
• the transformed extracted features and the templates are then correlated by a correlator or correlators 2207 .
• the parallelism of the transforms and correlators serves to increase the recognition speed of the device.
• the outputs of the correlators are input into an adaptive weighing network 2208 , to produce a probability of a match between a given feature and a given template.
• the recognition is completed in an identifier 2209 , which produces a signal identifying one or more objects in the video frame input.
• the identifier 2209 also has an output to the template database 2206 , which reinforces the recognition by providing feedback; therefore, if the same object appears again, it will be more easily recognized.
• the template database 2206 therefore also has an input from the feature extractor 2204 , which provides it with information regarding the features recognized.
• the parallel transform engines 2205 , correlators 2207 , and adaptive weighing network 2208 also allows the system to ignore features that, though complex, do not aid in recognition. For example, during dialogue, the soundtrack voice will correlate with the mouth movements. Thus, the mouth movements aid little in recognition, and may be virtually ignored, except in the case where a particular person's mouth movements are unique, e.g., “Gomer Pyle”.
• the complexity and parallelism in the intermediate recognition stages may actually simplify the later stages by allowing more abstract features to be emphasized in the analysis.
• the pattern recognition function of the present invention could be used, in a VCR embodiment of the present invention to, e.g., to edit commercials out of a broadcast, either by recognition of characteristics present in commercials, in general, or by pattern recognition of specific commercials in particular, which are often repeated numerous times at various times of the day, and on various broadcast channels.
• certain media present a recognizable audio or video cue when a commercial break has ended. (E.g. often sports events, such as the Olympic Games, will have theme music or distinctive video screens.)
• the present device need not respond immediately to such cues, and may incorporate a delay, which would store the information while a decision is being made.
• the temporary storage medium may be independent of the pattern recognition system.
• the first serving as the desired signal to be stored may actually be two data streams: the first serving as the desired signal to be stored, and the second to the pattern recognition system.
• This system is advantageous because is allows a broadcast quality temporary storage, which may be analog in nature, to be separate from the digital signal processing and pattern recognition stage, which need only retain significant information for the pattern recognition, and therefore may be highly compressed, and devoid of various types of information which are irrelevant or of little importance to the pattern recognition functions.
• the temporary storage may employ a different image compression algorithm, e.g. MPEG II or MPEG++, which is optimized for retention of visually important information, while the recognition system may use a compression system optimized for pattern recognition, which may retain information relevant to the recognition function which is lost in other compression systems, while discarding other information which would be visually important.
• the compression algorithm is integral to the recognition function, preparing the data for the pattern matching and characterization, and therefore should be optimized for high throughput.
• the initial compression may include redundant information, if necessary in order to achieve real-time or near real-time recognition, and, thus may actually result in a larger intermediate data storage requirement than the instantaneous data presented to the recognition system; however, the term “compression”, in this case, applies to the long term status of the device, and in a real-time recognition function, the amount of data stored for use in recognition will always be less than the cumulative amount of data presented, except during the very initial stages of data acquisition.
• the image may be compressed using the so called “fractal transform”, using the method of Barnsley and Sloan, which is implemented and available in product form from Iterated Systems, Inc., Norcross, Ga., as the FTC-II, which incorporates eight fractal transform integrated circuit chips, 1 MByte of RAM, and an Intel i80960CA-25 ⁇ P, and operates in conjunction with P.OEM software, which operates under MS-DOS.
• FTC-II hardware compression requires approximately 1 second per frame, while software decompression on an Intel 80486-25 based MS-DOS computer, using “Fractal Formatter” software, can be performed at about 30 frames per second, which allows approximately real time viewing.
• This method is advantageous because the transform allows compression up to about 2456:1, while still maintaining an aesthetically acceptable result.
• the fractal method could be used as a part of the image recognition system.
• the compression system might also be applicable to audio compression as well, so that a single hardware system could incorporate the basic functions of the device. It is noted that the audio compression and image recognition functions cannot be performed on the FTC-II board, and require a separate device. It should also be noted that an even more efficient compression-pattern recognition system could be constructed by using the fractal compression method in conjunction with other compression methods, which may be more efficient under certain circumstances, such as discrete cosine transform (DCT) or wavelet techniques.
• DCT discrete cosine transform
• This step includes the substep of executing, for each of the mapped range blocks, a corresponding procedure upon the one of the subsets of the stored image data which corresponds to the mapped range block.
• Unique identifiers are then assigned to corresponding ones of the mapped range blocks, each of the identifiers specifying for the corresponding mapped range block a procedure and a address of the corresponding subset of the stored image data.
• the one of the mapped range blocks which most closely corresponds according to predetermined criteria is selected.
• the image information is represented as a set of the identifiers of the selected mapped range blocks. This method allows a fractal compression of image data. In particular, Drs.
• the fractal method proceeds from an understanding that real images are made up of a plurality of like subcomponents, varying in size, orientation, etc.
• a complex block of data may be described by reference to the subcomponent, the size, orientation, etc. of the block.
• the entire image may thus be described as the composite of the sub-images. This is what is meant by iterative function systems, where first a largest block is identified, and the pattern mapping is repetitively performed to describe the entire image.
• the FTC-II board is used in conjunction with a frame-grabber board, such as Matrox, Quebec, Canada, Image-LC board, or a Data Translation DT1451, DT2651, DT2862, DT2867, DT2861 or DT2871, which may perform additional functions, such as preprocessing of the image signal, and may be further used in conjunction with an image processing system, such as the Data Translation DT2878.
• a frame-grabber board such as Matrox, Quebec, Canada, Image-LC board
• a Data Translation DT1451, DT2651, DT2862, DT2867, DT2861 or DT2871 which may perform additional functions, such as preprocessing of the image signal, and may be further used in conjunction with an image processing system, such as the Data Translation DT2878.
• a fractal-based system for real-time video compression, satellite broadcasting and decompression is also available from Iterated Systems, Inc. and Entertainment Made Convenient 2 , Inc. (EMC 2 ).
• EMC 2 Entertainment Made Convenient 2 , Inc.
• This system also incorporates anti-copy encryption and royalty and accounting documentation systems.
• the interface of the present invention could interact with the standard accounting system to allow royalty-based recording, and possibly implement a serial-copy recording prevention system.
• the EMC 2 system does not incorporate the intelligent features of the present invention. In particular, a user must still explicitly select a program, rather than allow an intelligent system to assist in selection and programming of the device. This system is described in “EMC 2 Pushes Video Rental By Satellite”, Electronic Engineering Times, Dec. 2, 1991, p. 1, p. 98, which is incorporated herein by reference.
• one embodiment of the device may incorporate a memory for storing a program, before being transferred to a permanent storage facility, such as tape.
• a memory may include a hard disk drive, magnetic tape loop, a rewritable optical disk drive, or semiconductor memories, including such devices as wafer scale memory devices. This is shown diagrammatically as the intermediate storage 2210 of FIG. 22 .
• the capacity of such a device may be effectively increased through the use of image data compression, which may be proprietary or a standard format, i.e.
• MPEG MPEG, MPEG-II, MPEG++ (Motion Picture Experts Group), JPEG (Joint Photographic Experts Group), Px64 (CCITT H.261, videoconferencing transmission standard), DVI (Digital Video Interactive), CDI (Compact Disk Interactive), etc.
• Standard devices are available for processing such signals such as the IIT Vision Processor (VP) chip, Integrated Information Technology Inc., Santa Clara, Calif., the C-Cube CL550B (JPEG) and CL950 (MPEG decoding), SGS-Thompson STI3220, STV3200, STV3208 (JPEG, MPEG, Px64), LSI Logic L64735, L64745 and L64765 (JPEG) and Px64 chip sets, and the Intel Corp. i750B DVI processor sets (82750PB, 82750 DB).
• IIT Vision Processor VP
• JPEG C-Cube CL550B
• CL950 MPEG decoding
• SGS-Thompson STI3220 STV3200
• STV3208 JPEG, MPEG, Px64
• LSI Logic L64735, L64745 and L64765 JPEG
• Px64 chip sets and the Intel Corp. i750B DVI processor sets (82750PB,
• the compression algorithm may be lossless or lossy, depending on the application.
• Various different methods and paradigms may be used.
• DCT discrete cosine transform
• wavelets wavelets
• fractals and other known transform methods
• a compressed image may also be advantageously used in conjunction with the image recognition system of the present invention, as described above. In such a case, the compression system would retain the information most important in the recognition function, and truncate the unimportant information.
• a further method of performing pattern recognition, especially of two dimensional patterns, is optical pattern recognition, where an image is correlated with a set of known image patterns represented on a hologram, and the product is a pattern according to a correlation between the input pattern and the provided known patterns. Because this is an optical technique, it is performed nearly instantaneously, and the output information can be reentered into an electronic digital computer through optical transducers known in the art.
• Such a system is described in Casasent, D., Photonics Spectra, November 1991, pp. 134-140, which is incorporated herein by reference. The references cited therein provide further details of the theory and practice of such a system, and they are also incorporated herein by reference. Lendaris, G.
• optical recognition systems are best suited to applications where an uncharacterized input signal frame is to be compared to a finite number of visually different comparison frames (i.e., at least one), and where an optical correlation will provide useful information.
• a user wished to detect one of, e.g., “Johnny Carson”, “Dan Rather”, “Willard Scott”, or “Jane Pauley”, a number of different views of these persons would be formed as a holographic correlation matrix, which could be superimposed as a multiple exposure, stacked in the width dimension, or placed in a planar matrix, side by side.
• the detection system produces, from the uncharacterized input image and the holographic matrix, a wavefront pattern that is detectable by photonic sensors.
• optical pattern recognition method is limited in that a holographic image must be prepared of the desired pattern to be detected, and that optically similar images might actually be of a different image, if the differences are subtle.
