A registration and correction method and system for multispectral remote sensing images

Abstract

The invention belongs to the technical field of image processing and provides a registration and correction method and system for multispectral remote sensing images. In the method and system, one of the multispectral remote sensing images that has undergone preliminary geometric correction is selected as a reference image, and the rest are used as images to be registered. First, the best image sub-block is extracted from the reference image, and then the best image sub-block is combined with each The image to be registered is matched at the sub-pixel level in the dual-scale space to obtain the matching point, and then the projection transformation matrix is ​​calculated according to the matching point, and the corresponding spectral image is projected into the coordinate space of the reference image, and then the necessary image interpolation is adopted, A multispectral image registered and corrected in the coordinate space of the reference image is obtained. Compared with the existing multispectral remote sensing image registration method, the method has high calculation efficiency and high registration accuracy, and is especially suitable for registration and correction of large format multispectral remote sensing images.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a registration and correction method and system for multispectral remote sensing images. Background technique [0002] In the imaging of multi-spectral remote sensing images, there are generally certain aberrations between large-scale remote sensing images with different spectra after preliminary geometric correction. The types of aberrations include: small spatial translation errors, small rotation errors, small The scaling error of , and the error resulting from the combined action of these three errors. In a specific image, it generally shows that in different regions of the image, the mismatch errors in vertical and horizontal directions between spectral images are generally different. It is often the case that two images with different spectra appear as translation and rotation errors in local observation, that is, the aberration can be basically elimin...

AI Technical Summary

Problems solved by technology

[0011] The purpose of the embodiments of the present invention is to provide a registration and correction method for multi-spectral remote sensing images, aiming at solving the problem of the existing registration methods for multi-spectral remote sensing images or by manually selecting feature point pairs, or using gray-based The realization of the automatic registration algorithm based on the degree of precision, or the realization of the automatic registration algorithm based on the feature, the three methods all have the problem of low calculation efficiency and / or low registration accuracy

Images