一种新的二维碎片的轮廓匹配方法  

New method of contour matching based on two dimensional fragments

作　　者：姬周强[1] 王冰[1] 

机构地区：[1]西北大学信息科学与技术学院,西安710127

出　　处：《计算机应用研究》2009年第8期3189-3191,共3页Application Research of Computers

基　　金：陕西省科技攻关基金资助项目(2004K05-G25)

摘　　要：以往的轮廓匹配算法中所用的轮廓表示方法大多需要大量繁琐的计算,大大增加了算法的时间复杂度,为此提出一种新的轮廓特征表示方法,简化了此部分的计算。首先求出待匹配图像单像素宽的轮廓曲线上像素点的坐标序列,然后利用轮廓上各点和与其相差六个点的像素点之间的位置关系及行列坐标差的平方代数和对轮廓进行表示,得到两轮廓曲线的表示序列后;接着采用寻找两轮廓表示序列的最长公共子序列(LCS)的方法进行匹配,并在匹配过程中引入了"断点续配"的概念,有效提高了算法的容错性。实验证明所用的轮廓表示方法简单明了,计算量小,在提高算法效率的同时仍可保证匹配的正确性。In the past contour matching algorithm, the contour describing methods always need too much calculation. It has greatly increased the complexity of the time. The contour describing method is easier, and need less calculation, it could improve the efficiency of the algorithm and also could ensure the correct rate. Firstly got the sequence of the contour points with one pixel width of the image to be assembled. Proposed a new approach of describing the fragments＇ contour based on the relationship between the contour points＇ position. By searching the longest common subsequence （LCS） of the description of the two fragments＇ contours to find the matching curves part. During the matching process, if a point was not matched, but the points before and after it were all matched, then also thought it＇ s matched. This method improves the matching efficiency and the fault-tolerant rate.

关 键 词：二维碎片 轮廓 最长公共子序列 匹配 拼合 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]