Research Of Motion Estimation Technology In Video Compression

Motion estimation and motion compensated are always the key technology in the current range of video coding standards. On the one hand, the accuracy of motion estimation is the key to measure the distortion of the image. The higher accuracy, the lower distortion in the image. The quality of the video image can be fully protected; on the other hand, the efficiency of motion estimation determines the real-time in the video encoding or decoding, the higher efficiency in the motion estimation algorithms, the lower complexity in the video encoding or decoding, the smaller proportion of time, it saves the time for the transmission of video image. Therefore, the research of the fast block-matching in the motion estimation has great significance, which improves the speed of video encoding and quality of prediction. The paper has researched the fast motion estimation algorithms which are easy to fall into the defect of local optimal, and completed the following work.First of all, the paper researches the key performance indicators which influence the block-matching motion estimation algorithm, it analyzes the influency that the indicators to the speed of the searching and accuracy. Secondly, it researches and analyzes the classics of the block-matching motion estimation algorithm, and analyzes the advantages and disadvantages of the motion estimation algorithm form the experimental data. Finally, on the base of researching and analyzing the block-matching motion estimation in image compression, the paper proposes a block motion estimation algorithm named Multi-Path Search using small diamond search pattern. Because some motion estimation which use fixed template has the shortcomings of falling into local optimal easily. According to the characteristic of the center-based motion vector distribution for the real-world video sequence, MP-SDSP utilizes more than one path to find the absolute minima. Wrong search paths are avoided early resulting in faster search. Not only ensuring the accuracy maximum in searching, but also avoiding the reduction of search speed when using a small cross pattern to find the optimal motion vector from the multiple paths. Finally, Experimental results indicate that MP-SDSP algorithm can reduce the number of search points to improve the search speed with the similar search precision, and it does well in stationary and large motion searching.