Context adaptation within video coding modules

摘要

Context adaptive video coding is described in which encoding and decoding relies on deriving context adaptive coding tools, which can be applied within a coder (encoder and/or decoder) to one or more coding processes. A discontinuity of the block is identified, and a context feature extracted having directionality and location of discontinuity. Coding then uses the context adaptive tools to enhance coding efficiency, without the need of additional side information. Context adaptation according to invention can be applied such as to inter prediction, intra prediction, transform, interpolation filtering, and/or scanning.

主权项

1. An apparatus for performing video coding, comprising:
(a) a computer configured for encoding and/or decoding of video; and (b) programming executable on said computer for context adaptation by performing steps comprising:
(i) assuming a block within a video frame is piece-wise smooth;(ii) identifying that a discontinuity exists in the block with respect to its context of neighbors;(iii) extracting a context feature of said discontinuity having directionality and location of discontinuity within the block, with location of discontinuity extracted in response to steps, including:
(iii)(1) mapping a line along a direction of discontinuity;(iii)(2) determining distance from said line to the center of a current block;(iii)(3) determining an average of said distance for said block among neighboring pixels that have significant gradients;(iii)(4) utilizing said average as the discontinuity location; and(iv) performing coding of said block as a context adaptive operation in response to said context feature.