Context adaptive hybrid variable length coding

A computer-readable medium encoded with computer-executable instructions that when executed by at least one processor of a processing system causes carrying out of a coding method comprising:

• for an ordered series of quantized transform coefficients of a block of image data, the ordering from low to high spatial frequency according to the transform, the series having a most likely-to-occur amplitude, and at least one other amplitude including a next-to-most likely-to-occur amplitude, for a first region which could be the whole series:
• encoding the position of events in the first region that include one quantized coefficient having other than the most likely-to-occur amplitude or a cluster of at least one quantized coefficient having other than the most likely-to-occur amplitude, the encoding using a plurality of predetermined variable length code (VLC) mappings, including initially selecting an initial position VLC mapping, and later switching between position VLC mappings according to one or more position VLC selection criteria, including at least one context-based position VLC selection criterion;
• and
• if the position coding of the events leave any amplitude values other than the most likely-to-occur amplitude to be encoded, encoding the amplitudes of the to-be-encoded amplitude values in the reverse order of the original ordering of the series or cluster if events includes a cluster, the encoding according to one of a plurality of pre-selected amplitude VLC mappings, including initially selecting an amplitude VLC mapping, and switching between the plurality of amplitude VLC mappings according to one or more amplitude VLC selection criteria, including at least one context-based amplitude VLC selection criterion,
• wherein by a context-based selection criterion is meant a criterion that during encoding is known or derivable from one or more previously encoded items of information,
• wherein the position coding of events includes jointly encoding the next joint position and amplitude event that is defined by the next cluster of one or more consecutive coefficients each having any other than the most likely-to-occur amplitude, each event defined by three parameters:
• the number, possibly zero, of coefficients having the most likely-to-occur amplitude preceding the cluster, the number of coefficients in the cluster having other than the most likely-to-occur amplitude, and a third parameter related to amplitudes in the cluster having other than the most likely-to-occur amplitude, the coding for the first cluster being position coded being according to a three-dimensional joint VLC mapping selected according to one or more initial position VLC selection criteria, including at least one context-based initial position VLC selection criterion, and after the first encountered is position coded, according to one or more position VLC selection criteria, including at least one context-based position VLC selection criterion, and
• wherein the third parameter indicates for all or at least some cluster lengths the number of consecutive trailing coefficients in the cluster that each have the next-to-most likely-to-occur amplitude, and/or wherein in the case that such third parameter is not applicable to all cluster lengths, for at least one cluster length, the third parameter includes an indication of which of the coefficients in the cluster have the next-to-most likely-to-occur amplitude, and/or wherein in the case that such third parameter is not applicable to all cluster lengths, for a cluster length of one, the third parameter is indicative of the amplitude of the coefficient in the cluster having amplitude other than the most likely-to-occur amplitude.

An apparatus configured to encode an ordered series of quantized transform coefficients of a block of image data, the ordering from low to high spatial frequency according to the transform, the series having a most likely-to-occur amplitude, and at least one other amplitude including a next-to-most likely-to-occur amplitude, the apparatus comprising:

• a first region position encoder configured to encode the position of events in a first region of the series, the first region possibly the whole region, each event including one quantized coefficient having other than the most likely-to-occur amplitude or a cluster of at least one quantized coefficient having other than the most likely-to-occur amplitude, the encoding of the position by the first region position encoder using a plurality of predetermined variable length code (VLC) mappings, the first region position encoder configured to initially select an initial position VLC mapping, and to later switching between position VLC mappings according to one or more position VLC selection criteria, including at least one context-based position VLC selection criterion;
• and
• a first region amplitude encoder configured to encode, in the case the first region position encoder leaves any amplitude values other than the most likely-to-occur amplitude to be encoded, the amplitudes of the to-be-encoded amplitude values in the reverse order of the original ordering of the series or cluster if events includes a cluster, the encoding by the first region amplitude encoder being according to one of a plurality of pre-selected amplitude VLC mappings, the first region amplitude encoder configured to initially selecting an amplitude VLC mapping, and to later switch between the plurality of amplitude VLC mappings according to one or more amplitude VLC selection criteria, including at least one context-based amplitude VLC selection criterion,
• wherein by a context-based selection criterion is meant a criterion that during encoding is known or derivable from one or more previously encoded items of information,
• wherein the position coding of events includes jointly encoding the next joint position and amplitude event that is defined by the next cluster of one or more consecutive coefficients each having any other than the most likely-to-occur amplitude, each event defined by three parameters:
• the number, possibly zero, of coefficients having the most likely-to-occur amplitude preceding the cluster, the number of coefficients in the cluster having other than the most likely-to-occur amplitude, and a third parameter related to amplitudes in the cluster having other than the most likely-to-occur amplitude, the coding for the first cluster being position coded being according to a three-dimensional joint VLC mapping selected according to one or more initial position VLC selection criteria, including at least one context-based initial position VLC selection criterion, and after the first cluster is position coded, according to one or more position VLC selection criteria, including at least one context-based position VLC selection criterion, and
• wherein the third parameter indicates for all or at least some cluster lengths the number of consecutive trailing coefficients in the cluster that each have the next-to-most likely-to-occur amplitude, and/or wherein in the case that such third parameter is not applicable to all cluster lengths, for at least one cluster length, the third parameter includes an indication of which of the coefficients in the cluster have the next-to-most likely-to-occur amplitude, and/or wherein in the case that such third parameter is not applicable to all cluster lengths, for a cluster length of one, the third parameter is indicative of the amplitude of the coefficient in the cluster having amplitude other than the most likely-to-occur amplitude.