• this method may be used in conjunction with electronic digital pattern recognition methods, to obtain the advantages of both.
• the block labelled intermediate processing 2211 of FIG. 22 shows that the intermediate storage need not retain the information as received from the frame buffer 2202 , and in fact, may prepare it for the feature extractor 2204 .
• the storage medium itself need not be normal videotape (VHS, Beta, 8 mm) and may be an adapted analog storage technique or a digital storage technique.
• the interface of the present invention need not be limited to audio-visual and multimedia applications, as similar issues arise in various programmable controller environments. Such issues are disclosed in Carlson, Mark A., “Design Goals for an Effective User Interface”, Electro/82 Proceedings, 3/1/1-3/1/4; Kreifeldt, John, “Human Factors Approach to Medical Instrument Design”, Electro/82 Proceedings, 3/3/1-3/3/6; Wilke, William, “Easy Operation of Instruments by Both Man and Machine”, Electro/82 Proceedings, 3/2/1-3/2/4; Green, Lee, “Thermo Tech: Here's a common sense guide to the new thinking thermostats”, Popular Mechanics, October 1985, 155-159; Moore, T. G.
• the pattern recognition function would be used to execute a contingent program.
• a sensor or sensor array could be arranged to detect a “door opening”. On the occurrence of the door opening, the system would recognize this pattern, i.e.
• the system would take appropriate action, including: halt of normal climate control and impose a delay until the door is closed; after closure, set a time constant for assimilation of the replaced air with the climate controlled air; based on the actual climatic condition after assimilation, or a predicted climatic condition after assimilation, begin a climate compensation control; optionally, during the door opening, control a pressure or flow of air to counterbalance the normal flow through the door, by using a fan or other device.
• the climate may differ in temperature, humidity, pollutants, or the like, and appropriate sensors may be employed. This generalized system is shown in FIG.
• the sensor array 2301 interfaces with a microprocessor 2302 with a serial data port 2302 a , is which transmits sensor data to a control 2303 .
• the control 2303 further interfaces or includes a data pattern recognition system 2304 and an interface and programming console 2305 of the present invention, using the intelligent features and adaptive pattern recognition techniques.
• the control 2203 controls the plant 2306 , which includes all the controlled actuators, etc.
• a characteristic of the computer file which is analogous to the broadcast program in temporary storage of a VCR, is classified according to some criteria, which may be explicit, such as an explicit header or identifying information, or implicit, such as a document in letter format, or a memorandum, as well as by words and word proximity.
• some criteria which may be explicit, such as an explicit header or identifying information, or implicit, such as a document in letter format, or a memorandum, as well as by words and word proximity.
• a recognition system could differentiate various clients or authors based on the content of the document, and these could be stored in different manner.
• the text analysis system of a text-based computer storage system is analogous to the program classification system of the VCR embodiment of the present invention.
• the VCR could incorporate optical character recognition of text displayed in the program material, or directly receive text information as a part of a closed caption or videotext system.
• the VCR device of the present invention could recognize and classify programs based on textual cues, and make decisions based on these cues. This might also provide a simple method of discriminating program material, for example, if a commercial does not include close caption or Second Audio Program (SAP), while the desired program does, or vice versa, then a commercial could be discriminated from a program with very little computational expenditure.
• SAP Second Audio Program
• Red text signifies instructions or errors
• yellow text represents data which must be entered or has not been changed
• blue text shows newly entered program data or status information.
• Blue buttons represent buttons which should normally be pressed during the programming sequence. Red buttons signify an erratic pattern in the data entry, such as the “cancel” and “return to main menu” buttons.
• these colors can be replaced by other display attributes, such as intensity, underline, reverse video, blinking and pixel dithering pattern, in addition to the use of various fonts. Such a situation would include a monochrome monitor or display.
• the date may be entered in the form of a calendar rather than as numbers (i.e., 9/6/91).
• This calendar method is advantageous because users may wish to input date data in one of three ways: day of the week, day relative to the present, and day of the month.
• the present method allows the current date to be highlighted, so that the calendar may be used to easily enter the absolute day, absolute date, and relative day.
• the choices “today” and “tomorrow”, the most frequently used relative recording times, are included in addition to a month-by-month calendar. This information is provided to avoid an unnecessary waste of time and user frustration.
• another aspect of the present invention is to provide a partially redundant interactive display input system which allows, according to the highest probability, the choices to be prominently displayed and easily available, in addition to allowing random access to all choices.
• the present device allows common user mistakes to be recognized and possibly addressed, such as the confusion between 12:00 PM and 12:00 AM with midnight and noon, respectively. Therefore, the options of “noon” and “midnight” are provided in addition to a direct numeric clock input. When entering time information, leading zeros need not be entered, and such information may be entered in either fashion.
• the criteria for system acceptance of input depends on how many keystrokes are required on the screen. If only one keystroke is required to complete input of the information, upon depressing the key, the programming sequence will continue. If more than one keypress is required, the user must depress the “OK” button to continue programming. This context sensitive information entry serves to avoid unnecessary input.
• An on-line “help” system and on-line feedback is preferably provided to the user throughout various aspects of the interface.
• Other features include minimizing the number of keypresses required to program the device.
• the interface of the present invention applied to a VCR control comprises a virtual keypad entry device, a directional input control for a cursor on a display screen, and a selection button.
• the input device has an input corresponding to a direction of movement relative to the cursor position.
• the display elements for a preferred embodiment of the present interface include:
• program type indicator program once, program once a week, program Monday-Friday, program everyday
• the interface of the present invention includes an internal clock, 4 program memory, and the capability to display a graphical color interface.
• this design is a unique implementation for an instrument to be used for programming an event driven controller via an interactive display. All information that the user needs is displayed on the screen to avoid the unnecessary searching for information. This information includes the current date and current time.
• the AKAI interface of the prior art hereinafter referred to as the prior art interface, was chosen because users made the fewest errors while using this machine, and no subject quit while programming, as compared to three other VCRs tested, a Panasonic PV4962 (Bar Coder), an RCA VKP950 (on-screen programming), Panasonic PV4700 (Display Panel).
• the present embodiment was constructed and tested using HyperPADTM, a rapid prototyping package for an IBM-PC Compatible Computer. It is, of course obvious that the present embodiment could be incorporated in a commercial VCR machine by those skilled in the art, or be implemented on many types of general purpose computers with output screens which allow on-screen feedback for the programming operation. Further, the present embodiment can control an infrared remote controlled VCR or translate the programming information and program an infrared remote control through an interface to an infrared transmitter.
• the present embodiment incorporates a mouse input device. It is understood that a small trackball with a button for selection, mounted on a remote control would be preferred. However, a computer mouse is easily available, and the mouse and trackball data are essentially similar for the type of task used in this study, with trackball performance being slightly faster. For daily use on a VCR however, a trackball would be a more preferable input device because it does not require a hard, flat surface, which is not always available to a user, such as in the situation where a person is watching television while sitting in a chair or sofa.
• a GeniusTM Mouse was used as the input device in the prototype of the interface of the present invention. With the mouse, the user could view all of the choices at once on the display screen, and then make a selection from the items on the screen by moving the cursor and then pressing the left mouse button.
• Subjective data was also collected; it was verbally supplied by the subject during and after the testing. Usability tests were run, using the “Thinking-Aloud” technique. This method requires users to verbalize their thoughts as they interact with the system. This technique is especially useful in discovering strategies which users employ in approaching tasks, pin-pointing problems, and discovering the reasons why they occur. In addition, demographic data, such as each subject's age, occupation, and experience using VCRs and mice was also recorded.
• the design was optimized according to the above-mentioned criteria through a procedure of testing, alteration of the simulation, and retesting.
• the alterations were maintained if they resulted in an improvement in subjective and/or objective criteria. Those alterations that did not result in improvement were reverted to a previous state.
• the interface be individually optimized for persons of various demographic groups, ages, education levels, etc., so that, in accordance with an object of the invention, the interface best matches a particular user's expectations.
• Simultaneous multivariate alterations were also implemented in order to demonstrate an interactive effect between various implementations. In such testing, subjective factors were weighted more heavily than objective factors because the purpose was to determine a qualitative effect, rather than a quantitative comparison.
• the final testing was conducted as a repeated measures experiment of naive subjects.
• the tasks required were to set the clock and 3 programs to simulate a situation where the subject might go on vacation and, upon arrival home, have the desired programs on tape. Three programs were set so that the learning time between programs could be more accurately studied. The subjects did not know which interface was the experimental one.
• Some of the interface factors affecting user performance include:
• the optimized interface of the present invention sought to minimize the first three of these aspects. Recall and Learning were, for the present purpose, considered as identical since data collected has shown that most people perform the task being studied (time-shift programming) less than once a month and thus have to re-learn the task each time. Functionality was also disregarded because it was a negligible factor between tasks.
• T task T acquire +T execute
• the execution time is the time, t j , for each of these operators j weighted by the frequency, n j , with which they occur, plus the total system response time, T R , to the steps performed by the user.
• t K 0.2 seconds for an average typist or mouse user
• the subjects' entry times, actions, and the computer response time were then subtracted from the total time required to perform the task in order to determine T acquire .
• This technique gives estimates accurate to about 20% of actual times required by users.
• T R the average computer response time, was calculated individually for each subject.
• the researcher, an expert user of the systems performed the tasks on both a 10 MHz (Intel 80286 based) and a 33 MHz (Intel 80386DX based) computer.
• the time using the faster computer was then subtracted from the time using the slower computer to achieve a measure of how much slower the 10 MHz computer was.
• Errors are often made by the users and they can be classified as follows:
• the errors of omission are remedied by the fact that the user cannot continue programming if the current step is not completed. In the prior art interface, this is not the case and critical actions may be overlooked.
• FIG. 2 shows the required and the extra keypresses for each interface.
