Region merging and coding parameter reuse via merging

A method of decoding a data stream comprising an array of information samples representing a picture at a resolution level, the method comprising:

• extracting, from the data stream, a merge indicator with respect to a region currently being decoded within the array of information samples, wherein the region currently being decoded is one of a plurality of regions obtained by dividing the array of information samples and the merge indicator is indicative of whether the region currently being decoded is to be reconstructed based on at least one coding parameters associated with a selected candidate region of the resolution level;
• identifying a set of candidate regions of the resolution level with respect to the region currently being decoded;
• if a number of candidate regions in the set is greater than one, extracting a candidate identifier from the data stream, the candidate identifier identifying a specific one of the set of candidate regions;
• selecting the specific one of the set of candidate regions as the selected candidate region in accordance with the candidate identifier;
• entropy decoding the candidate identifier using context modeling based on a single probability model;
• determining whether the region currently being decoded is to be reconstructed by copying or computing the at least one coding parameter based on one or more coding parameters associated with the selected candidate region;
• copying, if a determination is to copy, the at least one coding parameter from the one or more coding parameters;
• computing, if the determination is to compute, the at least one coding parameter based on a prediction of the at least one coding parameter using the one or more coding parameters and a prediction residual of the at least one coding parameter;
• and
• reconstructing the region currently being decoded based on the at least one coding parameter.

A non-transitory machine readable digital storage medium having computer instructions stored thereon for decoding a data stream comprising an array of information samples representing a picture at a resolution level, wherein the computer instructions, when read by the machine, causes the machine to at least perform:

• extracting, from the data stream, a merge indicator with respect to a region currently being decoded within the array of information samples, wherein the region currently being decoded is one of a plurality of regions obtained by dividing the array of information samples and the merge indicator is indicative of whether the region currently being decoded is to be reconstructed based on at least one coding parameter associated with a selected candidate region of the resolution level;
• identifying a set of candidate regions of the resolution level with respect to the region currently being decoded;
• if a number of candidate regions in the set is greater than one, extracting a candidate identifier from the data stream, the candidate identifier identifying a specific one of the set of candidate regions;
• selecting the specific one of the set of candidate regions as the selected candidate region in accordance with the candidate identifier;
• entropy decoding the candidate identifier using context modeling based on a single probability model;
• determining whether the region currently being decoded is to be reconstructed by copying or computing the at least one coding parameter based on one or more coding parameters associated with the selected candidate region;
• copying, if a determination is to copy, the at least one coding parameter from the one or more coding parameters;
• computing, if the determination is to compute, the at least one coding parameter based on a prediction of the at least one coding parameter using the one or more coding parameters and a prediction residual of the at least one coding parameter;
• and
• reconstructing the region currently being decoded based on the at least one coding parameter.

An encoder for encoding an array of information samples representing a picture at a resolution level into a data stream, the encoder comprising:

• a sub-divider configured for dividing the array of information samples into a plurality of regions;
• and
• a data stream generator configured for
• encoding, into the data stream, a merge indicator with respect to one of the plurality of regions that is currently being coded, wherein the merge indicator is indicative of whether the region currently being coded is to be reconstructed based on at least one coding parameter associated with a selected candidate region selected from a set of candidate regions of the resolution level, identified with respect to the region currently being coded,
• encoding, into the data stream, a candidate identifier that identifies a specific one of the set of candidate regions as the selected candidate region, if a number of candidate regions in the set is greater than one, the encoding of the candidate identifier including entropy encoding the candidate identifier using context modeling based on a single probability model,
• determining whether the region currently being coded is to be reconstructed by copying or computing the at least one coding parameter based on one or more coding parameters associated with the selected candidate region,
• copying, if a determination is to copy, the at least one coding parameter from the one or more coding parameters,
• computing, if the determination is to compute, the at least one coding parameter based on a prediction of the at least one coding parameter using the one or more coding parameters and a prediction residual of the at least one coding parameter, and
• encoding, into the data stream, the one or more coding parameters associated with the selected candidate region.

A method of encoding an array of information samples representing a picture at a resolution level into a data stream, comprising:

• dividing the array of information samples into a plurality of regions;
• encoding, into the data stream, a merge indicator with respect to one of the plurality of regions currently being coded, wherein the merge indicator is indicative of whether the region currently being coded is to be reconstructed based on at least one coding parameter associated with a selected candidate region selected from a set of candidate regions of the resolution level, identified with respect to the region currently being coded;
• encoding, into the data stream, a candidate identifier that identifies a specific one of the set of candidate regions as the selected candidate region, if a number of candidate regions in the set is greater than one, the encoding of the candidate identifier including entropy encoding the candidate identifier using context modeling based on a single probability model;
• determining whether the region currently being coded is to be reconstructed by copying or computing the at least one coding parameter based on one or more coding parameters associated with the selected candidate region;
• copying, if a determination is to copy, the at least one coding parameter from the one or more coding parameters;
• computing, if the determination is to compute, the at least one coding parameter based on a prediction of the at least one coding parameter using the one or more coding parameters and a prediction residual of the at least one coding parameter;
• and
• encoding, into the data stream, the one or more coding parameters associated with the selected candidate region.

A non-transitory machine readable digital storage medium having computer instructions stored thereon for encoding an array of information samples representing a picture at a resolution level into a data stream, wherein the computer instructions, when read by the machine, causing the machine to at least perform:

• dividing the array of information samples into a plurality of regions;
• encoding, into the data stream, a merge indicator with respect to one of the plurality of regions currently being coded, wherein the merge indicator is indicative of whether the region currently being coded is to be reconstructed based on at least one coding parameter associated with a selected candidate region selected from a set of candidate regions of the resolution level, identified with respect to the region currently being coded;
• encoding, into the data stream, a candidate identifier that identifies a specific one of the set of candidate regions as the selected candidate region, if a number of candidate regions in the set is greater than one, the encoding of the candidate identifier including entropy encoding the candidate identifier using context modeling based on a single probability model;
• determining whether the region currently being coded is to be reconstructed by copying or computing the at least one coding parameter based on one or more coding parameters associated with the selected candidate region;
• copying, if a determination is to copy, the at least one coding parameter from the one or more coding parameters;
• computing, if the determination is to compute, the at least one coding parameter based on a prediction of the at least one coding parameter using the one or more coding parameters and a prediction residual of the at least one coding parameter;
• and
• encoding, into the data stream, the one or more coding parameters associated with the selected candidate region.

A data stream stored in a non-transitory storage medium, the data stream comprising encoded information including an array of information samples representing a picture at a resolution level which is divided into a plurality of regions, the data stream further comprising,

• a merge indicator with respect to one of the plurality of regions, wherein the merge indicator is indicative of whether the region is to be reconstructed based on at least one coding parameter associated with a selected candidate region selected from a set of candidate regions of the resolution level, identified with respect to the region;
• a candidate identifier that identifies a specific one of the set of candidate regions as the selected candidate region, if a number of candidate regions in the set is greater than one, wherein the candidate identifier is entropy encoded into the data stream using context modeling based on a single probability model;
• and
• one or more coding parameters associated with the selected candidate region, from which the at least one coding parameter is copied or computed, and used to reconstruct the region, wherein computing the at least one coding parameter is based on a prediction of the at least one coding parameter using the one or more coding parameters and a prediction residual of the at least one coding parameter.