Adaptive Down-sampling Stereo Video Coding Based on Background Complexity

摘要

In the existing stereo video coding scheme, the efficiency of asymmetric encoding is higher than that of symmetric coding. However, for a complex background, asymmetric coding has problems such as the blocking artifacts and losing high frequency components and so on, which would degrade the image quality. Therefore, this paper proposed a method of the adaptive down-sampling encoding based on the background complexity. This method is an asymmetric coding scheme, in which, the left viewpoint uses coding standard H.264, while the right viewpoint uses the joint estimation prediction mode of disparity and motion. It proposes and designs sample size decision model which makes the right viewpoint select the down-sample size adaptively according to the complexity of the macro block background. Simulation results show that: the coding efficiency of this algorithm increased by 30% compared with JMVM2.3, and the image quality has improved significantly compared with the down-sampling algorithm with a fixed scale, which has improved 0.3dB and 2.4dB respectively.