Region-of-intrest based rate and computing resource allocation in video coding

摘要

In this paper, we propose a ROI (Region of Interest)-based rate and computing resource allocation algorithm for H.264/AVC video coding, targeting at the wireless video communication applications on mobile devices. The proposed algorithm includes three modules: ROI detection, computing resource allocation, and bit rate allocation. A saliency-map-based scheme is performed to detect the ROI. Considering the ROI is more attention-attractive, we propose the novel methods to allocate more computational resource and bits to the ROI. The experimental results demonstrate that the proposed scheme can achieve better subjective quality at the same computing constraint compared with the state-of-the-art algorithm.