• the average time required to complete the experimental tasks with the interface of the present invention was 560.1 seconds (9.3 minutes).
• the average computer response time, T R assuming it took 0.18 seconds per keypress, was 57.5, or 11% of the total time. When this is deducted from the total time, the new average is 502.7 seconds (8.4 minutes).
• the shortest length of time recorded for programming was 143.5 seconds (2.4 minutes) and the maximum was 1,187.7 seconds (19.8 minutes).
• Table 2 shows the subjects and the time it took each to successfully complete the tasks on the interface of the present invention. This table can be seen graphically in FIG. 4 , entitled “Differences In Seconds Between Total Time And (Total Time ⁇ Computer Time) For The Interface of the Present Invention.” Overall, 14 out of 16 of the test participants took less time using the interface of the present invention.
• the data was analyzed using SPSS-X, a statistical package.
• the tasks can be divided into six programming steps:
• FIG. 5 graphically shows the critical steps for the two interfaces.
• FIG. 6 shows the total time by subject. In FIG. 6 , subjects two and four took less time using the interface of the present invention in actuality, however, using adjusted times, the interface of the present invention took longer. In FIGS. 7 and 8 and show the minimum, maximum, average and standard deviations for both the prior art and the interface of the present inventions.
• the interface of the present invention reduced the average programming time by 54%. However, the standard deviations are high for both cases, and equal slightly more than half of the total time. T-Tests for each of the six programming steps showed no significance. However, when subjected to the Pearson Correlation, some significance was found (p ⁇ 0.10).
• the CHI Square test was performed to determine whether the probability of the times for each measurement is the same, or whether they are statistically different from one another. The results indicated that differences exist only between Clock Set, Program 2, and Program 3 and are significant at p ⁇ 0.01.
• T EXECUTE # OF KEYPRESSES ⁇ ( T M +T K +T P )
• T M Mentally Prepare
• T K Key in
• T P Point with mouse
• FIG. 9 shows keypresses per subject and FIG. 10 shows the differences between the theoretical and actual times for the keypresses, using Card's formulas, for each interface.
• the prototypes can be divided into screens which represent the programming steps.
• a minimum of 13 screens must be entered by the user.
• the average for the trials of the 16 subjects was 36.8 screens per subject, almost three times more screens than were necessary.
• Table 5 shows the screens, the minimum number of times they must be accessed, the average number of times that each was accessed, the average amount of time spent on each screen, and the standard deviation of the number of screens opened.
• Help screens were accessed either when the user was confused as to what to do, or just for curiosity (from users' comments).
• the one “Help” button that provided the most assistance was MAIN MENU HELP. It told the users to “POWER OFF” and then to set the Timer to “ON” when programming was finished.
• this button was pressed an average of 9.1 times by each the 16 subjects, which is almost double the number of keypresses required.
• Each keypress should theoretically have taken 2.1 seconds. However, in actuality, it took 12.1 seconds per “C” keypress, which is almost six times more than the theoretical value calculated using Card's formula.
• the extra keypresses can be attributed to the fact that five users thought that they had to press the “C” button after each category they changed.
• the extra time can be attributed to the fact that, as many subjects commented, they were doing the tasks by trial and error and had to recheck all of the information carefully to make sure it was correct. While test participants were using the prior art interface, many made comments that the numbers seemed to be changing by themselves. What had really happened was that the test participants were not looking at the correct place on the screen to see the changes and were confused by the blinking numbers, or they had entered information faster than it could be displayed, and as a result, pressed too many buttons and passed the desired choice.
• the interface of the present invention used a blue “OK” button on the bottom of each screen that required more than one keypress.
• These screens included: the current year, current time, start time, stop time, channel, and the confirmation screen. Pressing “OK” either brought the user to the next step of the programming sequence or back to the Main Menu from the confirmation screen. It was necessary for each subject to press this button a minimum of 14 times to complete the task, which would theoretically have taken 37.1 seconds. In the testing situation, “OK” was pressed an average of 18.5 times per subject. This was only 33% more than the minimum number of keypresses required. The average time was 6.9 seconds per “OK” pressed, which was 2.6 times more than the theoretical 2.65 per keypress found by applying Card's formula.
• the interface of the present invention took considerably less time. If the user is confident about the information he enters, it takes less time to confirm the data entered. When subjected to a t-Test, there was no significance for the number of times that “C” or “OK” was pressed, or between the time that it took for the two buttons to be pressed.
• the interface of the present invention reduced the number of incorrect recordings by 50%.
• the severity of the errors is unimportant here because one wrong entry will cause an irretrievable mistake and the user will not record the intended program.
• the “smart screen” aspect of the present invention is further explored in the present example.
• This aspect of the invention allows the interface to anticipate or predict the intent of the user, to provide, as a default user choice, the most likely action to be taken by the user of the programmable device as a default, which may be either accepted or rejected by the user, without delay to the user.
• the intelligent selection feature may also automatically choose an option and execute the selected option, without further intervention.
• VCR device When a user regularly applies the VCR device, for example, to record a given television show is which appears weekly on a given television channel, at a given time, on a given channel, such an action could be immediately presented to the user as a first option, without forcing him to explicitly program the entire sequence.
• the interface could actively determine whether the desired show is preempted, a repeat, changed in time or programming slot, etc.
• the interface could present information to the user, of which he might not be aware, and predict an action based on that information.
• Such a device could, if set in a mode of operation that allows such, automatically execute a sequence of instructions based on a predicted course of action.
• a user if a user is to be absent for a period, he could set the machine to automatically record a show, even if the recording parameters are not known at the time of setting.
• An on-line database system of known type may be used and need not be described in detail herein.
• the smart screens may be implemented as follows.
• the controller may be, for example, a Macintosh ci computer, operating under Macintosh 7.0 operating system.
• the Hypercard 2.0 software may be used to implement the screen interface, which incorporates the above-described features, which is generally compatible with the Hyperpad software described above. HyperCard is mentioned due to its capabilities to reference external programs, thus allowing interfacing to various software and hardware devices.
• a more global scripting language such as Frontier by UserLand Software Inc., may also be used, especially where low level hardware control of interfaced devices, such as a VCR, multimedia adapter, or the like is desired.
• Other scripting languages include versions of REXX, by IBM, available on many platforms.
• the input device is an Apple ADB mouse
• the output display is an 8 bit or 24 bit graphics color adapter connected to, e.g., a 14′′ color monitor.
• various parameters concerning the use of the interface are stored in the computer's memory, and a non-volatile mass storage device, such as a hard disk drive, or EEPROM or EPROM, as well as battery backed RAM could also be used.
• a predicted course of action or operation may be realized. This predicted operation is, in the context of the current user interface state, the most probable next action to be taken by the user.
• the predicted operation is based on: the identity of the user, if more than one user operates the interface and machine, the information already entered into the interface during the present programming is session, the presently available choices for data entry, settings for the use of the machine, which may be present as a result of a “setup” operation, settings saved during a prior session, and a database of programming choices.
• the interface software calls another program which has access to the necessary data in the memory, as well as access to any remote database which may be necessary for the function.
• a predictive technology such as Boolean logic, fuzzy logic, neural network logic, or other type of artificial intelligence, a most probable choice may be presented to the user for his approval, or another alternative choice may be selected.
• a number of most probable choices may be presented simultaneously or in sequence, in order to improve the probability that the user will be immediately or quickly presented with an acceptable choice. If multiple choices are presented, and there is limited room on the display, two (or more) similar choices may be merged into a single menu selection, which may be resolved in a secondary menu screen.
• FIG. 24 shows a system for correlating a user's preferences with a prospective or real-time occurrence of an event.
• the input device 2401 which is a remote control with a pointing device, such as a trackball, provides the user's input to the control 2402 .
• the program is stored in a program memory 2403 , after it is entered.
• the control 2402 controls a plant 2404 , which is a VCR.
• the control also controls an on-screen programming interface 2405 , through which the user interactively enters the program information.
• Each program entry of the user is submitted to the user history database and preferences module 2406 , which may also receive explicit preference information, input by the user through the input device 2401 .
• the prospective and real time event characterization unit 2407 uses any and all information available in order to determine the character of a signal input, which is a video signal, from the signal receiver 2408 .
• a signal analyzer 2409 provides a preliminary analysis and characterization of the signal, which is input to the prospective and real time event characterization unit 2407 .
• the prospective and real time event characterization unit 2407 also interacts and receives an input from a telecommunication module 2410 , which in turn interacts and receives information from an on-line database 2411 .
• a user preference and event correlator 2412 produces an output relating to a relatedness of an event or prospective event and a user preference.
• the control 2402 determines that the event or prospective event is a likely or most likely predicted action.
• the prospective event discussed above refers to a scheduled event, which is likely to occur in the future.
• the characterization unit also has a local database 2413 for storing schedule information and the like.
• the following is an example of a user wishing to program the machine interface of a VCR to record, e.g., “Married With Children” (Fox, Sunday, 9:00 p.m.), every time it occurs.
• the interface would first perform a self diagnostic to determine whether the machine is set up and operating correctly. This would include a determination of whether the clock has been set and thereafter operating continuously. Of course, the clock could have, in practice, a battery to minimize the occurrence of problems relating to clock function.
• the interface would then, if the clock is not properly set, and if there is no telecommunication or other external means for automatically determining the exact time, present the user with a menu selection to set the proper time.
• the step may be deferred. If the machine has access to an external source of the exact time, it would then preferably access this source first. This could include a telephone connection to a voice line which repeats the time. The computer would then perform a speech recognition algorithm which would be used to determine the time. Such a speech recognition algorithm could also be used as a part of the user interface for other purposes. Alternatively, a modem could be used to obtain the time in digitally coded form, which would alleviate the need for speech recognition capabilities for this function. An on-line connection could also be used in order to obtain information concerning television scheduling. A further alternative would be to access a video signal which contains time information.
