Discovision Associates

Kevin D Dewar

Martin William Sotheran

Anthony Mark Jones

Adrian Philip Wise

William P Robbins

Helen Rosemary Finch

Anthony Peter Claydon

David Andrew Barnes

Nicholas Birch

Mark Barnes

Andrew Peter Kuligowski

Colin Smith

Donald William Patterson

Anthony M Jones

An MPEG video decompression method and apparatus utilizing a plurality of stages interconnected by a two-wire interface arranged as a pipeline processing machine. Control tokens and DATA Tokens pass over the single two-wire interface for carrying both control and data in token format. A token decode circuit is positioned in certain of the stages for recognizing certain of the tokens as control tokens pertinent to that stage and for passing unrecognized control tokens along the pipeline. Reconfiguration processing circuits are positioned in selected stages and are responsive to a recognized control token for reconfiguring such stage to handle an identified DATA Token. A wide variety of unique supporting subsystem circuitry and processing techniques are disclosed for implementing the system, including memory addressing, transforming data using a common processing block, time synchronization, asynchronous swing buffering, storing of video information, a parallel Huffman decoder, and the like.

Technique for decoding variable and fixed length codes

Token-based adaptive video processing arrangement

A RAM implementation of asynchronous swing buffering is provided in which two buffers are operated asynchronously; one is written while the other is read. Accordingly, this allows for a data stream having a fast rate of through-put to be resynchronized to another rate, while still maintaining a desired rate. In the invention, the write control and read control both have state indicators for communicating which buffer they are using and whether the controls are waiting for access or are, in fact, accessing that buffer.

Technique for implementing a swing buffer in a memory array

Video decompression

A Huffman Decoder is provided for decoding a data stream of fixed length and variable length codes and is adapted for video decompression applications. In the case of variable length codes, a table index calculation is performed using a ROM whose addresses are controlled by a selector/shifter. This calculation is bypassed in the case of fixed length codes. An index-to-data circuit then yields decoded data.

Parallel Huffman decoder

Arrangement for processing packetized data

An IDCT, or Inverse Discrete Cosine Transform, method decimates a 2-D IDCT into two 1-D IDCT operations and then operates separately on the even and odd pixel input words. In a common processing step, selected input values are passed directly to output adders and subtractors, while others are multiplied by constant, scaled cosine values. In a pre-common processing step, the lowest-order odd input word is pre-multiplied by .sqroot.2, and the odd input words are summed pairwise before processing in a common processing step. In a post-common processing step, intermediate values corresponding to the processed odd input words are multiplied by predetermined coefficients to form odd resultant values. After calculation of the even and odd resultant values, the high-order and low-order outputs are formed by simple subtraction/addition, respectively, of the odd resultant values from/with the even resultant values. The input values are preferably scaled upward by a factor of .sqroot.2. Selected bits of some intermediate resulting data words are optionally adjusted by forcing these bits to either '1' or '0'. The IDCT system includes a pre-common processing circuit (PREC), a common processing circuit (CBLK), and a post-common processing circuit (POSTC), which perform the necessary operations in the respective steps. The system also includes a controller (CNTL) to generate signals to control the loading of system latches and, preferably, to time-multiplex the application of the even and odd input words to latches in the pre-common circuit.

Method and arrangement for transformation of signals from a frequency to

An IDCT method decimates a 2-D IDCT into two 1-D IDCT operations and then operates separately on the even and odd pixel input words. In a common processing step, selected input values are passed directly to output adders and subtractors, while others are multiplied by constant, scaled cosine values. In a pre-common processing step, the lowest-order odd input word is pre-multiplied by .sqroot. 2, and the odd input words are summed pairwise before processing in a common processing step. In a post-common processing step, intermediate values corresponding to the processed odd input words are multiplied by predetermined coefficients to form odd resultant values. After calculation of the even and odd resultant values, the high-order and low-order outputs are formed by simple subtraction/addition, respectively, of the odd resultant values from/with the even resultant values. The input values are preferably scaled upward by a factor of .sqroot. 2. Selected bits of some intermediate resulting data words are optionally adjusted by forcing these bits to either '1' or '0'. The IDCT system includes a pre-common processing circuit (PREC), a common processing circuit (CBLK), and a post-common processing circuit (POSTC), which perform the necessary operations in the respective steps. The system also includes a controller (CNTL) to generate signals to control the loading of system latches and, preferably, to time-multiplex the application of the even and odd input words to latches in the pre-common circuit.

Growing community of inventors

Bristol, United Kingdom

Kevin D Dewar