Fast 3D-HEVC intra-prediction for depth map based on a self-organizing map and efficient features

The ISO/IEC MPEG and ITU-T Video Coding Experts Group formed the Joint Collaborative Team on 3D Video Coding to develop the new 3D video coding standard 3D High-Efficiency Video Coding (3D-HEVC) which is based on MultiView plus Depth map data format which associates a depth map to each texture frame. The depth map intra-prediction provides the conventional HEVC intra modes and adds the depth modeling modes as new coding tools for depth map intra-predictions, which enables high coding efficiency, but also results in a very high encoding complexity and excludes the 3D-HEVC from the real-time applications. In this paper, we proposed a fast mode decision algorithm for depth map intra-coding based on efficient features. We used constructed training data and a self-organizing algorithm in the test model to skip unnecessary depth map intra modes for each prediction unit. The experimental results show that the proposed model brings a good complexity reduction with negligible loss of rate-distortion performance.