U.S. Philips Corporation

Hitachi, Ltd.

N.v. Philips' Gloeilampenfabrieken

Eric H Persoon

Christian J Vandenbulcke

Johannes A Bernsen

Geert Nijholt

Wilhelmus Jacobus Van Gestel

Eduard Ferdinand Stikvoort

Albert M Rijckaert

David E Penna

Peter H De With

Ronald W Saeijs

Junichi Higashino

Jurgen F Rosengren

Eddy A Odijk

Ratnayake M Abeysekera

Hendrikus Daalder

A device for deriving an ancillary signal from a compressed digital video signal (e.g. MPEG), wherein the ancillary signal includes selected parts of the main signal, for example, the DC coefficients of I-pictures, or the unscrambled parts, which parts can be used for display in a (multi-) picture-in-picture television receiver, or as an 'appetizer' in order to encourage the user to pay a subscription fee, is described. The ancillary signal can separately be recorded in digital video recorders so as to assist the user in finding the beginning of a scrambled program on tape. The ancillary signal can also be generated at the transmitter end and transmitted at a low bit rate. A decoder for decoding such an ancillary signal is considerably simpler and less expensive than a full-spec MPEG decoder. A decoding method is also described.

Method and apparatus for multiplexing and transmitting

A perceptron device has processing elements arranged in an input layer, one or more hidden layers, and an output layer. The training method of the perceptron device uses the back propagation method. In order to increase learning speed, the learning rate for the updates of a particular processing element is a function f(M,N,K) of the number N of inputs to that element, the number K of outputs of that element, and the number N of inputs to the next layer. Function f increases with increasing M, and decreases with increasing K and N.

Method for adjusting network parameters in a multi-layer perceptron

An image processing pipeline system and method utilizes a cascade of data processing modules for processing a stream of pixel data units organized in successive groups, which are presented to the system input, the groups being separated from one another by time slots consisting of dead time intervals produced, for example, by line or frame fly-back. The modules are controlled by control signals produced by sampling circuit type characteristic circuits in response to control pulses from a control signal source. The control signals applied to the modules are in synchronism with the presence of data units at the system input, and are absent during the dead time intervals. In order to avoid loss of information due to processing delays if data transport between the modules takes place during an occurrence of a dead time interval, a monitoring circuit stops or pauses processing by the modules during that time interval of data units to reach the system input prior to the interval.

Pipeline system which pauses processing of already input data during

Arrangements for recording and reproducing datareduced videoinformation are described. The recording arrangement comprises a datareduction unit (3) and a formatting unit (4). The formatting unit (4) generates packets (P.sub.i) each having a first, second and third packet portion (PP.sub.1, PP.sub.2 and PP.sub.3). The first packet portion comprises a sync word (SW) and an identification word. The second packet portion comprises datareduced videoinformation of a first and a second datablock (MB.sub.0, MB.sub.1). A datablock comprises DC coefficients (DC.sub.i) and most significant and least significant AC coefficients (MAC.sub.i, LAC.sub.i). The videoinformation comprised in the first and the second datablock are stored in the second packet portion (PP.sub.2), such that the DC coefficients of the first datablock (DC.sub.0) are stored first, the plurality of most significant AC coefficients (MAC.sub.0) of the first datablock are stored next, the DC coefficients of the second datablock (DC.sub.1) are stored last in the second packet portion, and the plurality of most significant AC coefficients (MAC.sub.1) of the second datablock are stored before the DC coefficients of the second datablock in the second packet portion. This enables an improved reproduction of the videosignal in a feature mode.

Arrangement for recording or reproducing data reduced video information

Method and apparatus for determining the presence of a predetermined pattern in a spatial field. The invention involves selecting a collection of amplitude values on the basis of location within a multivalued-amplitude field and arranging sequences of amplitude values in order of their magnitude. Amplitude values occurring at predetermined ranks are chosen as characteristic quantities representative of the portion of the field under consideration. By inter-relating a plurality of these characteristic quantities, pattern recognition is accomplished. Preferably rank value filters are employed for producing the characteristic quantities.

Method of processing pattern data on the basis of ranked amplitude

To generate quickly the several carry signals in a full adder circuit (for example a 40-bit circuit), the circuit is divided into a plurality of sub-circuits of a first type, in which complementary imaginary carry signals are generated parallel to each other. Carry look-ahead circuits are of a dual construction, each first carry look-ahead circuit receiving a logic '0' and each second carry look-ahead circuit receiving a login '1'. So the generated imaginary carry signals are complementary, from which the carry signal proper is selected with the aid of a multiplex switch. The multiplex switch is controlled by the carry signal generated in a preceding sub-circuit of the first type. Since a multiplex switch operates faster than 3-bit or 4-bit wide carry look-ahead circuits, which operate in parallel in groups, the carry signals are consequently generated faster.