• the interface could tune to this channel, acquire a representation of the screen image, and perform a character recognition algorithm to determine the time.
• This character recognition algorithm could also be used to decipher information regarding programming schedules, which may appear on certain cable broadcast channels.
• the interface determines a need for setting of the clock, and then takes measures to fulfill the necessary function, through any and all available resources, which may include speech recognition, character recognition, digital telecommunication means, radio wave reception and interpretation, and links to other devices.
• the system next must determine what function the user wishes to perform. In this regard, if more than one user has access to the system, the user identifies himself to the interface, in a user identification step 1701 or an analogous action, which may be a coded entry, or a selection from the menu. If the interface has voice recognition capability, then the user may be recognized by his voice pattern, or merely by stating his name. The interface then accesses the memory for a profile of the past use of the machine by the user, which may include the entire prior history, relevant abstracts of the history, or derived user preferences, as shown in the personalized startup based on user profile step 1702 , which information is also stored and used in the past user history determining element 2107 . These choices differ in the amount is of storage necessary in order to retain the desired information.
• the VCR to record
• the VCR to record
• the most likely current predicted choice would be the NBC 11 o'clock news.
• the interface were to present a number of choices, having lower probability, then it would interpret the recording history to be “news” based on a database of broadcast information. Therefore, a prediction of lower probability would be ABC or CBS news at, e.g., 11:00 p.m., and the NBC news at, e.g., 5:00 p.m.
• these three choices would be initially presented to the user, along with a menu selection to reject these predicted choices.
• the user would select the “reject” selection, and would be presented with a next predicted desired menu choice.
• the user history in this case, does not provide for another choice of high probability, the user would be prompted to explicitly choose the program sequence by day, time, channel, and duration.
• the user would then enter the starting time for recording according to the methods described above.
• the interface would then search its databases regarding the user and broadcast listings to present a most likely choice, as well as all available alternatives. In this case, the user history is of little help, and is not used to predict. In other cases, the system would use its intelligence to “fill in the blanks”, which could, of course, be rejected by the user.
• the most likely choices would then be those programs that begin at the selected time. If the user had input the channel, instead of starting time, then the presented choices would be the broadcast schedule of the channel, e.g. Fox, for the selected day. The user then selects one of the available choices, which would complete the programming sequence. If no database of broadcasts is available, then the user must then explicitly define all parameters of the broadcast. When the programming is completed, the interface must then update its user database, prompt the user to set the VCR to record, by, e.g., inserting a blank or recordable tape.
• the user would then proceed to explicitly program the VCR interface to record “Married with Children” on Fox at 9:00 p.m. on Sunday evening. If a database is available, it might also show that “Married with Children” is also syndicated in re-runs, and therefore may be available on other channels at other times. Thus, during the subsequent session, both the premier showing and re-run of “Married With Children” would be available predicted choices, along with the 11 o'clock News on NBC.
• the interface would then characterize the program. This would include, for example, a characterization of the soundtrack, the background, foreground, actors and actresses present, credits, etc.
• the interface would then attempt to is correlate the features present in the reference selection with other available selections. This comparison may be with a preformed database, providing immediate results, or prospectively, after entry of the reference selection. Of course, a number of correlation functions may proceed simultaneously, and various choices may be merged to form a compound reference selection. Further, as various “episodes” of the reference selection occur, the system appends and integrates the most recent occurrence with the stored reference information.
• the interface may be of further help.
• the interface may control a plurality of tuner elements 2502 of a video signal reception device 2501 , so that a plurality of broadcasts may be simultaneously received.
• a plurality of broadcasts may also be recorded simultaneously in the intermediate storage 2503 .
• the mass storage may be multiple VCRs, optical storage, or magnetic storage, including disk and tape.
• the optical recording tape produced by ICI, Inc. might also be a useful storage medium for large volumes of data, as might be generated by recording multiple video signals.
• the interface 2506 would access its associated database 2413 to determine, at a given time, which channels have “news”.
• the interface could also randomly or systematically monitor broadcasts for “special reports”. The interface would then monitor these channels for indicia of a “weather” broadcast. For example, the newscaster who appears to report the weather on a given show is usually the same, so that a pattern recognition system 2505 of the video frame could indicate the presence of that newscaster. In addition, the satellite photographs, weather radar, computer generated weather forecast screens, etc. are often similar for each broadcast. Finally, news segments, such as “weather” often appear at the same relative time in the broadcast. Using this information, the interface could begin recording at a beginning of a news segment, such as “weather”, stop recording during commercials, and continue recording after return from break, on all selected channels.
• the system of the present invention is intelligent, and may therefore “learn” either explicitly, or through training. Therefore, if the system made an error during the process, the user would define the error to the system, e.g., a substitute newscaster or rearrangement of news segments, so that the system is less likely to make the same error again.
• the interface provides simplicity, with inductive reasoning and deductive reasoning.
• a further example of the use of the advanced intelligent features of the present invention would be the use of the system to record, e.g., “live” musical performances. These occur on many “talk” shows, such as “Tonight Show with Johnny Carson” (NBC, 11:30 p.m. to 12:30 p.m., weeknights), “Saturday Night Live” (NBC 11:30 p.m. to 1:00 a.m. Saturday-Sunday), and other shows such as the “Grammy awards”.
• the interface if requested by the user to record such performances, would seek to determine their occurrence by, e.g.: analyzing a broadcast schedule; interacting with the on-line database 2411 ; and by reference to the local database 2413 .
• the interface When the interface determines with high probability that a broadcast will occur, it then monitors the channel(s) at the indicated time(s), through the plurality of tuners 2502 .
• an encryption/decryption unit 2509 is provided for decrypting the transmitted signal for analysis and viewing. This unit also allows encryption of material in other modes of operation.
• the interface system acquires the audio and video information being broadcast, through the signal receiver 2408 , and correlates this information with a known profile of a “live musical performance”, in the preference and event correlator 2412 .
• the system selects the broadcast for retention.
• the information in the intermediate storage 2503 is transferred to the plant 2507 , which includes a permanent storage device 2508 .
• the intermediate storage 2503 medium is used to record a “buffer” segment, so that none of the broadcast is lost while the system determines the nature of the broadcast. This, of course, allows an extended period for the determination of the type of broadcast, so that, while real-time recognition is preferred, it is not absolutely necessary in order to gain the advantages of the present invention.
• the determination storage need not be uncompressed nor lossless, so long as features necessary to determine the character of the broadcast are present.
• the storage be as accurate as possible, so that if a compression algorithm is implemented, it be as lossless as possible.
• the MPEG II standard would be applicable in this situation.
• broadcast material In a preferred situation, approximately 5 minutes of broadcast material is analyzed in order to make a determination of the content.
• This material is stored in two media. First, it is stored by normal means on video tape. Second, it is received in parallel by the computer control, where the data is subject to a number of recognition and characterization processes. These are performed in parallel and in series, to form an extracted feature storage matrix.
• a preferred method incorporates one or more digital signal processor based coprocessor elements, which may be present on, e.g., Nubus cards in the Macintosh ci or other computer type. These elements may be based on C-Cube CL550 (JPEG compression), AT&T DSP32C, AT&T DSP3210, AMD 29000 series, Motorola DSP 96000ADS, Texas Instruments TMS 32050, etc, or a combination of types.
• a typical board containing a DSP is the MacDSP3210 by Spectral Innovations Inc., containing an AT&T digital signal processor and an MC68020 CISC processor, and uses Apple Real-time Operating System Executive (A/ROSE) and Visible Cache Operating System (VCOS).
• A/ROSE Apple Real-time Operating System Executive
• VCOS Visible Cache Operating System
• processors employed be optimized for image processing, because of their higher throughput in the present applications, to process the video signals, and more general purpose signal processors to analyze the audio signals, because of the greater availability of software to analyze audio signals on these processors, as well as their particular strengths in this area.
• An array processor which may be interfaced with a Macintosh is the Superserver-C available from Pacific Parallel Research Inc., incorporating parallel Inmos Transputers. Such an array processor may be suitable for parallel analysis of the image segment and classification of its attributes.
• Pattern recognition especially after preprocessing of the data signal by digital signal processors and image compression engines, may also be assisted by logical inference engines, such as FUTURE (Fuzzy Information Processing Turbo Engine) by The Laboratory for International Fuzzy Engineering (LIFE), which incorporates multiple Fuzzy Set Processors (FSP), which are single-instruction, multiple data path (SIMD) processors.
• LIFE International Fuzzy Engineering
• FSP Fuzzy Set Processors
• SIMD single-instruction, multiple data path
• the processing system may provide a is best fit output to a set of inputs more efficiently than standard computational techniques, and since the presently desired result requires a “best guess”, rather than a very accurate determination, the present interface is an appropriate application of this technology.
• these processors may also serve other functions such as voice recognition for the interface, or extracting text from video transmissions and interpreting it.
• various methods are available for determining a relatedness of two sets of data, such as an image or a representation of an image. These include the determination of Hausdorff distance, fuzzy correlation, arithmetic correlation, mean square error, neural network “energy” minimization, covariance, cross correlation, and other known methods, which may be applied to the raw data or after a transformation process, such as an Affine transformation, a Fourier transformation, a warping transformation, a color map transformation, and the like. Further, it is emphasized that, in image or pattern recognition systems, there is no need that the entire image be correlated or even analyzed, nor that any correlation be based on the entirety of that image analyzed. Further, it is advantageous to allow redundancy, so that it is not necessary to have unique designations for the various aspects of the data to be recognized, nor the patterns to be identified as matching the uncharacterized input data.
• the MSHELL from Applied Coherent Technology is a software system that runs on a Mercury MC3200 array processor, in conjunction with a Data Translation DT2861 or DT2862.
