轮廓线图象的零交叉──二元树识别算法  

Zero Crossing and Binary Tree Recognition Arithmetic for Contour Images

作　　者：王耀明[1] 张刚[1] 董建萍[1] 俞时权[1] 

机构地区：[1]上海师范大学理工信息学院,上海200234

出　　处：《计算机工程》2001年第4期107-108,115,共3页Computer Engineering

摘　　要：采用尺度空间函数──改进的 样条函数──对轮廓线的曲率函数进行波形处理，从而得出它们的分布状况而构造出波形的层次结构。并利用其曲率函数的零交叉指纹图的交汇点建立相应的二元树。从而，根据二元树的相关度可以判别轮廓线的相似程度。这种判别方法，在图象识别中取得了较好效果。This article puts forward the wave-form procession to the curvature of contour line by using the improved spline function as scale-function and gets their distribution and establishes the hierarchy of the wave-form. Then it makes use of the crossing points of the zero-crossing fingerprint graph of the curvature to establish the corresponding binary tree. At last we can judge the likelihood of the contours according to the correlation of the binary trees. This method of judgment can acquire a good result in image recognition.

关 键 词：图象识别 零交叉指纹图 二元树 轮廓线图象 计算机视觉 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]