A Fast Image Matching Algorithm Based on Improved SURF

Abstract

To improve the accuracy of image matching and reduce the computational complexity, a fast image matching algorithm was proposed based on improved SURF. Firstly, determine the edge regions of the image, and extract the SURF features of these regions, Secondly, according to neighborhood Shannon entropy of the key points, remove those with less information, Thirdly, form the SURF feature vector, and match images by the Quasi-Euclidean distance, Finally, use the RANSAC and the least squares method to remove false matching points, and determine the matching areas between the two images. The experimental results showed that the algorithm maintains the robustness and stability of SURF, and the matching accuracy and the running speed are better than those of SURF.