A quad-tree and statistics based fast CU depth decision algorithm for 3D-HEVC

Guangsheng Chi, Xin Jin, Qionghai Dai

Abstract: 3D-HEVC is the latest 3D video coding project of MPEG based on High Efficiency Video Coding (HEVC). The encoding complexity of it is remarkably high due to the coding units (CU) size decision process together with the rate distortion optimization (RDO) process. In this paper, we proposed a fast CU depth decision algorithm taking advantages of the interview correlations. With the CU depth correlation analysis between the independent view and the dependent view, the CU depth of each CU in the dependent view is automatically determined based on inter-view prediction and quad-tree structure constraints. It eliminates the brute force RDO process in determining CU depth, which provides an average of 60% saving in texture encoding complexity with a negligible loss in compression efficiency.