US 5,963,154 A · Technique for decoding variable and fixed length codes

Abstract

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.

Claims13

Dependent claims

In a video decoding and decompression system, a parallel Huffman decoder, comprising:

an input accepting a mixed data stream comprising Huffman coded variable length codes, fixed length codes, and data tokens;
a memory for "don't care" handling, comprising:
a set of memory address lines, inverted address lines and data lines, said address lines and inverted address lines being connected according to a decoding format to access the addressed information in the form of a data word, "don't care" address locations on the data lines being unconnected with the associated address lines and said inverted address lines, said memory being addressed according to said Huffman coded variable length codes, and having data stored therein representing decoded data and information representative of a length of the Huffman coded variable length codes;
a selector, responsive to an output of said memory, wherein in a first mode of operation variable length codes are provided to said memory address lines, and in a second mode of operation a signal representative of a length of fixed length codes is output by the decoder without reference to said memory.

1 more element

A method of decoding a mixed data stream comprising Huffman coded variable length codes, fixed length codes, and data tokens, the stream including a first code having a first length, and a second code having a second length, comprising the steps of:

memorizing data in a plurality of locations of a memory, wherein said memory has a plurality of address lines for addressing said locations, a first set of said address lines
and a second set of said address lines being defined to correspond with a pattern of bits of said first code and a pattern of bits of said second code respectively;
identifying variable length codes in the data stream;
respectively associating each bit of a said variable length code with a said address line;
activating said address lines responsive to the bits associated therewith;
coupling a first wordline driver to said first set of address lines;
coupling a second wordline driver to said second set of address lines;
activating a selected one of said first wordline driver and said second wordline driver in response to signals on all of their respectively coupled address lines;
obtaining data from a location in said memory that is driven by said activated wordline driver;
from said obtained data determining the length of the variable length code;
according to said determined length, aligning bits of a succeeding variable length code with said address lines for association therewith;
identifying fixed length codes in the data stream;
and outputting a decoded value thereof without accessing said memory.

10 more elements

Apparatus for decoding a mixed data stream, comprising:

a memory, having a table of data disposed in locations thereof, and a plurality of address lines for addressing said locations;
an input register for accepting the data stream, wherein the data stream comprises Huffman coded variable length codes, fixed length codes, and data tokens, and includes a first code having a first length, and a second code having a second length;
a selector for aligning the variable length codes in said input register with said address lines, wherein a first set of said address lines is defined to correspond to the pattern of bits of said first code and a second set of said address lines is defined to correspond to the pattern of bits of said second code;
said selector further comprising electrical circuitry for distinguishing variable length codes from fixed length codes;
a first wordline driver coupled to said first set of address lines;
a second wordline driver coupled to said second set of address lines, wherein a selected one of said first wordline driver and said second wordline driver is activated in response to signals on all of their respectively coupled address lines;
an output bus having data thereon derived from a location in said memory that is driven by said activated wordline driver;
means coupled to said output bus for determining the length of the variable length code, and coupled to said selector;
wherein in response to a fixed length code said detector outputs a signal representative of a length of the fixed length code without accessing said memory.

6 more elements

Dependencies are inferred from the claim text.

Technique for decoding variable and fixed length codes

Life of this patent

Technique for decoding variable and fixed length codes

Life of this patent

Abstract

Claims13