Position coding for context-based adaptive variable length coding

A non-transitory computer-readable storage medium storing instructions that when executed by one or more processors cause carrying out a method, the method comprising:

• for an ordered sequence of quantized transform coefficients of a block of image data, the sequence having a most likely-to-occur amplitude, and at least one other amplitude including a next-to-most likely-to-occur amplitude, encoding the ordered sequence using a context-based adaptive variable length coding method, the context-based adaptive variable length coding method including:
• encoding the positions of the coefficients having a value other than the most likely-to-occur amplitude using at least one position coding method selected from the set of position coding methods consisting of:
• a first position coding method including repeatedly alternating encoding of (a) a run-length of a run of most likely-to-occur amplitude coefficients preceding a coefficient having other than the most likely-to-occur amplitude, and (b) a run length of a run of coefficients having amplitudes other than the most likely-to-occur amplitude and determining if alternating produces shorter codewords than no alternating, the alternating occurring according to whether alternating or not alternating is determined to produce shorter codewords;
• and
• a second position coding method including determining a first parameter and a second parameter that are indicative of the number of most likely-to-occur amplitude coefficients and the number of coefficients having other than the most likely-to-occur amplitude whose positions remains to be coded;
• and jointly encoding the positions of the coefficients according to the first and second parameters using a variable length coding mapping that provides for any pair of parameter values all possible position patterns that have those parameter values,
• such that relatively more likely-to-occur sequences of quantized coefficients are represented by relatively short sets of codewords and relatively less likely-to-occur sequences of quantized coefficients are represented by relatively long sets of codewords,
• wherein the encoding the positions uses one or more code mappings determined using assumed or measured statistics for likelihood of occurrence of events.

An apparatus comprising:

• an input buffer operative to accept a bitstream of codewords by a context-based adaptive variable length coding method that encodes an ordered sequence of quantized transform coefficients of a block of an image, the quantized coefficients, each quantized coefficient having an amplitude from a set of amplitudes that includes a most likely-to-occur amplitude and at least one other amplitude, including a next-to-most likely-to-occur amplitude;
• a codeword recognizer coupled to the input buffer and operative to recognize a codeword of the codewords in the bitstream, the codeword recognizer comprising one or more processors and a non-transitory computer readable medium comprising instructions to instruct at least one of the one or more processors to recognize the codeword;
• a decoder operative to decode the recognized codeword,
• wherein the context-based adaptive variable length coding method includes encoding the positions of the coefficients having a value other than the most likely-to-occur amplitude using at least one position coding method selected from the set of position coding methods consisting of:
• a first position coding method including repeatedly alternating encoding of (a) a run-length of a run of most likely-to-occur amplitude coefficients preceding a coefficient having other than the most likely-to-occur amplitude, and (b) a run length of a run of coefficients having amplitudes other than the most likely-to-occur amplitude and determining if alternating produces shorter codewords than no alternating, the alternating occurring according to whether alternating or not alternating is determined to produce shorter codewords;
• and
• a second position coding method including determining a first parameter and a second parameter that are indicative of the number of most likely-to-occur amplitude coefficients and the number of coefficients having other than the most likely-to-occur amplitude whose positions remain to be coded;
• and jointly encoding the positions of the coefficients according to the first and second parameters using a variable length coding mapping that provides for any pair of parameter values all possible position patterns that have those parameter values,
• such that relatively more likely-to-occur sequences of quantized coefficients are represented by relatively short sets of codewords and relatively less likely-to-occur sequences of quantized coefficients are represented by relatively long sets of codewords.

A non-transitory computer-readable storage medium storing instructions that when executed by one or more processors cause carrying out a method comprising:

• accepting a bitstream, wherein the bitstream comprises a concatenation of codewords determined by a coding method that encodes an ordered sequence of quantized transform coefficients of a block of image data of an image into codewords;
• recognizing of codewords;
• and
• decoding recognized codewords,
• wherein the coding method comprises:
• accepting the ordered sequence, wherein the quantized coefficients in the sequence have a most likely-to-occur amplitude, and at least one other amplitude including a next-to-most likely-to-occur amplitude;
• encoding the ordered sequence using a context-based adaptive variable length coding method, the context-based adaptive variable length coding method including encoding the positions of the coefficients having a value other than the most likely-to-occur amplitude using at least one position coding method selected from the set of position coding methods consisting of:
• a first position coding method including repeatedly alternating encoding of (a) a run-length of a run of most likely-to-occur amplitude coefficients preceding a coefficient having other than the most likely-to-occur amplitude, and (b) a run length of a run of coefficients having amplitudes other than the most likely-to-occur amplitude and determining if alternating produces shorter codewords than no alternating, the alternating occurring according to whether alternating or not alternating is determined to produce shorter codewords;
• and
• a second position coding method including determining a first parameter and a second parameter that are indicative of the number of most likely-to-occur amplitude coefficients and the number of coefficients having other than the most likely-to-occur amplitude whose positions remain to be coded;
• and jointly encoding the positions of the coefficients according to the first and second parameters using a variable length coding mapping that provides for any pair of parameter values all possible position patterns that have those parameter values,
• such that relatively more likely-to-occur sequences of quantized coefficients are represented by relatively short sets of codewords and relatively less likely-to-occur sequences of quantized coefficients are represented by relatively long sets of codewords.