Fast mode decision in HEVC for depth image coding

MVD (Multi-view video plus depth) can synthesize virtual viewpoint effectively and reduce the number of desired views in transmission. Among previous works for depth image coding, HEVC has outperformed many other algorithm. However, the mode decision in HEVC is really time-consuming for its high computational complexity. In order to improve the encoding efficiency of depth image in MVD, a fast mode decision algorithm is proposed. By utilizing the depth information, the proposed algorithm gains residual characteristics of depth image by statistics and sets the corresponding threshold to reduce alternative modes which cause early termination of CU´ division. Thus, the encoding time is greatly decreased. The experimental results show that the encoding time is sharply decreased with negligible video quality degradation.