Low Complexity Hybrid View Synthesis Optimization for 3D-HEVC

Abstract

In this paper, we propose a low complexity Hybrid View Synthesis Optimization (HVSO) scheme for 3D extension of high efficiency video coding (3D-HEVC). In particular, based on the analyses regarding the impact of depth map compression on the quality of synthesized view rendering, we locally classify the coding tree units (CTUs) of depth map into two categories, including synthesized view distortion change (SVDC) based and view synthesis distortion estimation (VSDE) based CTUs. This allows us to dynamically distinguish the rate-distortion optimization process of these CTUs and obtain the synthesized view distortion of the VSDE based ones in a high efficiency way. More specifically, a new interpolation-based distortion estimation algorithm dedicated for VSDE based CTUs is proposed such that the actual view rendering process can be successfully avoided. In this manner, a good trade-off between the rate-distortion performance and computational complexity in depth map compression is achieved. Experimental results show that with our proposed HVSO method, 75% of the computational complexity on average can be saved for the view synthesis optimization process with only 1.6% BD-rate loss compared with the state-of-the-art scheme in the 3D-HEVC platform.