• the NDS1000 Development System from Nestor, Inc. provides image recognition software which runs on a PC compatible computer and a Data Translation DT2878.
• the above mentioned processing hardware and software, as known, is incorporated herein.
• the C-Cube CL550 is fully disclosed in “C-Cube CL550 JPEG Image Compression Processor”, Preliminary Data Book, August 1991, and addendum dated Nov. 20, 1991, incorporated herein by reference, and products incorporating the CL550 include the JPEG Video Development Kit (ISA bus card with Chips and Technologies PC video 82C9001A Video Window Controller), and the C-Cube CL550 Development Board/PC for ISA Bus (CL550, for use with Truevision TARGA-16 or ATVista cards) or for NuBus (Macintosh).
• the so-called C-Cube “CL950” (unofficially announced) is a MPEG decoder device. Such a device as the CL950 may be particularly useful for use in the present VCR for reproducing is compressed program material, which may be compressed by the present apparatus, or may be used for decompressing pre-compressed program material.
• the interface of the present invention incorporates an intelligent user interface level determination. This function analyzes the quality of the user input, rather than its content. Thus, this differs from the normal interface user level determination which requires an explicit entry of the desired user level, which is maintained throughout the interface until explicitly changed.
• the present interface may incorporate the “smart screen” feature discussed above, which may, through its analysis of the past user interaction with the interface predict the most likely predicted user input function.
• the predictive aspects of Example 1 may be considered a species of the intelligent user level interface of Example 2. However, the following better serves to define this aspect of the invention.
• the input device in addition to defining a desired command, also provides certain information about the user which has heretofore been generally ignored or intentionally removed.
• this information includes a velocity component, an efficiency of input, an accuracy of input, an interruption of input, and a high frequency component of input.
• FIG. 21 which has a speed detector 2104 , a path optimization detector 2105 , a selection quality detector 2106 , a current programming status 2108 , an error counter 2109 , a cancel counter 2110 , a high frequency signal component detector 2112 , an accuracy detector 2113 and a physio-dynamic optimization detector 2114 .
• FIG. 21 also shows that the interface also uses a past user history 2107 , an explicit user level choice 2111 and an explicit help request 2115 .
• the velocity or speed component refers to the speed of movement of the sensing element, i.e. the rotating ball. This may also be direction sensitive, i.e., velocity vector. It is inferred that, all other things being equal, the higher the velocity, the more likely that the user “knows” what he is doing.
• the efficiency of input refers to two aspects of the user interface. First, it refers to the selection of that choice which most simply leads to the selection of the desired selection. For example, if “noon” is an available choice along with direct entry of numbers, then the selection of “noon” instead of “12:00 p.m.” would be more efficient.
• the second aspect of efficiency has to do with the path taken by the user in moving a menu selection cursor from a current position to a desired position. For example, a random curve or swiggle between locations is less efficient than a straight line. This effect is limited, and must be analyzed in conjunction with the amount of time it takes to move from one location of a cursor on the screen to another; if the speed of movement is very rapid, i.e.
• an inefficiency in path is likely due to the momentum of the mouse and hand, momentum of the rolling ball, or a physiological arc of a joint.
• This aspect is detected by the physio-dynamic optimization detector 2114 .
• arcs of movement, as well as uncritical damping of movement around the terminal position may be more efficient, and a straight path actually inefficient, so that the interface may act accordingly where indicated.
• an “efficient” movement would indicate an user who may work at a high level, and conversely, an inefficient movement would indicate a user who should be presented with simpler choices.
• FIG. 21 also shows the use of a past user history 2107 , an explicit user level choice 2111 and an explicit help request 2115 .
• the interface may incorporate screen buttons which are smart, i.e. those which intelligently resolve ambiguous end locations
• the accuracy of the endpoint is another factor in determining the probable level of the user.
• a 14′′ color monitor screen having a resolution of 640 by 480 pixels
• an accurate endpoint location would be within a central area of a screen button of size about 0.3′′ by about 1.0′′, would be an area of about 0.25′′ by about 0.75′′.
• a cursor location outside this location, but inside the screen button confines would indicate an average user, while a cursor location outside the screen button may be inferred to indicate the button, with an indication that the user is less experienced in using the pointing device.
• a high frequency component may be extracted from the pointer signal by the high frequency signal component detector 2112 , which would indicate a physical infirmity of the user (tremor), a distraction in using the interface, indecision in use, or environmental disturbance such as vibration.
• the presence of a large amount of high frequency signal indicates that, at least, the cursor movement is likely to be inaccurate, and possibly that the user desires a lower user level. While this is ambiguous based on the high frequency signal content alone, in conjunction with the other indicia, it may be interpreted.
• the response of the user level adjust system would be to provide a screen display with a lowered required accuracy of cursor placement, without necessarily qualitatively reducing the implied user level of the presented choices, thus, it would have an impact on the display simplification 2103 , with only the necessary changes in the current user level 2101 .
• the adaptive user level interface is of use in uncontrolled environments, such as in a moving vehicle, especially for use by a driver.
• An intelligent system of the present invention would allow the driver of such a vehicle to execute control sequences, which may compensate for the limited ability to interact with an interface while driving.
• the driver need not explicitly control all individual elements, because the driver is assisted by an intelligent interface.
• the interface would predict the windshield wipers should be actuated, the windows and any roof opening closed, and possibly the headlights activated.
• the driver could immediately assent to these actions, without individually actuating each control.
• the screen interface would provide a small number of choices, which may be simply selected.
• the present interface could be used to control complex telecommunications functions of advanced telephone and telecommunications equipment.
• the user display interface would be a video display, or a flat panel display, such as an LCD display.
• the interface would hierarchically present the available choices to the user, based on a probability of selection by the user.
• the input device would be, for example, a small track ball near the keypad.
• Another aspect of the present invention relates to the cataloging and indexing of the contents of a storage medium.
• random access media normally incorporate a directory of entries on a disk
• devices such as optical juke boxes normally are used in conjunction with software that indexes the contents of the available disks
• serial access mass storage devices such as magnetic tape
• an area of the tape is encoded to hold information relating to the contents of the tape.
• This encoding is shown in FIG. 19 , which shows a data format for the information.
• This format has an identifying header 1901 , a unique tape identifier 1902 , an entry identifier 1903 , a start time 1904 , an end time 1905 and/or a duration 1906 , a date code 1907 , a channel code 1908 , descriptive information 1909 of the described entry, which may include recording parameters and actual recorded locations on the tape, as well as a title or episode identifying information, which may be a fixed or variable length entry, optionally representative scenes 1910 , which may be analog, digital, compressed form, or in a form related to the abstract characterizations of the scenes formed in the operation of the device.
• error correcting codes 1911 for the catalog entry may also include advanced block encoding schemes to reduce the affect of non-Gaussian correlated errors which may occur on video tape, transmission media and the like.
• This information is preferably a modulated digital signal, recorded on, in the case of Hi-Fi VHS, one or more of the preexisting tracks on the tape, including the video, overscan area, Audio, Hi-Fi stereo audio, SAP or control tracks. It should be noted that an additional track could be added, in similar fashion to the overlay of Hi-Fi audio on the video tracks of Hi-Fi VHS. It is also noted that similar techniques could be used with Beta format, 8 mm, or other recording systems, to provide the necessary indexing functions.
• the recording method is preferable a block encoding method with error correction within each block, block redundancy, and interleaving.
• Methods are known for reducing the error rate for digital signals recorded on unverified media, such as videotape, which are subject to burst errors and long term non-random errors. Such techniques reduce the effective error rate to acceptable levels. These are known to those skilled in the art and need not be discussed herein in detail.
• a standard reference related to this topic is Digital Communications by John G. Proakis, McGraw-Hill (1983), which is incorporated herein by reference.
• the digital data recording scheme is best determined according to the characteristics of the recording apparatus. Therefore, if an, e.g.
• the catalog of entries is also preferably stored in non-volatile memory, such as hard disk, associated with the VCR controller.
• non-volatile memory such as hard disk
• This allows the random selection of a tape from a library, without need for manually scanning the contents of each tape.
• This also facilitates the random storage of recordings on tape, without the requirement of storing related entries in physical proximity with one another so that they may be easily located.
• This allows more efficient use of tape, because of reduced empty space at the end of a tape.
• the apparatus is shown schematically in FIG. 20 , in which a tape drive motor 2001 , controlled by a transport control 2002 , which in turn is controlled by the control 2003 , moves a tape 2005 past a reading head 2004 .
• the output of the reading head 2004 is processed by the amplifier/demodulator 2006 , which produces a split output signal.
• One part of the output signal comprises the analog signal path 2007 , which is described elsewhere.
• a digital reading circuit 2008 transmits the digital information to a digital information detecting circuit 2009 , which in turn decodes the information and provides it to the control 2003 .
• the user In order to retrieve an entry, the user interacts with the same interface that is used for programming the recorder functions; however, the user selects different menu selections, which guide him to the available selections.
• This function instead of focusing mainly on the particular user's history in order to predict a selection, would analyze the entire library, regardless of which user instituted the recording. Further, there would likely be a bias against performing identically the most recently executed function, and rather the predicted function would be an analogous function, based on a programmed or inferred user preference. This is because it is unlikely that a user will perform an identical action repeatedly, but a pattern may still be derived.
• the present library functions differ from the prior art VHS tape index function, because the present index is intelligent, and does not require the user to mark an index location and is explicitly program the VCR to shuttle to that location. Rather, the index is content based.
• Another advantage of the present library function is that it can automatically switch media. Such a system might be used, for example, if a user wishes to record, e.g., “The Tonight Show With Johnny Carson” in highly compressed form, e.g. MPEG at 200:1 compression, except during the performance of a musical guest, at which time the recording should be as lossless as possible.
• a normal VCR could hardly be used to implement such a function even manually, because the tape speed (the analogy of quality level) cannot be changed in mid recording.
