Essential Algorithm Research In Video Compression

In the video encoding and video processing, motion estimation and motion compensation technology to reduce the redundancy of video sequence time, improve the coding efficiency plays the very essential role. The accuracy of motion estimation directly determines the coding efficiency of video encoder. Moreover, the motion estimation that takes the entire encoder coding-time 50% to arrive at 80%[1], its searching algorithm's complexity will decide the video compression coding system's difficulty degree. Therefore how to improve the accuracy of motion estimation ,cause its search process to be fast highly effective, has been the hot spot which the people study.This paper first studies several classic search algorithms for fast motion estimation, analysis their respective technical characteristics in detail and summarizes the key factors on the impact of the efficiency of motion estimation algorithms. Based on this proposed, an adaptive hybrid template fast search algorithm is made. This algorithm is in the hexagon searching algorithm foundation, according to the center-biased characteristics of motion vector, combineding with the initial search point prediction, adding the suspension of identification techniques and so on. The hexagon searching algorithm is improved to choice the template between small diamond and small square shape adaptively, which causes it to conform to the realistic video. In addition, in the field of pixel search algorithms, an attempt based on classification quadrant pixel search algorithm is to be made. This algorithm does not need conventional interpolation filters, but only under the integer pixel motion estimation search results, forecast obtained spots by the pixel motion estimation. Thus it saves a large number of complex calculations. The simulation experiments indicat that the two algorithms above enhance the efficiency of motion estimation effectively and the the video premise quality is guaranteed.