Fast mode decision algorithm for scalable video coding based on luminance coded block pattern

Abstract

A fast mode decision algorithm is proposed to reduce the com- putation complexity of adaptive inter layer prediction method, which is a motion estimation algorithm for video compression in scalable video cod- ing (SVC) encoder systems. SVC is standard as an extension of H.264/ AVC to provide multimedia services within variable transport environments and across various terminal systems. SVC supports an adaptive inter mode prediction, which includes not only the temporal prediction modes with varying block sizes but also inter layer prediction modes based on correlation between the lower layer information and the current layer. To achieve high coding efficiency, a rate distortion optimization technique is employed to select the best coding mode and reference frame for each MB. As a result, the performance gains of SVC come with increased com- putational complexity. To overcome this problem, we propose fast mode decision based on coded block pattern (CBP) of 16 × 16 mode and refer- ence block of best CBP. The experimental results in SVC with combined scalability structure show that the proposed algorithm achieves up to an average 61.65% speed up factor in the encoding time with a negligible bit increment and a minimal image quality loss. In addition, experimental results in spatial and quality scalability show that the computational com- plexity has been reduced about 55.32% and 52.69%, respectively. © 2013