A fast two-step algorithm for large-area thick cloud removal in high-resolution images

Abstract

ABSTRACT Clouds are inevitable in optical satellite images, and affect the subsequent processing and application of the images. Thick cloud removal in high-resolution imagery is challenging due to the complex spatial and radiometric variations. In this letter, we present a two-step thick cloud removal method based on fast radiometric adjustment and residual correction (FRARC), which is effective for large-area thick cloud removal in high-resolution images. Compared with other thick cloud removal methods, the proposed FRARC method achieved both satisfactory results and a high efficiency in simulated and real-data experiments. In addition, as the area covered by clouds in the imagery increases, FRARC shows a more significant improvement in efficiency compared to the other methods. The proposed FRARC method can be applied to generate cloud-free images to support seamless mapping with high-resolution images.