A novel sub-pixel motion estimation algorithm using wavelet-based contourlet transform

High-precision matching and compensation can reduce the prediction error and improve the compression effect of video image. The motion compensation has been accurate to1/4 pixel. But the computation of sub-pixel full search (SPFS) algorithm is great, which makes the encoder´s coding rate very low. Aimed at this problem, a novel sub-pixel motion estimation algorithm using wavelet-based contourlet transform (WBCT) is proposed in this paper. Firstly, the optimal low-frequency sub-band integer pixel point of the image block in WBCT domain is found out. Then the optimal point is regarded as a center to search the optimal motion vector of the subpixel by chaotic particle swarm optimization (PSO) algorithm, which can be regarded as the initial motion vector of corresponding high-frequency sub-band in the image block. Finally, we refinedly search the high-frequency sub-band to get the optimal motion vector and complete motion compensation. As the experimental results show, the proposed algorithm has superior performance in aspects of search time and motion compensation effect.