Prediction based fractional pixel motion estimation algorithm for AVS-M

Abstract

AVS-M is the latest video coding standard developed by China for mobile application, which promises improved performance close to H.264/AVC with lower complexity. In an AVS-M encoder, the most time-consuming component is variable block-size motion estimation (ME). Since the complexity of integer pixel search has been greatly reduced by numerous fast ME algorithms, the computation overhead required by fractional pixel ME has become relatively significant. To reduce the complexity of fractional pixel ME, a prediction based fractional pixel ME algorithm is proposed in this paper. More accurate motion vector predictions are utilized in the algorithm to achieve better computation reduction. Additionally, an early termination method to decrease the amount of search is proposed. Experimental results show that compared to the full search sub-pel ME and the fast sub-pel ME proposed in H.264/AVC, the proposed method can reduce up to 84% and 63% of fractional pixel search points respectively, with a negligible degradation in quality.