一种用于图像拼接的角点匹配算法

针对基于自适应光学技术的人眼眼底待拼接图像灰度、清晰度不一致以及细节的差异使检测出的角点无法匹配,进而导致图像不能拼接的问题,提出一种在OpenCV配置环境下角点检测的基础上,进行角点匹配的新方法.该方法利用角点间的距离和斜率在眼底很小的区域内基本不变的特性,将两幅图像的角点对相互匹配,进而找到两幅图像中三对一一对应的角点.实验结果表明,用该方法可以解决人眼图像拼接的角点检测和角点匹配问题,而且运算速度快,检测效率高.

Method of Corner Matching Used in Image Registration

Considering the difficulty of the registration of two images of the eye-ground gotten by adaptive optics principles,a new corner matching method was proposed,which was based on corners detection in the OpenCV environment.This method applied the invariability of the distance and slope of two image’s corner couples in a small area of the eye-ground to filtrate out three couples of corners which were parallelism in two images.The result shows that three corner couples are found rapidly and exactly by this method,so the corner detection and the corner matching of two images in the eye-ground are solved.