基于线性搜索的快速运动估计算法

为了减小快速运动估计算法的计算复杂度和提高运动补偿的准确性,提出了一种新的块匹配运动估计算法,称为线性正方形搜索算法.该算法采用运动估计的线性搜索策略,对于不重要的搜索区域利用线性搜索技术进行快速搜索以减小算法的计算复杂度,而对于重要搜索区域,即最佳点所在区域,用9点的正方形模块进行精细搜索以提高算法的搜索精度.实验结果证明,该算法与菱形算法相比不仅计算复杂度减小了10%以上,而且视频编码效率可以提高约0 1dB.

New Fast Motion Estimation Algorithm Based on Line Search

In order to reduce the computational complexity of the fast motion estimation and improve the accuracy of motion compensation, a new block-matching algorithm called line-square search (LSS) algorithm was proposed, in which the strategy of the line search was introduced. The LSS algorithm performed the line search for the unimportant area to reduce the computation complexity. For the important search area in which optimal points were existed, a square search pattern consisted of 9 checking points was used to carry out the refined search, thus the search accuracy and the prediction quality were improved. Compared with the diamond search algorithm, experimental results showed that the computational complexity could be reduced up to 10% and the coding efficiency could be increased about 0.1 dB by the LSS algorithm.