A fast mode determination algorithm for multi-view video coding

Abstract

Multi-view video coding by using the hierarchical B picture structure utilizes intra-view and inter-view predictions to reduce redundant information. The best coding mode is determined by exhaustively searching over all possible partition modes using rate distortion comparison in motion estimation for each macroblock. However, there is a high computational complexity in exhaustive searching. This study analyzes the statistics of coding mode distribution in the inter- and intra-view and then a fast mode decision algorithm is proposed to select a proper coding mode, in which the probability density function of RD cost and motion homogeneity degree are set as the multi-threshold in the algorithm to achieve the mode determination. For multi-view, the mode correlation of the neighboring views is utilized to select the coding mode from interview or the intra-view prediction. Experimental results show that the encoding time for the basic view and multi-view is saved up to 75% and 65%, respectively, and the quality of the reconstruction video is almost remained.