一种基于边缘积分与邻域提纯的图像匹配方法

摘要

For questions that Harris algorithm detect corner,the threshold is set to be repeated,the paper has been im-proved on the Harris algorithm,using the edge of the integral image,the number of edge point on neighborhood to deter-mine the corresponding threshold.Meanwhile,the NCC algorithm was further improved by comparing the point of the be-ginning of the inner case of matching the distribution of edge points,the matching point of the early eluting reduces the run-ning time of the algorithm.%在利用 Harris 算子检测角点时，不同的图片有时需要需多次设置阈值，才能得到理想的角点分布，本文对 Harris 算子作改进，采用边缘积分图像，统计邻域内边缘点的数量，确定对应的阈值。同时，对NCC 算法作了进一步的改进，通过比较初匹配点对内边缘点分布的情况，对初匹配点进行提纯，降低了算法的运行时间。