HEVC encoder optimization for HDR video coding based on irregularity concealment effect

In this paper, we propose HEVC encoder optimization for HEVC Main 10 Profile-based HDR video coding using irregularity concealment effect. The human visual system (HVS) is more sensitive to the visual contents with regularity than those without it, i.e., the irregularity concealment effect. We adopt the irregularity concealment effect for HEVC encoder optimization. First, we calculate orientations of pixels, and then obtain visual regularity in a local region using the orientation distribution. Next, we combine the visual regularity with 10bit luminance adaptation and spatial masking to generate 10bit just-noticeable-distortion (JND) map, called visual regularity-based JND (VRJND). Finally, based on VRJND, we perform perceptual block merging for HEVC Main 10 Profile-based HDR video coding. Experimental results demonstrate that the proposed method achieves average bit-rate reductions of 4.2% in PSNR-DE100 and 5.24% in HDR-VDP-2.2 with only increase of 4.09% runtime over HEVC. Visual comparison further verifies that the proposed method preserves more structural and edge information than HEVC in reconstructing HDR videos.