Video processing for masking coding artifacts using dynamic noise maps

A video decoder system, comprising:

• a video decoding engine to generate recovered video from a data stream of coded video data,
• a noise database storing a plurality of noise patches, and
• a noise mapping system to retrieve noise patches from the database based on an indicator from an encoder present in the data stream, the indicator identifying a patch from the plurality of noise patches stored in the noise database, and a type of scaling to be applied,
• a post-processing unit to merge the retrieve noise patches and integrate them into the recovered video according to the identified scaling.

A video decoding method, comprising:

• decoding coded video data to generate recovered video data,
• estimating a location of coding artifacts present in the recovered video,
• selecting previously-stored noise patches from memory to mask the coding artifacts according to a noise mask identifier received in the data stream from an encoder, wherein the noise mask identifier identifies at least one noise patch from a plurality of noise patches stored in the database, and
• merging the selected noise patches into the recovered video.

A video decoding method comprising:

• decoding coded video data to generate recovered video data,
• based on a patch identifier provided from an encoder in a received data stream, selecting previously-stored noise patches from memory, wherein the patch identifier identifies at least one noise patch from a plurality of noise patches stored in memory,
• scaling the retrieved noise patches according to a scaling identifier provided in the received data stream, and
• merging the selected noise patches into the recovered video.

A video encoding system, comprising:

• a video coding engine to code source video data into a coded video data,
• a video decoder to decode the coded video data into recovered video data,
• an artifact estimator to identify locations of artifacts in the recovered video data,
• a noise database storing noise patches, and
• a patch selector to select stored noise patches that mask the artifacts when integrated with the recovered video data during a post-processing operation, the patches selected according to a noise mask identifier received in the data stream from an encoder, wherein the noise mask identifier identifies at least one noise patch from a plurality of noise patches stored in the database.

A method, performed at a video encoder, comprising:

• coding source video data into a coded video data,
• decoding the coded video data into recovered video data,
• emulating a noise patch derivation to be performed at a decoder, the emulating identifying a first noise patch of a local noise database that would be used by the decoder,
• comparing recovered video data generated according to the emulation with a source video data,
• determining whether the noise database includes noise patches other than the first noise patch that better mask the artifacts,
• when other noise patches are identified by the determining, transmitting an identifier of the other noise patches in a channel with the coded video data in which the artifacts are located, wherein the identifiers each identify at least one noise patch from a plurality of noise patches stored in the database.