Spatiogram based fast mode decision in spatial scalable video coding

The scalable video coding extension of H.264/AVC standard specifies the usage of layered scalable bit streams to achieve temporal, spatial and quality scalability. The inter-layer redundancies are exploited in addition to spatial and temporal predictions to obtain a better coding efficiency. The addition of inter-layer modes increases the computational complexity of the mode decision process in the enhancement layer. We propose a fast mode decision algorithm based on spatiograms for reducing the computational overhead during the encoding process. Our algorithm is based on constructing spatiograms for the macroblocks and comparing them to find the similarity of the macroblocks. This similarity measure is used to evaluate only a subset of modes leading to reduction in coding time. Our experiment results show that the time saving is around 29% to 67% based on the video sequence with a quality loss between 0 to -0.04dB and bit rate increase in the range -0.21% to 0.94%. The quality loss and the bit rate increase from our proposed algorithm is extremely insignificant when compared to the original JSVM 9.8 reference implementation. © 2010 Springer-Verlag.