Virtual view synthesis oriented fast depth video encoding algorithm

A virtual view rendering oriented fast algorithm is proposed to reduce computational complexity of multiview depth video coding. Firstly, the depth image is divided into depth-continuity and depth-discontinuity regions by a macroblock deviation factor. Then, imbalance macroblock mode distribution of depth-continuity regions and effect of image quality degradation in depth-discontinuity regions on virtual view rendering are discussed and analyzed in detail. Finally, the fast algorithm is presented. Two methods are utilized in different regions of depth video. The first method, directly searching SKIP and Intra modes is adopted in smooth regions. The second method which is the same as the JMVM reference software is utilized in depth discontinuous regions. The experimental results show that the proposed algorithm significantly promotes the encoding speed of depth video by 3.91~11.33 times, while maintains high virtual rendering performance in comparison with full search algorithm.