• the present system could recognize the desired special segment, record it as desired, and indicate the specific parameters on the information directory.
• the recorded information may then be retrieved sequentially, as in a normal VCR, or the desired selection may be preferentially retrieved.
• the interface of the present invention is set to automatically record such special requests, the catalog section would then be available for the user to indicate which selections were recorded based upon the implicit request of the user. Because the interface has the ability to characterize the input and record these characterizations in the index, the user may make an explicit request different from the recording criteria, after a selection has been recorded. The controller would then search the index for matching entries, which could then be retrieved based on the index, and without a manual search of the entire tape.
• Other advantages of the present system are obvious to those of ordinary skill in the art.
• a library system is available from Open Eyes Video, called “Scene Locator”, which implements a non-intelligent system for indexing the contents of a videotape. See NewMedia, November/December 1991, p. 69.
• standard audio frequency modems and recording/receiving methods are available. These standard modems range in speed from 300 baud to 19,200 baud, e.g. v.FAST, v.32bis, etc. While these systems are designed for dial-up telecommunications, and are therefore slower than necessary and incorporate features unnecessary for closed systems, they require a minimum of design effort and the same circuitry may be multiplexed and also be used for telecommunication with an on-line database, such as a database of broadcast listings, discussed above.
• Videotext standard may also be used to record the catalog or indexing information on the tape. This method, however, if used while desired material is on the screen, makes it difficult to change the information after it has been recorded, because the videotext uses the video channel, during non-visible scan periods thereof.
• on-line database listings may be used by the present interface to provide information to be downloaded and incorporated in the index entry of the library function, and may also be used as part of the intelligent determination of the content of a broadcast. This information may further be used for explicitly programming the interface by the user, in that the user may be explicitly presented with the available choices available from the database.
• the present invention may incorporate character recognition from the video broadcast for automatic entry of this information. This is shown schematically in FIG. 24 , with the inclusion of the videotext and character recognition module 2414 . This information is shown to be transmitted to the event characterization unit 2407 , where the detected information is correlated with the other available information. This information may also be returned to the control 2402 . Examples of the types of information which would be recognized are titles of shows, cast and crew from programming material, broadcast special alerts, time (from digital display on special access channels), stock prices from “ticker tape” on special access channels, etc. Thus, this technology adds functionality to the interface. In addition, subtitled presentations could be recognized and presented through a voice synthesizer, to avoid the necessity of reading the subtitle. Further, foreign language subtitles could be translated into, e.g., English, and presented.
• the character recognition is performed in known manner on a buffer memory containing a frame of video, from a device such as a Data Translation DT2851, DT2853, DT2855, DT2867, DT2861, DT2862 and DT2871.
• a contrast algorithm run on, for example, a Data Translation DT2858, DT2868, or DT2878, first removes the background, leaving the characters. This works especially well where the characters are of a single color, e.g. white, so that all other colors are masked. After the “layer” containing the information to be recognized is masked, an algorithm similar to that used for optical character recognition (OCR) is employed. These methods are well known in the art.
• This may be specially tuned to the resolution of the video device, e.g. NTSC, S-VHS, IDTV, EDTV, MUSE, PAL, SECAM, etc.
• a spatial-temporal image enhancement algorithm may be employed to improve the quality of the information to be recognized.
• the present invention may also be incorporated into other types of programmable controls, for example those necessary or otherwise used in the control of a smart house. See, “The Smart House: Human Factors in Home Automation”, Human Factors in Practice, December 1990, 1-36.
• the user interface in such a system is very important, because it must present the relevant data to the user for programming the control to perform the desired function.
• a smart house would likely have many rarely used functions, so that the presentation of both the data and the available program options must be done in the simplest manner consistent with the goal of allowing the user to make the desired program choice.
• a smart house system might be used to execute the program “start dishwasher, if more than half full, at 9:00 p.m.” A user who wishes to delay starting until 11:00 p.m.
• the smart house system also controls the climate control system. Thus, it could coordinate temperatures, air flow and other factors, based on learned complex behaviors, such as individual movement within the dwelling. Since the goal of the programming of the smart house is not based on the storage of discrete information, but rather the execution of control sequences at various times and under certain circumstances, the control would differ in various ways from that of a VCR. However, the user interface system, adaptive user level, help system, and the like would be common to both types of system. This differs from the Fuzzy Logic controlled air conditioner available (in Japan) from Mitsubishi in that that device does not have an intelligent interface of the present invention. It should also be noted that the control for the VCR could be the same control as that for the smart house, so that the common elements are not redundant. Therefore, by applying a single control to many tasks, a common user interface is used, and the cost is reduced.
• the present Example relates to a programmable environmental controller application.
• a sensor or sensor array is arranged to detect a change in the environment which is related to a climatic condition, such as an open door.
• the system would apply a pattern recognition analysis to recognize this particular sensor pattern, i.e. a mass of air at a different temperature entering the environment from a single location, or a loss of climate controlled air to a single location.
• These sensor patterns must be distinguished from other events, such as the action of appliances, movement of individuals in the vicinity of the sensor, a shower and other such events.
• a neural network based adaptive controller may be more efficient, because the installation and design of such a system is custom, and therefore it would be difficult to program a priori.
• a learning system such as a neural network
• the training procedure could be fully automated, so long as sufficient sensors are provided for controlling the system, and also that an initial presumption of the control strategy is workable during the training period.
• the initial strategy incorporated is the prior art “bang-bang” controller, which operates as a simple thermostat, or multi-zone thermostat.
• a fuzzy logic temperature controller may be modeled and employed.
• Other known strategies which are not often used in environmental control include the proportional-integral-differential controller (PID).
• a sensor array 2301 sufficient sensors in a sensor array 2301 are provided, being light, temperature, humidity, pressure, air flow and possibly a sensor for determining an event proximate to the sensor, such as door opening. While a single sensor array 2301 could provide input to the present control, a plurality of sensor arrays are preferably employed in complex installations, such as that described here.
• the sensors with the possible exceptions of the flow sensor and event sensor, are housed in a single sensor head. Further, the temperature and pressure sensors may be combined in a single integrated circuit by known means.
• the light and temperature sensors are known to those skilled in the art, and need not be described herein.
• the pressure sensor may be a Sensym strain gage pressure transducer, a Motorola pressure transducer device, or the like, and may also be a derivative of the Analog Devices monolithic accelerometer. These devices are known in the art.
• the humidity sensor is preferably an electronic type, producing an electrical signal output. It need not be internally compensated for the other measured environmental factors.
• the air flow sensor may be based on pressure differentials, using the pressure sensor described above, or may be a mechanical vane type. In most applications, a single flow axis will be sufficient, however, in some circumstances, a two or greater axis sensor will be required. Further, in the case of large volume areas, complex turbulent flow patterns may be relevant, for which known sensors exist.
• the event sensor may be of any type, and depends particularly on the event being measured. In the present case, where a door opening is to be detected, it is preferred that the environmental control be interfaced with a perimeter intrusion alarm system, which, for example, provides a magnet embedded in the door and a magnetic reed switch in the door frame. Individual sensors are normally wired to the alarm control panel, thus providing central access to many or all of the desired event detection sensors while minimizing the added cost.
• the event detector may also be an ultrasonic, infrared, microwave-Doppler, mechanical, or other type of sensor.
• the preferred method of receiving sensor information is through a serial digital or multiplexed analog (i.e., 4-20 mA transmitter) data transmission scheme, with minimal local processing of the sensor data by the microprocessor 2302 with the serial link 2302 a in the sensor head.
• This system allows the central control 2303 to incorporate the desired processing, e.g., by the pattern recognition system 2304 , etc., while minimizing the installation expense.
• a simple microprocessor device 2302 in the sensor head interfaces the sensing elements, and may provide analog-to-digital conversion, or other conversion which may be necessary, of the sensor signal.
• the local microprocessor formats the sensor data, including a code indicating the sensor serial number and type, the sensor status (i.e., operative, defective, in need of maintenance or calibration, etc.), the sensor data, and an error correcting code.
• the microprocessor also arbitrates for bus usage and the messaging protocol.
• the control it must be understood, has a number of available operative systems at its disposal, comprising the plant 2306 .
• the system is a forced air heating and cooling system.
• This system has a heating unit, a humidifier, blowers, a cooling unit (which also dehumidifies), ducts, dampers, and possible control over various elements, such as automated door openers.
• the system is installed with a complete array of sensors, some of which may be shared with other control systems in the environment, and begins operation with a basic acceptable initial control protocol.
• the system receives data from the sensors, and correlates data from the various sensors, including the event sensors, with the operation of the systems being controlled. In such a case, a “door open” event may be correlated with a change in other measured variables.
• the system then correlates the control status with the effect on the interrelation of the measured variables.
• the system would detect that if the blower is operating while the door is open, then there is a high correlation that air will flow out of the door, unless a blower operates to recirculate air from a return near the door.
• the system will learn to operate the proximate return device while the door is open and the blower is on. Once this correlation is defined, the system may further interrelate the variables, such as a wind speed and direction outside the door, effects of other events such as other open doors, the absolute and relative speeds of the blowers and the return device, the effect of various damper devices, etc. It is further noted that, under some circumstances, an exchange of air through an open door is desired, and in such instance, the system may operate to facilitate the flow through such an open door. Finally, the system must be able to “learn” that conditions may exist which produce similar sensor patterns which should be handled differently. An example is a broken or inoperative sensor. In such a case, the system must be able to distinguish the type of condition, and not execute an aggressive control algorithm in an attempt to compensate for an erroneous reading or otherwise normal event. This requires the intelligent control of the present invention.
• auxiliary device which suitably affects only a part of the building.
• the aggregate efficiency of the system may be improved, even if the individual efficiency of an element is lower.
