Research and implementation of image feature point matching method based on OpenCV

摘要

This paper described the distance metric of the Euclidean distance of SIFT operator and Hamming distance of BRISK operator represented based on OpenCV to extract and descript feature points on the two relatively low altitude remote sensing images, and then chose BruteForceMatch (violent match) and FlannBasedMatch (approximate Closest match) two matching methods to match in the same matching operator. Lastly, the RANSAC algorithm was chose to estimate the fundamental matrix, eliminating the effects of error matching on the fundamental matrix accuracy and stability. By using extreme geometric constraints to reject the mistake matching points in error matching, and improve the robustness and accuracy of the matching. The results show that the performance of FlannBasedMatch method is more advantageous.