Cloud Removal Advances: A Comprehensive Review and Analysis for Optical Remote Sensing Images
Jin Ning, Lianbin Xie, Jie Yin, Yiguang Liu
Abstract
Cloud cover significantly decreases the quality of Optical Remote Sensing (ORS) images, adversely impacting its effectiveness in geographic monitoring, disaster prevention, and advanced visual applications. This phenomenon has made cloud removal a critical preprocessing step in ORS image processing. This paper comprehensively reviews cloud removal techniques and classifies them based on the type of auxiliary data used: single-image, multimodal, and multitemporal. The discussed methods include physical modeling, deep learning, multispectral analysis, and Synthetic Aperture Radar (SAR) fusion strategies. This paper analyzes the core concepts and fundamental processes of these techniques and addresses the challenges encountered in actual scenarios. The paper also includes future research directions. Moreover, the paper outlines the benchmark datasets and evaluation metrics commonly used in cloud removal, thereby establishing a standardized reference for algorithm development and performance evaluation. A thorough comparative analysis was performed to assess their performance variations using visualization outcomes from the most recent and representative methodologies.