基于层次基元结构的优化匹配算法

匹配是计算机视觉中的一个重要而又必不可少的步骤.本文针对运动图象的对应问题,给出了一个新的基于层次基元结构的匹配算法.整个算法分成两步,第一步是层次基元结构的构建.在使用~2G算子获取零交点后,基于一定规则进行聚类,可以构建零交点、零交轮廓线片段、零交轮廓线这样的层次基元结构.第二步是基于相似度准则的由粗至精的匹配过程.本文采用松驰迭代与约束树搜索相结合的策略,获取优化的匹配结果.对真实图象的实验表明,该算法是有效的.

AN OPTIMAL MATCHING ALGORITHM BASED ON HIERARCHICAL PRIMITIVE STRUCTURE

A new matching agorithm based on Hierarchical Primitive Structure(HPS) is presented for sequential images analysis. It consists of two main steps. The first is the construction of HPS. We extract zerocrossings using Laplacian of Gaussian operator (2G) and assemble them according to certain rules. HPS is established as hierarchical collections of zerocrossing points, zerocrossing segments and zerocrossing contours. The second is a coar-se-to-fine matching process based on similarity criteria. The strategy which combines relaxation and constrained tree-searching is employed to obtain the optimal matching result. Experimental results are demonstrated for real images.