摘要
在研究相关视觉中的X光图像点匹配问题时,依据相关视觉中两幅图像的点具有位置相似性的特征,提出了基于“最小欧氏距离和约束”的匹配准则。该准则不同于一般点匹配问题中的基于灰度信息和基于区域特征的方法,它是利用顺序匹配算法原型,并依据两幅图像中标志点满足局部范围内对应点的相对位置不变这个条件,采用进化规划搜索算法不断调整标志点序列的顺序,通过使其对应点欧氏距离和最小来逼近最佳结果,从而在较低时间内取得了绝大部分点正确匹配的效果。经过实例测试和比较,验证了算法的有效性。最后总结了该算法的思想,并推广了它的应用。
According to the position similarity of corresponding points in multi images of correlation based on computer vision, a corresponding point matching rule based on “the regulation of the minimum summation of euclid distance” is presented when studying the problem of corresponding points matching from X ray images. This rule is different from the conventional corresponding points matching methods based on gray level and based on regional geometric feature. It employs the model of sequence matching algorithm. According to the condition that the relative position of corresponding points in two images are almost unchanged in adjacent area, this rule minimizes the summation of corresponding points distance to match the corresponding points by adjusting the sequence of feature points array with evolutionary programming searching algorithm. By the experiment on PC, the result demonstrates that this algorithm can match the most of corresponding points correctly in a low time cost just based on the position similarity of corresponding points. The idea of this rule is summarized and its application is generalized in the end.
出处
《中国图象图形学报（A辑）》
CSCD
北大核心
2005年第1期81-86,共6页
Journal of Image and Graphics
基金
陕西省科技计划项目(JHJYB20046)
关键词
点匹配
图像
匹配算法
搜索算法
进化规划
灰度
视觉
标志点
局部
位置
multi view, position similarity, corresponding points matching, the regulation of the minimum summation of euclid distance, evolutionary programming
