Image Matching In Remote Sensing Based On Local Binary Pattern With Equidistant Rings And A Method Of Combining Coarse With Fine Matching

Matching method based on local invariant feature is a hotspot of image processing in recent years,and it has been widely used in various fields.In this paper,local invariant features approach is adopted to study image matching algorithm of remote sensing.The matching methods to the remote sensing images,including homologous and heterologous remote sensing images,has its some research difficulties.Therefore,to homologous remote sensing images matching,the difficulty is how to improve real-time responding,while to heterogeneous remote sensing images matching,the problem is how to reduce nonlinear intensity differences between the images.To solve the above-mentioned problems,the proposed approaches are as following:In order to solve the real-time problem of remote sensing image matching,the current solution is using the binary descriptor method in image matching,and to achieve rotation invariance,the binary descriptors are generally associated with principal direction.However,deviation of the principal direction of the descriptors usually occurring in computational procedure cannot achieve ideal matching results based on these descriptors.Therefore,a binary descriptor,named local binary pattern with equidistant rings,is proposed to solve this problem.The sampling model of local binary pattern with equidistant rings and its group matching strategy are introduced into descriptor constructing and descriptor matching in order to guarantee the descriptor is rotation-invariance.Thus,the deviation problem of principal direction is solved and the speed of descriptor matching is improved.The experimental results show that our approach is more accurate and more speedy,and also can effectively solve the deviation problem of principal direction.It is difficult to match the heterogeneous remote sensing images.The main reason is that the heterogeneous images are obtained from different imaging devices,which produce the large nonlinear intensity differences.Consequently,the traditional gray-based and gradient-based matching method cannot have better performance in heterogeneous image matching.To solve this problem,a new method of combining coarse matching with fine matching is presented in this paper.In the coarse matching step,the displacement,rotation and scale among the images are eliminated,while in the fine matching step,to improve matching accuracy,the non-linear intensity difference between the images is eliminated and the template matching is employed.Experimental results demonstrate that our method can effectively solve the iamge matching of heterogeneous remote sensing images.