In-loop selective processing for contour artefact reduction in video coding
Proposed is an in-loop decontouring algorithm that utilises block-based processing for video coding. In order to avoid uniform filtering that brings about unnecessary processing, candidate blocks that may contain noticeable false contours using different kinds of features in a block are sequentially refined. Then, pseudorandom noise masking is applied to them to reduce contour artefacts. It is demonstrated that visual quality could be efficiently improved with a negligible rate-distortion (RD) loss.