Research On Remote Sensing Image Registration Method Based On Point Feature

Image registration is the process of matching two or more images of the same scene with different time,different viewpoints,and different sensors.Image registration technology has been widely applied to many fields in society,for example,panoramic image stitching in computer vision,evaluation and diagnosis of diseases in medical image analysis,change detection and image fusion in remote sensing images and so on.With the development of image acquisition technology,the types of remote sensing images are increasing,and the difference between the images is also increasing.However,traditional methods can not achieve accurate registration,so there are still many problems to be solved in remote sensing image registration.Feature-based method is currently the most widely used method for remote sensing image registration,among them,the point feature is not only good for the geometric transformation of the image,but also requires a low degree of coincidence between the images,in practice,it also has high computational efficiency and has been widely used.Therefore,according to the characteristics of remote sensing image,this thesis makes a deep research on image registration technology based on the classical scale invariant feature transform(SIFT)algorithm and anisotropic scale space theory,and proposes two remote sensing image registration algorithms based on point features:In view of the large gray difference between remote sensing image pair can easily result in matching failures,this thesis propose a remote sensing image registration algorithm based on modified SIFT and enhanced feature matching to this problem.In this letter,a new gradient definition is first introduced,experimental results show that the proposed gradient definition method can effectively overcome the difference between remote sensing image pairs.Then,the enhanced feature matching is achieved by combining the position,scale,and orientation of each keypoint.The experimental results show that the enhanced feature matching method can effectively increase the number of correct matching points.In view of the traditional Gauss scale space fuzzy the image edge,thus leads to the problem of poor positioning accuracy of feature points,this thesis propose a remote sensing image registration algorithm based on anisotropic diffusion and Harris.In this algorithm,the anisotropic scale space is constructed according to the anisotropic diffusion principle,which can preserve the edge information of the image and increase the precision of the detected feature points.Then the Harris operator is used to detect the feature points in the anisotropic scale space,which increases the anti noise properties of feature points.Experimental results show that the algorithm not only increases the number of feature points,but also improves the registration accuracy.