Affine invariant feature and coastline constraint-based automatic coastal zone remote-sensing image registration method

Abstract

An automatic registration method for coastal zone remote sensing images based on affine invariant features and coastline constraints. The steps are: coastline and land area image extraction; generation of SIFT feature points and feature vectors; matching of SIFT feature points; calculation of homography matrices; and the corresponding When coastlines overlap, calculate the average cross-section distance between coastline pairs; adjust the distance threshold of the RANSAC method, and recalculate the cross-section distance between coastline pairs; take the instance before the cross-section distance mutation as the optimal instance. The optimal instance homography matrix is ​​the optimal homography matrix; image registration is completed using the optimal homography matrix. The invention overcomes the problem of difficulty in image registration due to complex land features and lack of water features in coastal zone images, and can effectively perform automatic registration of coastal zone remote sensing images with high registration accuracy and automation.

Description

technical field [0001] The invention relates to an automatic registration method of remote sensing images of coastal zones, in particular to an automatic registration method of remote sensing images of coastal zones based on affine invariant features and coastline constraints. Background technique [0002] Coastline and coastal land use mapping is crucial to waterway safety, coastal resource management, coastal environmental protection and sustainable development of the coastal zone. With the increasing development and utilization of coastal areas, the application of remote sensing technology in coastal areas is becoming more and more extensive. In order to make full use of a large number of remote sensing images with different resolutions, different time phases, and different ranges, it is necessary to automatically register remote sensing images instead of manual registration. The coastal area is at the junction of the land and the ocean, and the water area is vast, but t...

AI Technical Summary

Problems solved by technology

This method can register remote sensing images of coastal zones to a certain extent, but the registration accuracy depends on the image quality, and the slow registration speed does not fully utilize the information of coastal zone images

The few existing registration methods for coastal remote sensing images mostly improve the corresponding parameters on the traditional method, and the registration accuracy is not high

Images