Research On Multi-source Remote Sensing Image Registration Based On LBPH And SIFT

Image registration refers to the process of finding the optimal geometric transformation of the same scene image with overlapping areas captured at different times,different sensors or different viewing angles.Image registration technology is a key step in image processing,mainly used in remote sensing,military,medical analysis,computer vision and other fields.Remote sensing image registration is not only the main research content of remote sensing image processing,but also the key to whether many practical application problems such as image fusion,target detection and recognition,image mosaic,and scene reconstruction can be solved smoothly.In recent years,with the rapid development of remote sensing photography technology,the methods of acquiring remote sensing images have become more and more diversified.The grayscale differences between the captured images are obvious.Image registration faces huge challenges and has become a research hotspot at this stage.The SIFT algorithm is the most widely used method based on image local feature registration.It can maintain stability for rotation,scaling,and illumination changes,but its calculation complexity is high.Our article analyzes the SIFT algorithm and the face recognition algorithm based on the LBPH feature,and proposes a method of image registration using the LBPH feature.The algorithm in our article mainly has the following innovations:first,by analyzing the implementation method of CS-LBP operator,a new and simpler 4-LBP operator is proposed;second,the use of new 4-LBP features combined with image spatial information It constitutes a LBPH feature with a simpler feature vector dimension.Third,image registration is achieved by combining the SIFT algorithm with the new LBPH feature.In order to verify the wide applicability of the algorithm in our article,Our article randomly selected four sets of remote sensing images from the spatial geographic data cloud for registration research.Through comparative experiments,it is found that compared with the traditional SIFT algorithm,the algorithm can better meet the real-time requirements of image registration.