Full adder circuit

A method of generating a two-dimensional image representing a three-dimensional scene is describerd. The method comprises: (a) storing a set of foreground pixel data and at least one set of background pixel data, each such set defining a color and a depth for each pixel of the image; (b) selectively modifying the stored foreground and background data in response to a series of input pixel data defining color and depth for a series of pixels of the image; and (c) forming output pixel data by combining the stored foreground and background color for each pixel in proportions defined by a mixing value associated with that pixel. In one embodiment, step (b) comprises selectively modifying the stored data depending on the input depths so that: (i) if the input depth is greater than the stored foreground and background depths, the stored foreground and background pixel data are left unchanged, (ii) if the input depth lies between the stored foreground depth and the stored background depth, the input pixel data is stored in place of the background data for that pixel, and (iii) if the input depth is less than both the stored foreground and background depths, the stored current foreground data for that pixel are stored in place of the stored background data and the input data are stored in place of the current foreground data. An arrangement is described for performing the method. The arrangement may form part of an electronic graphics system.

Method and apparatus for storing a two-dimensional image representing a

A pipeline system which is suitable for the processing of data accompanied by data identifying signals comprises interconnected modules which process the data and which controllably transfer the data to one another. Data processing in a module introduces a delay of the data. For correct data processing in a module, the data identifying signals must be in synchronism with the data. In a pipeline system in accordance with the invention, the data identifying signals are applied in parallel to all modules, these signals being delayed in a module by a delay means until the data whereto the signals relate have reached the relevant module. The foregoing has the advantage that the number of connections between the modules is reduced.

Pipeline system with parallel data identification and delay of data

A data transmission system comprising a control unit, a transmitter and a receiver, the transmitter and receiver being connected by a transmission line for the serial transmission of a series of data words in a sequence of subsequent data bits. The control unit is provided with a word select signal generator for generating a two level word select signal for selecting, under control of a level change, a data word to be transmitted. The time length between two successive level changes of the word select signal is variable.

Data transmission system having transmission intervals which are

An image processing device for two-dimensional raster images is described. The receiving means thereof are followed by a sub-sampling circuit (28) for increasing the pitch of the raster received in accordance with an adjustable scale. To the sub-sampling circuit there are connected a buffer (30, 32) and a correlation circuit (36, 38) for comparing a number of columns of pixels with a reference image. The comparison result is subjected to a logic operation. The device is suitable for modular use and may include a growth circuit for converting rasters portions onto one color or brightness when a predetermined minimum number of pixels therein contains the relevant color or brightness, respectively.

Image processing device for the real-time processing and recognition of

An integrated data processor which includes a multiplier element provided with non-symmetrical operand inputs and an arithmetic and logic unit which is connected to the output of the multiplier element by way of a connection over the full product width. The ALU also comprises an accumulator device. Multiple-precision operations can thus be performed on the quantities received on each of the two inputs of the multiplier element. The data processor is notably suitable for performing each time identical operations on sequentially arriving signal quantities.

Integrated processor for the processing of word-wise receivable data

A device for binarizing an image consisting of grey values by means of a dynamically adjusted black/white discrimination threshold comprises:

Device for the dynamic adjustment of a black/white discrimination

A device for recognizing overall patterns of digital signals which are arranged according to at least one coordinate comprises: first storage means for the storage of a pattern to be presented for comparison; a window device for selecting a local pattern of a number of digital signals from the overall pattern; a stepping mechanism for coordinate-wise adjustment of the window device over a step; a gradient determining device for determining, for predetermined coordinate values within the window, the absolute value of the coordinate-wide gradient of the values of the digital signals prevailing at said coordinate values; a moment generator for receiving these absolute values, for determining the moment therefrom with respect to a focussing point of the window, and for generating an output signal when a predetermined minimum value of said moment is reached in order to indicate the reaching of a local maximum in said coordinate-wise gradient; second storage means for the storage of a number of reference patterns; and a difference determining device for comparing, under control of the output signal of the moment generator, a local pattern selected by the window device with the reference patterns and for alternatively forming a non-correspondence/correspondence signal. If the difference is too large, the 'new' pattern is stored as a further reference, so that the learning phase and the use phase occur simultaneously. The comparison does not take into account a difference near substantially coincident acute amplitude gradients.

Growing community of inventors

Eindhoven, Netherlands

Eric H Persoon