Remote Sensing Image Matching Of Different Sources

Image is one of the most important means for human to obtain information from the world.However,just an image can not give us the exact information we want to get.Through image matching technology,we can quickly get the information we want from the image.With the development of science and technology,we can obtain images in different ways,which gives birth to a new problem-heterogeneous image matching.With the development of science and technology,we can obtain images not only through optical imaging,but also through a series of new methods such as synthetic aperture radar images.Optical imaging satellites often can’t get real-time images because of weather interference,the appearance of clouds,vegetation or artificial camouflage,but we can get them through radar.However,the gray level of SAR image is different from that of optical image,so it is difficult to match the image.In this paper,the heterogeneous image matching is divided into two steps.Firstly,because the heterogeneous image is interfered by various kinds of interference,it will produce different degrees of degradation.Therefore,we build a deep network drnfnb to solve the degradation of heterogeneous image.Secondly,we introduce multi-directional Gabor wavelet to extract the optimal direction information of the point,and match the two heterogeneous images,It is converted to the matching of the optimal direction of two heterogeneous images.Based on our method,we participated in the optical image and SAR image matching competition held by the rocket army in October 2020,and achieved good results.The experimental results show that our method is successful.This paper is divided into four parts: the first part introduces the current situation of heterogeneous remote sensing image matching;the second part introduces the difficulties in heterogeneous remote sensing image matching;the third part describes the commonly used image matching algorithms and the better methods in recent years;the fourth part describes the heterogeneous remote sensing image matching based on multi-directional Gabor wavelet filtering.Finally,the experimental results of remote sensing image matching based on Gabor wavelet are introduced.The experimental results of this paper are obtained on the real data,but because the real experimental data is secret,all the pictures and principle demonstration of this paper do not involve the real data,so it is hereby explained.