A New Method for Stereo Video Encoding Based on AVS

A new method for stereo video encoder based-on AVS is presented in this paper. In the system, an AVS encoder encodes the left-view sequence. For the right-view sequence, combined predictions from the temporal or disparity domains are exploited, with a hierarchical two-stage neural classifier for fast mode decision. The result of first-stage classifier produces different candidates for variable block partitions of each macro-block (MB). The second-stage classifier chooses the most suitable prediction sources between forward motion and disparity frames. The input of two-stage classifier is the simple calculation between current and its reference. Experiment results show that the proposed method can achieve up to 80% of time savings with nearly ignored quality degradation and only 7% of bit-rate increase.