基于最小生成树相似测度的耳廓匹配

Ear matching based on similarity measure of MST

将树匹配引入身份识别,将每只耳廓表示为一棵最小生成树,通过两棵树的相似测度衡量两只耳廓的相似度,以完成耳廓匹配。首先从耳廓点云数据中选取一定数量的关键点;然后分别对两只耳廓的关键点集采用Kmskal算法生成最小生成树,计算两只耳廓匹配产生的3个相似度测量值;最后经过置信加权求和,求出两只耳廓之间的整体相似度,进而完成耳廓匹配。

Tree matching is applied to identification in this paper . Each ear is represented by a MST and the ear matching is accomplished by comparing the similarity of two trees to measure the ear′s similarity . A certain number of keypoints are selected from the 3D ear firstly . The MST is constructed by adopting Kruskal to the keypoint sets of the two ear secondly . We match the trees by computing the similarity measures between the two ear , finally accomplish the recognition by using a confidence weighted sum rule to obtain the combined similarity .