• the present intelligent control allows a fine degree of control, making use of all available control elements, in an adaptive and intelligent manner.
• the system would take appropriate action, including: interruption of normal climate control until after the disturbance has subsided and normal conditions are achieved; based on the actual climatic conditions or predicted climatic conditions begin a climate compensation control, designed to maximize efficiency and also maintain climatic conditions during the disturbance, as well as return to normal after the disturbance; optionally, during the door opening disturbance, the system would control a pressure or flow of air to counterbalance a flow through the door, by using a fan, blower or other device, or halting such a device, if necessary.
• the climatic control system could also be outfitted with actuators for opening and closing doors and windows, or an interface with such other system, so that it could take direct action to correct the disturbance, e.g., by closing the door.
• the climate between the internal and external ambients may differ in temperature, humidity, pollutants, or the like, and appropriate sensors may be employed.
• a remote control of the present invention may be constructed from, for example, a Micromint (Vernon, Conn.) RTC-LCD, RTC-V25 or RTC-HC11 or RTC180 or RTC31/52, and RTC-SIR, in conjunction with an infrared transmitter and receiver, input keys and a compatible trackball, which may provide raw encoder signals, or may employ a serial encoder and have a serial interface to the processor module.
• a power supply such as a battery, is used. The use, interfacing and programming of such devices is known to those skilled in the art, and such information is generally available from the manufacturer of the boards and the individual circuit elements of the boards.
• the function of such a remote control is to receive inputs from the trackball and keys and to transmit an infrared signal to the controller.
• the processor and display may provide added functionality by providing a local screen, which would be useful for programming feedback and remote control status, as well as compressing the data stream from the trackball into a more efficient form.
• certain of the extracted information may be relevant to the determination of the user level, so that information related to the user level would be analyzed and transmitted separately to the controller by the infrared transmitter. If the local LCD screen is used in the programming process, then the main controller would transmit relevant information to the remote display, by a reverse infrared link.
• the interface and intelligent control of the present invention are applicable to control applications in medicine or surgery.
• This system may also be described with reference to the generic system drawings of FIGS. 23 and 24 .
• an operator identifies himself and enters information regarding the patient, through the interface 2305 .
• the interface 2305 automatically loads the profile 2406 of both the operator and the patient, if the device is used for more than one at a time, and is connected to a database containing such information, such as a hospital central records bureau.
• the interface may be connected to various sensors, of the input device 2401 , such as ambient conditions (temperature, humidity, etc.), as well as data from the patient, such as EKC, EMG, EEG, Evoked Potentials, respirator, anesthesia, temperature, catheter status, arterial blood gas monitor, transcutaneous blood gas monitor, urinary output, IV solutions, pharmaceutical and chemotherapy administration data, mental status, movement, pacemaker, etc. as well as sensors and data sources separate from the patient such as lab results, radiology and medical scanner is data, radiotherapy data and renal status, etc.
• the interface 2405 using the simple input device and the display screen described above, presents the most important information to the operator, along with a most probable course of action.
• the system described has a large memory in the signal analysis module 2409 for recording available patient data from the signal receiver 2408 , and thus assists in medical record keeping and data analysis, as well as diagnosis. While various systems are available for assisting in both controlling medical devices and for applying artificial intelligence to assist in diagnosis, the present system allows for individualization based on both the service provider and the patient. Further, the present invention provides the improved interface for interaction with the system.
• medical events may be characterized in the characterization unit 2407 and recorded by the plant 2404 , so that a recording of the data need not be reviewed in its entirety in order to locate a particular significant event, and the nature of this event need not be determined in advance.
• the compression feature of the recorder of the present invention could be advantageously employed with the large volume of medical data that is often generated.
• the system could be used by the operator to create notes and discharge summaries for patients, using the database stored in the local database 2413 , as well as the user history and preferences 2406 .
• it would also perform an additional function, that of synthesizing the data, based on medical significance.
• the present example also comprises a control 2402 , and may interface with any of the sensors and devices, performing standard control and alarm functions.
• the present control 2402 is intelligent and has pattern recognition capability, in addition to full data integration from all available data sources, it may execute advanced control functions. For example, if the present control 2402 is interfaced to a controlled infusion pump for, e.g., morphine solution, in e.g., a terminally ill patient, then certain parameters must be maintained, while others may be flexible. For example, a maximum flow rate is established as a matter of practice as a safety measure; too high a flow rate could result in patient death.
• a patient may not need a continuous infusion of a constant dose of narcotic.
• the level of infusion may be advantageously altered.
• the controller had a urinary output monitor, it could immediately suppress the morphine infusion as soon as the renal output is recognized as being decreased, and further indicate an alarm condition.
• it may be advantageous to provide a diurnal variation in the infusion rate, to provide a “sleep” period and a period of heightened consciousness with correspondingly lower levels of narcosis.
• the control 2402 could be interfaced with a cardiac catheter monitor, as a part of the signal receiver 2408 .
• alarms are set based on outer ranges of each sensor measurement, and possibly a simple formula relating two sensor measurements, to provide a useful clinical index.
• the present device including the present control, may be more easily programmed to execute control and alarm functions, may provide a centralized source of patient information, including storage and retrieval, if diverse sources of such information are linked, and may execute advanced, adaptive control functions.
• the present control 2402 is equipped to recognize trends in the sensor data from the signal receiver 2408 , which would allow earlier recognition and correction of various abnormal conditions, as well as recognizing improvements in conditions, which could allow a reduction in the treatment necessary. Further, by allowing a fine degree of control, parameters may be maintained within optimal limits for a greater percentage of the time. In addition, by monitoring various sensors, various false alarms may be avoided or reduced. In particular, false alarms may occur in prior art devices even when sensors do not indicate a dangerous condition, merely as a safety precaution when a particular parameter is out of a specified range.
• a cause of such abnormal condition may be identified, such as patient movement or the normal activities of the patient's caretakers, then such condition may be safely ignored, without indicating an alarm.
• a sensor parameter does in and of itself indicate a dangerous condition
• the alarm may be ignored, or at least signaled with a different level of priority.
• the present invention is also of use in automated securities, debt, variable yield and currency trading systems, where many complex functions are available, yet often a particular user under particular circumstances will use a small subset of the functionality available at a given time. Such a situation would benefit from the present interface, which provides adaptive user levels, prioritized screen information presentation, and pattern recognition and intelligent control.
• a securities trading system is disclosed in U.S. Pat. No. 5,034,916, for a mouse driven Fast Contact Conversational Video System, incorporated herein by reference.
• the present system relates primarily to the user terminal, wherein the user must rapidly respond to external events, in order to be successful. In such a case, the advantages of the interface aspects are obvious, and need not be detailed herein.
• the pattern recognition functions of the present invention may be applied to correspond to the desired actions of the trader, unlike in prior intelligent trading systems, where the terminal is not individually and adaptively responsive to the particular user.
• the system exploits the particular strengths of the user, facilitating his actions, including: providing the desired background information and trading histories, in the sequence most preferred by the user; following the various securities to determine when a user would execute a particular transaction, and notifying the user that such a condition exists; monitoring the success of the user's strategy, and providing suggestions for optimization to achieve greater gains, lower risk, or other parameters which may be defined by the user.
• Such a system rather than attempting to provide a “level playing field”, allows a user to use his own strategy, providing intelligent assistance.
• the fractal method employing Affine transforms may be used to recognize images.
• This method proceeds as follows.
• a plurality of templates are stored in a memory device, which represent the images to be recognized. These templates may be preprocessed, or processed in parallel with the remainder of the procedure, in a corresponding manner.
• Image data which may be high contrast line image, greyscale, or having a full color map, the greyscale being a unidimensional color map, is stored in the data processor, provided for performing the recognition function.
• a plurality of addressable domains are generated from the stored image data, each of the domains representing a portion of the image information. It is noted that the entire image need not be represented, only those parts necessary for the recognition, which may be determined by known means.
• a plurality of addressable mapped ranges are created, corresponding to different subsets of the stored image data. Creating these addressable mapped ranges, which should be uniquely addressable, also entails the step of executing, for each of the mapped ranges, a corresponding procedure upon the one of the subsets of the stored image data which corresponds to the mapped ranges. Identifiers are then assigned to corresponding ones of the mapped ranges, each of the identifiers specifying, for the corresponding mapped range, a procedure and a address of the corresponding subset of the stored image data.
• the treatment of the template and the image data is analogous, so that the resulting data is comparable.
• the domains are optionally each subjected to a transform, which may be a predetermined rotation, an inversion, a predetermined scaling, and a predetermined frequency domain preprocessing transform.
• This transform is used to optimize the procedure, and also to conform the presentation of the image data with the template, or vice versa.
• Each of the domains need not be transformed the same way.
• the one of the mapped ranges which most closely corresponds according to predetermined criteria is selected.
• the image is then represented as a set of the identifiers of the selected mapped ranges.
• a template is selected which most closely corresponds to the set of identifiers representing the image information. It is preferred that, for each domain, a most closely corresponding one of the mapped ranges be selected.
• the mapped range is selected which is the most similar, by a method which is appropriate, and may be, for example, selecting minimum Hausdorff distance from the domain, selecting the highest cross-correlation with the domain, the minimum mean square error with the domain and selecting the highest fuzzy correlation with the domain. Neural network energy minimization may also yield the best fit, and other techniques may also be appropriate.
• the step of selecting the most closely corresponding one of mapped ranges according to the minimum modified Hausdorff distance includes the step of selecting, for each domain, the mapped range with the minimum modified Hausdorff distance calculated as D[db,mrb]+D[1 ⁇ db,1 ⁇ mrb], where D is a distance calculated between a pair of sets of data each representative of an image, db is a domain, mrb is a mapped range, 1 ⁇ db is the inverse of a domain, and 1 ⁇ mrb is an inverse of a mapped range.
