Research And Improvement Of Algorithm For Video Compression Coding Based On H.264

This thesis analyze the new generation of video coding standard H.264motionestimation algorithm, be aimed at the classic algorithm in motion estimation algorithm,have carried out research and improvement on entire pixel UMHexagonS algorithm.After deeply researching on the entire pixel UMHexagonS algorithm, we can find thatthe UMHexagonS algorithm synthesis function is advantageous, but it also hasdisadvantages, such as heavy arithmetic, highly complicated, much reconnaissancecount etc. According to the defects in this algorithm, this article has brought forwardsome optimized and improved methods on the feasibility theory. on one hand, in orderto decrease the arithmetic measures and reduce the complicated degreed, on the otherhand, in order to reduce the search points, the article use diamond-octagon searchpattern replace spiral searching templates and use multi-level octagon searchingtemplate replace multi-level hexagon searching template, and finally put forward a kindof quadrant regional segmentation method. After optimization and improvement, theUMHexagonS in the whole process of the searching algorithm could reduce predicttime obviously and get a good exercise vector searching results.Finally, take the simulation test to the UMHexagonS algorithm which has beenimproved, in order to checking the performance effect of the algorithm. Use H.264reference software model JM10.1, compiled by VC++6.0, realized the algorithm, basedon the experiment of small, medium and large sports movement characteristics’s videoimage test series, we can prove that the UMHexagonS algorithm after optimization andimprovement which compared with the original algorithm, under the condition ofinvariant in the PSNR-Y, bit rate and reconstruction images quality, movementestimates time has decreased17%-30%, reduced the computational complexity, andimproved the code efficiency.