No abstract

No abstract

Satellite image have recently been wide used in academia and industry, and thus providing high quality and even cloud-free satellite image is a fundamental and important issue. Globally, the land scenes are on average about 35% cloud covered, as reported by Ju and Roy [1], indicating that cloud covers are generally present in optical satellite images. This phenomenon limits the usage of optical images and increases the difficulty of image analysis. Considerable research efforts have been devoted to the issue of cloud removal to ease the difficulties caused by cloud covers [2][3][4][5][6][7][8][9][10][11][12][13]. These efforts and studies focus on how to detect clouds and how to reconstruct the information of cloudcontaminated pixels. However, information reconstruction is generally sensitive to the feature structures lying on the boundaries of the cloud-contaminated areas. In this paper, we address the topic of determining optimal boundaries of cloud-contaminated areas for the purpose of cloud removal.