| Motion estimation(ME) process is the most computationally expensive operation in the entire video compression process for H.264. Because it has a high coding efficiency and it is very scalable, UMHexagonS algorithm is considered to be the most popular ME algorithm. However, the main drawback of the UMHexagonS is that it is computationally intensive, which limits the scope of its application. This paper researched and analysed the UMHexagonS with JM14.2, and improved it based on the analysis results. Some new search templates and the motion acuteness degree of image were used by the proposed algorithm. These new search templates can flexibly predict optimal matching object’s region, the motion acuteness degree of image can reduce the scope of the search and number of searches about the matching objects, they are able to reduce the computational complexity of motion estimation. Simulation results illustrate that the proposed algorithm achieve reduction of15%encoding time on average compared with UMHexagonS algorithm,with almost the same PSNR about video image, in the premise of effectively improving the encoding speed. |
PDF Full Text Request