一种计算复杂度可调整的块匹配运动估计算法

提出一种计算复杂度可调整块匹配运动估计算法，该方法综合传统三步法中收缩搜索策略和四步法及钻石法中步进搜索策略的优点，对于分散分布的运动矢量及聚中分布的运动矢量搜索都表现出良好的鲁棒性．搜索模式的规整性也使得该算法的VLSI结构设计比以往的算法实现更加简洁，通过对搜索步数和块匹配中降采样率的控制，实现了对运动估计计算复杂度的调整，从而可在计算复杂度与图像处理质量之间折中．该特征为视频编码系统整体的低功耗设计提供前提条件，适用于低复杂度低功耗视频编码器．

Algorithm and VLSI architecture for computational complexity scalable block marching motion estimation

A computational complexity scalable algorithm(CSS) is presented for block matching motion estimation in video encoding.The CSS shares the feature of fast convergence in TSS and the advantage of center bias searching in FSS and DS,and with the uniform shape of searching pattern in algorithm,the VLSI implementation of the CSS becomes easier than that of other fast motion estimation algorithms like DS.The scalability of computation payloads has been achieved based on searching steps adjusted and block pixel sub-sampled.The proposed algorithm and architecture are suitable for the low power implementation of video encoding that needs flexible scalable capability.