• the method includes a matching of the color map, which as stated above, includes a simple grey scale.
• the method is modified to optionally transform the color map values of the pixels of each domain by a function including at least one scaling function, for each axis of said color map, each of which may be the same or different, and selected to maximize the correspondence between the domains and ranges to which they are to be matched.
• the one of the mapped ranges having color map pixel values is selected which most closely corresponds to the color map pixel values of the domain according to a predetermined criteria, wherein the step of representing the image color map information includes the substep of representing the image color map information as a set of values each including an identifier of the selected mapped range and the scaling functions.
• the correspondence method may be of any sort and, because of the added degree of complexity, may be a different method than that chosen for non-color images.
• the method of optimizing the correspondence may be minimizing the Hausdorff distance or other “relatedness” measurement between each domain and the selected range.
• the recognition method concludes by selecting a most closely corresponding stored template, based on the identifier of the color map mapped range and the scaling functions, which is the recognized image.
• the method is further modified to accommodate time varying images. These images usually vary by small amounts between frames, and this allows a statistical improvement of the recognition function by compensating for a movement vector, as well as any other transformation of the image. This also allows a minimization of the processing necessary because redundant information between successive frames is not subject to the full degree of processing. Of course, if the image is substantially changed, then the statistical processing ceases, and a new recognition function may be begun, “flushing” the system of the old values.
• the basic method is thus modified by storing delayed image data information, i.e., a subsequent frame of a moving image. This represents an image of a moving object differing in time from the image data in the data processor.
• a plurality of addressable further domains are generated from the stored delayed image data, each of the further domains representing a portion of the delayed image information, and corresponding to a domain.
• an analogous transform is conducted so that the further domains each are corresponding to a domain.
• a plurality of addressable mapped ranges corresponding to different subsets of the stored delayed image data are created from the stored delayed image data.
• the further domain and the domain are optionally matched by subjecting a further domain to a corresponding transform selected from the group consisting of a predetermined rotation, an inversion, a predetermined scaling, and a predetermined frequency domain preprocessing transform, which corresponds to a transform applied to a corresponding domain, and a noncorresponding transform selected from the group consisting of a predetermined rotation, an inversion, a predetermined scaling, a translation and a predetermined frequency domain preprocessing transform, which does not correspond to a transform applied to a corresponding domain.
• the one of the mapped ranges is selected which most closely corresponds according to predetermined criteria.
• a motion vector is then computed between one of the domain and the further domain, or the set of identifiers representing the image information and the set of identifiers representing the delayed image information, and the motion vector is stored.
• the further domain is compensated with the motion vector and a difference between the compensated further domain and the domain is computed.
• the one of the mapped ranges is selected which most closely corresponds according to predetermined criteria.
• the difference between the compensated further domain and the domain is represented as a set of difference identifiers of the selected mapping ranges and an associated motion vector.
• FIG. 27 is a basic flow diagram of the recognition system of the present invention.
• FIG. 28 provides a more detailed description, including substeps, which are included in the major steps shown in FIG. 27 .
• the image, or a part thereof is decomposed into a compressed coded version of the scene, by a modified fractal-based compression method.
• this differs from the prior compression algorithms in that only a part, preferably that part containing objects of interest, need be processed. Thus, if a background is known (identified) or uninteresting, it may be ignored. Further, the emphasis is on matching the available templates to produce an image recognition, not achieving a high degree of compression.
• the image, or domains thereof may be transformed as required in order to facilitate the matching of the templates.
• the templates are represented in analogous form, having been processed similarly, so that a comparison of the relatedness of an object in an image and the templates may be performed.
• an oblique view of an object is presented, then either the object may be transformed to achieve a predicted front view, or the template transformed or specially selected to correspond to the oblique view.
• the system need only ensure that the scene has not changed, and need not continually process the data. This has implications where multiple recognition processes are occurring simultaneously, either in a single scene or in different images, wherein the throughput of the recognition apparatus need not meet that required for de novo recognition of all aspects of all the objects or images.
• FIG. 30 shows a flow diagram of a cartoon-like representation of an image recognition method of the present invention. It shows initially, an input image 3001 , having a degree of complexity. A windowing function 3002 isolates the object from the background. A first order approximation of the image is generated 3003 , here called a mapping region. The first order approximation is then subtracted from the initial image to produce a difference 3004 . The first order error is then subjected, iteratively, to successive transform and subtract operations 3005 and 3006 , until the error is acceptably small, at which point the input image is characterized by a series of codes, representing the first order approximation and the successive transforms, which are stored 3008 . These codes are then compared with stored templates 3009 . The comparisons are then analyzed to determine which template produces the highest correlation 3010 , and the match probability is maximized 3011 . The recognized image is then indicated as an output 3012 .
• a sensor 2602 provides data, which may be image data, to a control 2601 .
• the control 2601 serves to control the plant 2603 , which has an actuator.
• the plant 2603 may be a VCR or the like.
• the control 2601 has associated with it an intermediate sensor data storage unit 2611 , which may be, for example a frame buffer or the like.
• the control 2601 also has associated with it a transform engine 2612 , which may perform a reversible or irreversible transform on the data or stored data.
• the system also has a template input 2610 , which may receive data from the sensor 2602 , if accompanied by identifying information.
• the pattern storage memory 2609 stores a pattern, such as an image pattern, along with an identifier.
• the control 2601 also has an input device 2604 , an on-screen display interface 2605 , and a program memory 2606 , for inputting instructions from a user, providing feedback to the user, and recording the result of the user interaction, respectively.
• a characterization network 2607 characterizes the sensor 2602 data, which may be provided directly from the sensor 2602 or preprocessing circuitry, or through the control 2601 .
• a correlator 2608 correlates the output of the characterization network with the stored patterns, representing the templates from the template input 2610 . The system therefore operates to recognize sensor patterns, based on the correlator 2608 output to the control 2601 .
• the error between the movement and transform compensated delayed image and the image is quantified, to determine if the compensation is valid, or whether the scene is significantly changed.
• a template is selected, from the stored templates, which most closely corresponds or correlates with both the set of identifiers of the image data and the set of identifiers of the delayed image data, thus improving recognition accuracy, by allowing a statistical correlation or other technique.
• the system would score the first template as a better match to the first image.
• ADVANTAGES AND DISADVANTAGES DEVICE ADVANTAGES DISADVANTAGES Touch-Screen a device which allows users accurate. fast. “natural” pointing Doesn't show location of the cursor to point directly to the screen to enter their device. Hand obscures view on the screen. Requires an overlay. choices. Difficult with curved screens. Requires frequent cleaning. Expensive. Must be within reach envelope. Light Pen: a pen shaped device with which Points to the screen. Inexpensive. Inaccurate. Awkward to the users touch the screen to select their use. Pen needs a storage location. choices. Must be within reach envelope.
• Trackball a ball mounted on a stationary Can be mounted and used object; the ball's rolling motion controls the anywhere. Does not require a cursor. horizontal surface.
• Joystick a stick mounted on a stationary Can be mounted and used Clumsy for cursor control. object; the sticks movement controls the anywhere. cursor. Does not require a horizontal surface.
• Mouse a ball mounted on the bottom of a Most effective for pointing and Requires a horizontal surface area.
• movable object which is rolled on a selecting objects on the screen. horizontal surface to control the cursor.
• Touch-Tablet a pad which sits on a Activated with fingers or stylus. Small interface. horizontal surface on which selections are Remote from display. made by using a finger or stylus.
• Keyboard a device which lies on a Requires a horizontal surface. Large. horizontal surface and which has Many keys. alphanumeric keys on which to type information.
• Multi-Function Keys buttons which serve Inexpensive. Confusing. more than one function. Space efficient. Bar Code Scanner: a wand which must be Quick if Barcode is present in TV May require several tries to send wiped over a bar code to type enter directory. data. Tedious if Barcode is not information. Pressing a button then signals available in the TV directory. the controlling device.
• Voice the use of the human voice to give Frees hands. Requires training. Affected by speech prompts or to accept commands. Enables disabled persons to use surrounding noises. Low accuracy. the device. Expensive.
• Eye Tracker an optical scanner which is Frees hands. Expensive. activated by the human eye. Enables disabled persons to use Inaccurate. the device.
• Data Suit/Data Glove a suit or glove which is Reacts to hand and body Expensive. controlled by manipulation of an on-screen gestures. Computer intensive. “Virtual Image”. It is controlled by optical Gives a 3-D image. fibers which measure the degree of bending.

Landscapes

• Engineering & Computer Science (AREA)
• Theoretical Computer Science (AREA)
• Physics & Mathematics (AREA)
• General Engineering & Computer Science (AREA)
• General Physics & Mathematics (AREA)
• Software Systems (AREA)
• Mathematical Physics (AREA)
• Fuzzy Systems (AREA)
• Computing Systems (AREA)
• Artificial Intelligence (AREA)
• Evolutionary Computation (AREA)
• Data Mining & Analysis (AREA)
• Mechanical Engineering (AREA)
• Combustion & Propulsion (AREA)
• Chemical & Material Sciences (AREA)
• Computational Mathematics (AREA)
• Mathematical Analysis (AREA)
• Pure & Applied Mathematics (AREA)
• Mathematical Optimization (AREA)
• Automation & Control Theory (AREA)
• Life Sciences & Earth Sciences (AREA)
• Biomedical Technology (AREA)
• General Health & Medical Sciences (AREA)
• Molecular Biology (AREA)
• Algebra (AREA)
• Health & Medical Sciences (AREA)
• Signal Processing (AREA)
• Computational Linguistics (AREA)
• Computer Vision & Pattern Recognition (AREA)
• Medical Informatics (AREA)
• User Interface Of Digital Computer (AREA)
• Air Conditioning Control Device (AREA)

Abstract

Description