Macroblock-Based Frequency Selective Weighting for Visual Scalable Video Coding of H.264/AVC

In this paper, a novel macroblock-based frequency selective weighting scalable video coding (MBFSW-SVC) algorithm is investigated by exploring the characteristics of the human visual system (HVS) in SVC H.264/AVC for improving the coding efficiency. Compared with previous SVC schemes, the MBFSW-SVC can extend the quality layer with different frequency weighting for each MB according to the size of region of interests. For reasonable adaptation of the frequency weighting into the quality layer extension, the spatial context characteristics and the motion activities in each picture are taken into account. The experiment results show that the proposed scheme can improve the coding performance up to a 10-20% reduction of the bit rate with almost the same subjective video quality. In particular, the subjective quality is significantly improved in lower scalable quality layers with a small number of the encoding bit budget.