Low delay picture coding

A method for decoding information from a datastream to reconstruct a picture, the method comprising:

• entropy decoding portions of the picture based on wavefront parallel processing (WPP) in which a current portion of the portions is entropy decoded according to one of at least two modes indicated by a syntax element portion associated with the current portion, wherein the entropy decoding includes:
• in accordance with a first mode of the at least two modes, decoding data related to the current portion from the datastream using context adaptive entropy decoding to obtain a residual signal, the context adaptive entropy decoding including deriving contexts across portion boundaries and initializing a symbol probability,
• wherein, in initializing the symbol probability in accordance with the first mode, the entropy decoding includes:
• checking as to whether a current coding block associated with the current portion is the left-most coding block in a current row of the picture associated with the current portion,
• if so, initializing the symbol probability based on a saved symbol probability as acquired in context adaptive entropy decoding a previously-decoded portion up to a previous coding block of the picture associated with the previously-decoded portion, and
• if not, initializing the symbol probability based on a symbol probability as acquired at the completion of context adaptive entropy decoding a last coding block of the picture associated with the previously-decoded portion, and
• in accordance with a second mode of the at least two modes, decoding data related to the current portion from the datastream using context adaptive entropy decoding to obtain a residual signal, the context adaptive entropy decoding including restricting the derivation of the contexts so as to not cross the portion boundaries and initializing symbol probabilities independent of a previously-decoded portion;
• generating a prediction signal based on prediction parameters related to the current portion from the datastream;
• and
• reconstructing the picture related to the current portion based on the residual signal and the prediction signal.

An encoder for encoding picture into a datastream, the encoder comprising:

• a residual module configured to generate, using a processor, a residual signal related to a current portion of portions of the picture;
• a predictor configured to generate, using the processor, prediction parameters based on a prediction signal, wherein the prediction parameters are encoded into the datastream;
• and
• an entropy encoder configured to determine, using the processor, a syntax element portion for the current portion and entropy encode, into the datastream, the portions based on wavefront parallel processing (WPP) in which the residual signal related to the current portion is entropy encoded according to one of at least two modes indicated by the syntax element portion, wherein the entropy encoder is configured to:
• if the current portion is to be coded in accordance with a first mode of the at least two modes, encode the residual signal using context adaptive entropy encoding that includes deriving contexts across portion boundaries and initializing a symbol probability associated with the current portion,
• wherein, in initializing the symbol probability associated with the current portion, the entropy encoder is configured to:
• check as to whether a current coding block associated with the current portion is the left-most coding block in a current row of the picture associated with the current portion,
• if so, initialize the symbol probability based on a saved symbol probability as acquired in context adaptive entropy encoding a previously-encoded portion up to a previous coding block of the picture associated with the previously-encoded portion, and
• if not, initialize the symbol probability based on a symbol probability as acquired at the completion of context adaptive entropy encoding a last coding block of the picture associated with the previously-encoded portion, and
• if the current portion is to be coded in accordance with a second mode of the at least two modes, encode the residual signal using context adaptive entropy encoding with restricting the derivation of the contexts so as to not cross the portion boundaries and initializing symbol probabilities independent on any previously-encoded portion.

A non-transitory computer-readable medium for storing data associated with a video, comprising:

• a data stream stored in the non-transitory computer-readable medium, the data stream comprising encoded information related to a picture of a video, wherein the picture is partitioned in portions, wherein the encoded information is coded into the data stream for storing in the non-transitory computer-readable medium by operations including:
• generating a residual signal related to a current portion of portions of the picture;
• generating prediction parameters based on a prediction signal, wherein the prediction parameters are encoded into the datastream;
• and
• determining a syntax element portion for the current portion and entropy encoding, into the datastream, the portions based on wavefront parallel processing (WPP) in which the residual signal related to the current portion is entropy encoded according to one of at least two modes indicated by the syntax element portion, wherein the entropy encoding includes:
• if the current portion is to be coded in accordance with a first mode of the at least two modes, encoding the residual signal using context adaptive entropy encoding that includes deriving contexts across portion boundaries and initializing a symbol probability associated with the current portion,
• wherein, in initializing the symbol probability associated with the current portion, the entropy encoding includes:
• checking as to whether a current coding block associated with the current portion is the left-most coding block in a current row of the picture associated with the current portion,
• if so, initializing the symbol probability based on a saved symbol probability as acquired in context adaptive entropy encoding a previously-encoded portion up to a previous coding block of the picture associated with the previously-encoded portion, and
• if not, initializing the symbol probability based on a symbol probability as acquired at the completion of context adaptive entropy encoding a last coding block of the picture associated with the previously-encoded portion, and
• if the current portion is to be coded in accordance with a second mode of the at least two modes, encoding the residual signal using context adaptive entropy encoding with restricting the derivation of the contexts so as to not cross the portion boundaries and initializing symbol probabilities independent on any previously-encoded portion.