A Fast and Robust Registration Method of Visible Light Image and SAR Image Based on Structural Similarity

Abstract

The invention discloses a fast and robust registration method for visible light images and SAR images based on structural similarity. First, the rational polynomial coefficient model RPC is used for rough registration of the visible light image to be registered and the SAR image; then, the improved corner detection method is used to detect the interest points of the visible light image, and then a certain number of interest points are uniformly selected as control points; then use the improved direction Harris corner histogram HIOHC to construct a similarity measure to perform key point matching in the SAR image; after removing obvious mismatches, use the least square method LSM to obtain the global transformation matrix and obtain the final Registration results. This method breaks through the dependence of traditional methods on image descriptors, and at the same time breaks through the sensitivity of traditional methods to different data sources, and achieves ideal results. It realizes high-precision registration of visible light images and SAR images. Change detection, loss estimation, etc. have important practical significance.

Description

technical field [0001] The invention relates to a registration method for heterogeneous remote sensing images, in particular to a fast and stable registration method for visible light images and SAR images based on structural similarity. Background technique [0002] With the rapid development of geographic information technology, remote sensing sensors have significantly improved both in data quality and characteristics, and more and more remote sensing sensors are installed on remote sensing satellites. According to the method of obtaining information from remote sensing satellites, remote sensing satellites can be briefly divided into two categories, the first is optical sensors that passively receive information, such as GF1, Ikons, etc., and the other is synthetic aperture radar (SAR) sensors that actively acquire information , such as GF3, TerraSAR-X, etc. SAR sensors can emit microwave signals and record signals emitted by the surface, so they can penetrate the inter...

AI Technical Summary

Problems solved by technology

[0005] However, for the registration of visible light images and SAR images, the methods based on MI or PC are relatively slow, and are often not suitable for large-scale remote sensing images, and SAR images have certain changes in brightness and contrast, which lead to the existing registration methods. difficult to achieve desired results

Images