Abstract: Motion estimation is a core process for video coding schemes, which enables the transmission and storage of video signals with a low bit rate. In order to decrease calculation times and enhance matching efficiency of block matching search, a fast block matching motion estimation algorithm based on edge partial distortion search was proposed. Based on the block matching criterion, the algorithm forecasts the motion vector by sorting the similarity between the current image block and its neighborhood image block, and then chooses the motion vector of neighborhood image block with the highest similarity as its own motion vector. At the same time, edge partial distortion strategy was introduced, and the computing of neighborhood image block with the bigger deviation was prematurely terminated. The simulation result shows that the algorithm reduces the calculation amounts and promotes the matching precision in comparison with other fast standard block matching methods.

Key words: Edge partial distortion, Sort, Block matching, Motion estimation,  Similarity

摘要： 运动估计能够将视频信号以较低的比特率进行传输和存储, 是视频编码中核心的处理过程之一。为
了降低块匹配搜索的计算次数, 提高匹配效率, 提出一种基于边缘局部偏差搜索的快速块匹配运动估计算法, 根据
块匹配准则将当前块的邻域块与当前块的相似度进行排序, 将相似度最高的邻域块的运动向量作为当前块的预测
运动向量, 同时引入边缘局部偏差搜索策略, 可提前终止偏差较大的邻域块的计算。仿真结果表明, 与其他快速标
准块匹配方法相比, 本文算法减少了计算次数, 提升了匹配精度。

关键词: 边缘局部偏差, 排序, 块匹配,  , 运动估计,  , 相似度