Integrating a spatial just noticeable distortion model in the under development HEVC codec

Although great achievements have been obtained in the past, recent developments towards ultra high definition video have been asking for further compression efficiency gains regarding the H.264/AVC state-of-the-art. As an answer to these needs, ITU-T and MPEG started a new standardization project called High Efficiency Video Coding. This paper extends and integrates a perceptual visual model in the High Efficiency Video Coding experimental codec to improve its RD performance. The designed extensions are related to the new Integer Discrete Cosine Transform block sizes as well as the new Mode Dependent Directional Transform. Moreover, an estimation mechanism for the perceptual model at the decoder side is also integrated to avoid the rate burden of sending to the decoder the model parameters. The RD performance is assessed using a methodology taking into account the variability of subjective scores which is reflected into a nonlinear sensitivity of the subjective quality metric. The RD performance obtained for the perceptually driven codec shows a bitrate reduction of up to 43% regarding the non-perceptual video codec.