Research On Hierarchical Registration Of Remote Sensing Images

Image registration is a process of geometrically aligning two or more images obtained under different conditions in the same area,through which the homologous points and surfaces in the images are in the same space.The registration of remote sensing images is to eliminate the effects of the aberrations on comprehensive analysis of multi-modality remote sensing image information,and make different types of remote sensing images can be used to complement the information of the same geographic target at different time and different spatial scales,which is of great significance for disaster emergency assessment,urban planning,military investigation,land use,weather forecasting and other applicationsIn allusion to the complex deformation problem of rigid deformation and elastic deformation exist simultaneously affected by the factors as imaging angle,terrain undulation load type and geographical environment change etc.,this paper makes a thorough and system study on rigid registration method and elastic registration method and improves the traditional remote sensing image registration method.Based on the registration strategy of from whole to local,from course to fine,layer by layer registration,this paper combines the rigid registration method based on affine transformation and the elastic registration method based on B-spline function to achieve high precision registration result.Using the algorithm to carry out the registration experiments of the aerial images and satellite remote sensing images with complex image distortion.Through visual analysis and objective analysis,it is proved that the algorithm can obtain high precision remote sensing image registration results and the hierarchical registration method is applicable and effective in the field of remote sensing image registration.