An adaptive segment-based view synthesis optimization method for 3D-HEVC

Abstract

This paper presents an adaptive segment-based view synthesis optimization (VSO) for depth map coding in 3D-HEVC. The original method skips lines of pixels where there is no change in depth data, and exhaustively renders pixels from all the rest of lines to get distortions from synthesized views. It brings unnecessary coding operations and ignores smaller regions which can also be skipped in VSO. The proposed method is designed to skip proper segments of pixels for the non-skipped lines based on both of the texture smoothness and depth data unchangeableness. Experimental results show that the proposed method can achieve low coding complexity without significant performance loss for the synthesized views.