Multi-scale dithering for contouring artefacts removal in compressed UHD video sequences

Abstract

Compared to HD (high definition), UHD (ultra-high definition) video provides higher visual quality. However, artefacts in UHD content are more apparent due to increased spatial resolution. False contouring is one of the visual artefacts in compressed UHD video sequences which affect the visual experience. This paper proposes a dithering method to remove the visual contouring while keeping artefact-free regions unaffected. The proposed multi-scale dithering algorithm is able to remove contouring of different widths. By considering the gradient smoothness and block boundary smoothness with neighbouring blocks, the dithering approach is applied to adjust each pixel value to break down the contouring bands. The implementation is block-based and of low complexity. Our results show that the contouring artefacts are effectively removed in all test sequences without generating new artefacts.