Multi-source remote sensing image registration method

Abstract

The invention discloses a multi-source remote sensing image registration method which is characterized in that size ratio of multi-source remote sensing images is considered, target images and source images are respectively subjected to appropriate partitioning processing, surface (SURF) feature points are subjected to partitioning detection and matching, areas of interest (ROI) are determined, the ROI are matched according to a certain strategy, extraction is carried out according to the partitioning image matching condition, and finally the integral matching relation of the target image and the source image is obtained. The multi-source remote sensing image registration method had the advantages that quadrant division is achieved through detection and matching of a few of partitioned SURF feature points, accordingly the ROI of images are determined, and the purpose of improving registration efficiency is achieved by combining direction selection and feature point number threshold screening. Simultaneously, error rate of the original SURF feature matching is reduced by determining the radiation transformation relationship of the maximum matching points and image partitions serving as the original images.

Description

technical field [0001] The present invention relates to the technical field of remote sensing image processing and application, in particular to a multi-source remote sensing image registration method, the method is based on quadrant division and direction selection block improved SURF algorithm, mainly used for fast Accurately determining the fusion area is the premise of realizing multi-source remote sensing image fusion. Background technique [0002] As an important means of obtaining information, remote sensing can obtain a large number of remote sensing images with different spatial resolutions, spectral resolutions, and temporal resolutions. Obvious limitations. Using effective methods to fuse multi-source remote sensing images can lay a data foundation for subsequent specific applications such as target segmentation, recognition, and tracking. Remote sensing image registration is the process of matching and superimposing local features on multiple remote sensing ima...

AI Technical Summary

Problems solved by technology

[0004] Although the SURF feature has greatly improved the time efficiency compared with the previous SIFT, PCA-SIFT, GLOH and other algorithms, the SURF algorithm still uses the main direction determination and gradient value summation process in the descriptor calculation. Descriptor calculation still has high complexity

Moreover, the error rate of SURF feature registration needs to be further reduced. For large-scene remote sensing images, its time complexity is still high

Images