Spatial transform skip in the emerging High Efficiency Video Coding standard

To meet the still growing compression efficiency needs for high definition content, ITU and MPEG are now defining the High Efficiency Video Coding (HEVC) standard. The HEVC codec still relies on a hybrid motion compensated predictive video coding architecture as its H.264/AVC ancestor though novel coding tools are introduced. These novel coding tools provide a highly uncorrelated prediction residue for which classical frequency decomposition methods as the discrete cosine transform may not provide an effective energy compaction. Therefore this paper proposes a transform skip mode which allows skipping one or both directions where the transform is applied. The proposed transform skip mode is integrated in the HEVC codec and is able to provide bitrate reductions of up to 5% at the same objective quality when compared with the HEVC reference codec.