Adaptive Group-of-Picture Size Control Based on Side Information Quality Analysis for Distributed Video Coding

Abstract

The group of picture (GOP) size control in distributed video coding (DVC) is a significant task since it has a direct influence on DVC compression performance. In this paper, a novel adaptive GOP size control algorithm based on the evaluation of side information (SI) quality is presented and integrated into the state-of-the-art DVC architecture in discrete wavelet transformation domain. In order to achieve adaptive GOP size control and without raising the complexity of encoder obviously, the quality of the coarse SI for a whole frame is assessed online firstly. Then the index positions of key frames in a video sequence are adaptively determined according to the statistics of the SI quality. The experimental results show that for tested sequences with different motion scenarios, the compression performance of the targeted DVC system with the proposed algorithm is improved especially at the high bit-rates, compared with the fixed GOP size.