TOTAL VARIATION REGULARIZED LOW-RANK SPARSITY DECOMPOSITION FOR BLIND CLOUD AND CLOUD SHADOW REMOVAL FROM MULTITEMPORAL IMAGERY

Yong Chen,Wei He,Naoto Yokoya,Ting-Zhu Huang
DOI: https://doi.org/10.1109/igarss.2019.8900086
IF: 12.7
2019-01-01
ISPRS Journal of Photogrammetry and Remote Sensing
Abstract:This paper proposes a spatial-spectral total variation (TV) regularized low-rank sparsity decomposition model for blind cloud and cloud shadow (cloud/shadow) detection and removal of multitemporal remote sensing imagery. Our concept is to decompose the contaminated image into the surface-reflected component and the cloud/shadow component. Low-rank regularization is utilized to model the spectral-temporal correlation of the surface-reflected component, meanwhile, the ` 1 -norm and spatial-spectral total variation regularization is employed to describe the sparse prior and spatial-spectral continuity of the cloud/shadow component. To better preserve the information in cloud/shadow-free areas, the cloud/shadow detection results obtained as a by-product of our method are used to guide the information compensation from the original contaminated images. Several experiments are presented to demonstrate the effectiveness of the proposed method.
What problem does this paper attempt to address?