A novel direction-based JND model for perceptual HEVC intra coding

Abstract

In High efficiency video coding (HEVC), rate distortion optimization (RDO) technique plays a crucial part in improving video coding efficiency. However, the perceptual characteristics of human visual system (HVS) are not taken into consideration in the process of RDO. In this paper, given the prediction patterns of intra frame, a novel direction-based just-noticeable-distortion (DRJND) model as a perceptual feature is introduced to HEVC encoder, which adaptively adjust Lagrange multiplier. Meanwhile, the rate-distortion (RD) cost also needs to be perceptually calculated once again. Experimental results demonstrate that compared with the anchor HM12.0, the proposed algorithm has achieved better experimental results in rate-SSIM performance while keeping its visual quality of the same test sequences.