Faster mode determination algorithm using mode correlation for multi-view video coding

Multi-view video coding with a hierarchical B picture structure utilises intra-view and inter-view predictions to reduce the quantity of redundant information. The optimal coding mode is determined by exhaustively searching through all possible partition modes; however, a high degree of computational complexity is involved in such exhaustive searches. In this study, the authors statistically analyse the coding mode distribution in inter-view and intra-view and propose a fast mode decision algorithm to select the optimal mode in terms of rate-distortion optimisation. The probability density function of the rate-distortion cost and the degree of the homogeneity in motion are set as the multi-threshold in the algorithm to determine the optimal mode for base view coding. For the multi-view coding, the correlation of the modes in neighbouring views with similar regions is utilised to select the coding mode from the inter-view or intra-view predictions. The experimental results show that the encoding time for the base view and the multi-view is reduced by up to 85 and 69%, respectively, and the quality of the reconstructed video is nearly unchanged.