A fast depth map coding method based on optimized modes decision

摘要

Multiview video plus depth (MVD) has been widely used recently as the representation of 3D scenes. Due to the tremendous data volume, transmission of MVD sequences becomes difficult especially under networks circumstance. Hence, an efficient 3D video coding scheme is required to compress the video data. In this paper, a fast depth map coding method based on optimized inter-frame modes decision in H.264/AVC is proposed to improve the coding efficiency of depth maps. In our study, depth maps are divided into edge, foreground and background parts. For each part, the distribution of inter-frame modes is measured among sequences and only the most frequently used modes are selected as the candidate modes. Based on the statistic results, different modes decision strategies are applied to those three parts during coding process. Experiments show that the proposed method saves up to 76% coding time with negligible PSNR loss and bit rate overhead.