一种改进的SURF算法的红外和可见光图像的配准

Image registration is the basis of image processing. As the Harris algorithm is depend on scale and the SIFT(Scale Invariant Feature Transform) Algorithm is depend on the hardware, an image registration based on SURF(Speeded Up Robust Features) algorithm is proposed. First, the correspondences are found by SURF algorithm. Second the descriptor is made by computing the sum of Haar wavelet responses. Then matching points are using Euclidean distance. To improve the matching speed, the features which have the same type of contrast are only compared. The effectiveness of the method is also verified by the experiment.