The remote sensing image matching based on SURF algorithm

Abstract

Through the simultaneous photography technology of multi-spectral camera, the remote sensing image is composed of the different spectral of scenes’ electromagnetic spectrum. Having multi-spectral characteristics and rich image information, the remote sensing image is now used widely in many occasions. At the same time, how to achieve the remote sensing image matching rapidly and accurately has been becoming a research focus in the image processing filed. To improve the matching accuracy, this paper, based on the SURF algorithm, uses the Mahalanobis distance affine-invariant character and gets rid of the initial wrong matching points. Numerous experiments were carried out on remote sensing images with geometric distortion such as scale, rotation and so on. The result shows that the effectiveness of the method.