作者: Bo Huang , Ying Li , Xiaoyu Han , Yuanzheng Cui , Wenbo Li
DOI: 10.1109/LGRS.2014.2377476
关键词:
摘要: This letter presents a cloud removal method for reconstructing the missing information in cloud-contaminated regions of high-resolution (HR) optical satellite image (HRI) using two types auxiliary images, i.e., low-resolution (LR) composite (LRI) and synthetic aperture radar (SAR) image. The LRI contributes low-frequency information, SAR high-frequency restoring HRI. approach is implemented structure correspondences established by sparse representation. Specifically, dictionary pairs are trained jointly: One pair generated from HRI gradient patches, other patches. Experimental reconstructions HR Thematic Mapper images performed three MODIS 16-day only, both SAR, respectively. It shown that or data alone not sufficient to restore whereas combination can provide low- information. proposed achieve highly accurate result has potential areas where land-cover change may occur.