Study on the Unified Theory of Thin Cloud Detection and Removal based on Physical Model

Clouds greatly affect the quality of optical remote-sensing image data. Algorithms for detecting and removing clouds can significantly enhance the utilization of optical data. Numerous studies highlight the crucial role of cirrus bands in cloud detection and removal, although only a few satellites p...

Full description

Saved in:
Bibliographic Details
Main Authors: Lyu, Haitao, Qian, Jiang, Wang, Yong
Format: Conference Proceeding
Language:English
Subjects:
Artificial satellites
Clouds
detection
Earth
Optical imaging
Optical reflection
physical model
Reflectivity
removal
Satellites
Thin clouds
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Clouds greatly affect the quality of optical remote-sensing image data. Algorithms for detecting and removing clouds can significantly enhance the utilization of optical data. Numerous studies highlight the crucial role of cirrus bands in cloud detection and removal, although only a few satellites possess this capability. In this research, a unified theory based on physical model is proposed and validated for thin cloud detection and removal. First, top of reflectance (TOA) values of thin clouds are detected in a specific band. Subsequently, the detected cloud image is used to remove thin clouds from other bands via spatial transformation. Experiments with actual Landsat-9 Operational Land Imager 2 (OLI-2) data confirm the effectiveness of the proposed approach both qualitatively and quantitatively. Even if thin clouds are undetectable in the cirrus bands, or cirrus bands are unavailable, this research still presents a novel paradigm for detecting and removing thin clouds.
ISSN:2153-7003
DOI:10.1109/IGARSS53475.2024.10642238