Research On Depth Map Coding In3D Video

As a kind of novel multimedia form, three dimensional video (3DV) can provide viewers a immersive experience, and enable the viewers to freely choose an arbitrary view-point within a viewing angle.Multi-view video plus depth (MVD) representation is one of the main data formats of3DV, which needs to transmit2D texture video and the corresponding depth map, while it brings huge data. Under the limitations of transmitting and storage, the depth map sequence should be compressed efficiently. The paper focuses on the research of the depth map compression algorithm. Major the main contents and innovations of this paper are summarized as follows.There exists correlation between the motion information of depth map and texture video. And the rich detail in the texture video enables the texture video to provide more precise motion vectors. Based on the motion information of the texture video, this paper proposes a new depth map coding algorithm. Instead of searching motion vectors in the depth map, the depth map utilizes the motion information generated by the corresponding texture video directly. It is proved that the algorithm brings a good compression gains.In some cases the inter-view prediction algorithm does not work appropriately. This paper analyses the reasons and proposes a selecting algorithm of the inter-view prediction for the depth map. The algorithm decides whether using inter-view prediction or not for every frame of the depth map according to the rate-distortion optimization criterion. Experiment results show that this algorithm improves the coding efficiency of the depth map.