Research Of Multi-source Image Registration And 3D Terrain Matching

With the development of science and technology,the resolution of the images obtained by people is getting higher and higher,and the size of the images is also getting larger and larger,which makes the amount of data processed by the image multiplied.Using a general optimization algorithm,it usually takes tens of seconds or even minutes to process such an image,and for large images(up to the GB level of data),it takes a lot of time and still cannot obtain good results.For occasions where the real-time performance of large image registration processing is high,there are currently few good algorithms to solve this problem.This paper first describes the research background and general research status of the field of large-scale/super-fast image registration.Based on the basic concepts of image registration,the basic methods and classification of image registration techniques are summarized.The contrast experiments of the image registration method show that the traditional feature-based matching method not only takes long time to extract features,but also has low efficiency,especially when performing feature-based matching calculations for large/super-large images.Based on this,this paper proposes a new fast matching method for large/super-large images based on overlapping sub-regions.It uses the constructed image pyramid to quickly index low-resolution images for rough matching,and determines the approximate overlapping area.Subsequent precise registration only needs to be performed for overlapping areas,which greatly reduces the amount of data involved in feature matching,greatly accelerates the time required for the matching process,and improves the efficiency of the algorithm.Experiments show that compared with the general registration algorithm,this method can greatly reduce the registration time under the premise that the accuracy of registration is still within one pixel.Unlike two-dimensional images,three-dimensional terrain contains more information.In practical applications,more and more analysis and processing of three-dimensional terrain.This paper first introduces three-dimensional terrain visualization technology and realizes visualization of three-dimensional terrain.On the basis of visualization,this paper transforms 3D terrain matching problem into 2D image registration problem,and uses image registration to indirectly achieve 3D terrain matching.Experiments show that this method can effectively match three-dimensional terrain.Due to the imaging mechanism of the image and the illumination intensity and the shooting angle of the scene during imaging,there are large differences between the heterologous images,especially the loss of the detailed texture in the infrared image.The feature point-based algorithm can't extract matching feature points from multi-sesor images.Therefore,a straight line-based image registration algorithm is proposed in this paper.The straight line group labels are used to describe the extracted straight line segments.After similarity measures,so as to achieve the match Experiments show that the algorithm can achieve the registration of heterogeneous images with large differences in gray levels.