New motion estimation and compensation algorithms for video compression combining global and local motions

Abstract

In this paper, we propose algorithms for estimating and compensating the motion in the temporal images including global motion. Those images have two kinds of motions, i.e., global and local motions, and conventionally the motions are separately handled in two stages. But when the two classes of motions have a fractional resolution, the separation brings about two successive bilinear interpolations in predicting the current frame, resulting in bandwidth reduction in the reconstructed images. To solve this problem, we propose to combine two motion compensations for the global and local motions into just one motion compensation. We further propose a combined estimation algorithm which is suitable for the proposed combined motion compensation. Since the combined estimation algorithm has large computational complexity, the fast combined estimation algorithm is proposed which has the same computational complexity as the conventional estimation algorithm. The simulation results show that the proposed algorithm gives better prediction results than those of the conventional separate algorithm.