A method of decoding using a decoding apparatus, the method comprising:

• recognizing codewords in a set of concatenated codewords, the codewords formed by a context adaptive coding method that includes forming codewords based one or more context based criteria that during encoding are known or derivable from one or more previously encoded items of information, the context adaptive coding method comprising:
• for an ordered series of quantized transform coefficients of a block of image data, the ordering from low to high spatial frequency according to the transform, the series having a most likely-to-occur amplitude, and at least one other amplitude including a next-to-most likely-to-occur amplitude, for a first region which could be the whole series:
• encoding the position of events in the first region that include one quantized coefficient having other than the most likely-to-occur amplitude or a cluster of at least one quantized coefficient having other than the most likely-to-occur amplitude, the encoding using a plurality of predetermined variable length code (VLC) mappings, including initially selecting an initial position VLC mapping, and later switching between position VLC mappings according to one or more position VLC selection criteria, including at least one context-based position VLC selection criterion;
• and
• if the position coding of the events leave any amplitude values other than the most likely-to-occur amplitude to be encoded, encoding the amplitudes of the to-be-encoded amplitude values in the reverse order of the original ordering of the series or cluster if events includes a cluster, the encoding according to one of a plurality of pre-selected amplitude VLC mappings, including initially selecting an amplitude VLC mapping, and switching between the plurality of amplitude VLC mappings according to one or more amplitude VLC selection criteria, including at least one context-based amplitude VLC selection criterion,
• wherein the position coding of events includes jointly encoding the next joint position and amplitude event that is defined by the next cluster of one or more consecutive coefficients each having any other than the most likely-to-occur amplitude, each event defined by three parameters:
• the number, possibly zero, of coefficients having the most likely-to-occur amplitude preceding the cluster, the number of coefficients in the cluster having other than the most likely-to-occur amplitude, and a third parameter related to amplitudes in the cluster having other than the most likely-to-occur amplitude, the coding for the first cluster being position coded being according to a three-dimensional joint VLC mapping selected according to one or more initial position VLC selection criteria, including at least one context-based initial position VLC selection criterion, and after the first cluster is position coded, according to one or more position VLC selection criteria, including at least one context-based position VLC selection criterion, and
• wherein the third parameter indicates for all or at least some cluster lengths the number of consecutive trailing coefficients in the cluster that each have the next-to-most likely-to-occur amplitude, and/or wherein in the case that such third parameter is not applicable to all cluster lengths, for at least one cluster length, the third parameter includes an indication of which of the coefficients in the cluster have the next-to-most likely-to-occur amplitude, and/or wherein in the case that such third parameter is not applicable to all cluster lengths, for a cluster length of one, the third parameter is indicative of the amplitude of the coefficient in the cluster having amplitude other than the most likely-to-occur amplitude;
• and
• decoding the codewords,
• wherein during decoding of a codeword, any context-based selection criterion is known or derivable from one or more previously decoded items of information.

An apparatus configured to decode a set of concatenated codewords formed by a context adaptive coding method, the apparatus comprising:

• a codeword recognizer configured to recognize codewords in the set of concatenated codewords that is formed by the context adaptive coding method, including forming codewords based one or more context based criteria that during encoding are known or derivable from one or more previously encoded items of information,
• the context adaptive coding method comprising:
• for an ordered series of quantized transform coefficients of a block of image data, the ordering from low to high spatial frequency according to the transform, the series having a most likely-to-occur amplitude, and at least one other amplitude including a next-to-most likely-to-occur amplitude, for a first region which could be the whole series:
• encoding the position of events in the first region that include one quantized coefficient having other than the most likely-to-occur amplitude or a cluster of at least one quantized coefficient having other than the most likely-to-occur amplitude, the encoding using a plurality of predetermined variable length code (VLC) mappings, including initially selecting an initial position VLC mapping, and later switching between position VLC mappings according to one or more position VLC selection criteria, including at least one context-based position VLC selection criterion;
• and
• if the position coding of the events leave any amplitude values other than the most likely-to-occur amplitude to be encoded, encoding the amplitudes of the to-be-encoded amplitude values in the reverse order of the original ordering of the series or cluster if events includes a cluster, the encoding according to one of a plurality of pre-selected amplitude VLC mappings, including initially selecting an amplitude VLC mapping, and switching between the plurality of amplitude VLC mappings according to one or more amplitude VLC selection criteria, including at least one context-based amplitude VLC selection criterion,
• wherein the position coding of events includes jointly encoding the next joint position and amplitude event that is defined by the next cluster of one or more consecutive coefficients each having any other than the most likely-to-occur amplitude, each event defined by three parameters:
• the number, possibly zero, of coefficients having the most likely-to-occur amplitude preceding the cluster, the number of coefficients in the cluster having other than the most likely-to-occur amplitude, and a third parameter related to amplitudes in the cluster having other than the most likely-to-occur amplitude, the coding for the first cluster being position coded being according to a three-dimensional joint VLC mapping selected according to one or more initial position VLC selection criteria, including at least one context-based initial position VLC selection criterion, and after the first cluster is position coded, according to one or more position VLC selection criteria, including at least one context-based position VLC selection criterion, and
• wherein the third parameter indicates for all or at least some cluster lengths the number of consecutive trailing coefficients in the cluster that each have the next-to-most likely-to-occur amplitude, and/or wherein in the case that such third parameter is not applicable to all cluster lengths, for at least one cluster length, the third parameter includes an indication of which of the coefficients in the cluster have the next-to-most likely-to-occur amplitude, and/or wherein in the case that such third parameter is not applicable to all cluster lengths, for a cluster length of one, the third parameter is indicative of the amplitude of the coefficient in the cluster having amplitude other than the most likely-to-occur amplitude;
• and
• a codeword decoder configured to decode the recognized codeword,
• wherein during decoding of a codeword by the codeword decoder, any context-based selection criterion is known or derivable from one or more previously decoded items of information.