Curvature-Driven Remote Sensing Image Cloud Coverage Area Restoration Method and System

Abstract

This invention provides a curvature-driven method and system for restoring cloud-covered areas in remote sensing images. The method includes performing curvature transformation on registered target and reference images to generate corresponding image curvature maps; constructing a curvature cost function based on image curvature differences to extract optimized cloud-covered area boundaries; constructing an image curvature equation based on the optimized cloud-covered area boundaries to generate an initial cloud-free image; and further refining the initial cloud-free image to obtain the final restoration result. This includes calculating the deviation between known and reconstructed values ​​of pixels in the newly added curvature-optimized area within the optimized cloud-covered area, identifying pixels with errors in the initial reconstructed image, and correcting these errors to obtain the final reconstructed image. Compared to existing methods, this invention proposes for the first time to transform the target image to the curvature domain for restoration and then inversely transform it back to the natural domain, overcoming the influence of radiometric differences between images on the restoration result.

Images