An automatic cloud detection algorithm for Landsat remote sensing image

Due to the limitation of the band range of satellite sensor and the influence of the earth's atmosphere, the Landsat remote sensing images are inevitably be covered with clouds, which bring many difficulties to subsequent using and interpretation(environmental monitoring, resource investing-ation, agricultural production, ecological protection, et al). So detecting cloud accurately in remote sensing image is very importance for subsequent image processing and applications. A new algorithm combined random forest classifier and multi-features is proposed in this paper. The basic idea of proposed approach is: first, K- means clustering method is used to get the clustering center of the input features, and then a certain number of labels around the center of the cluster sample are selected. Second, random forest model is chosen for image classification. In order to verify the effectiveness of proposed method, several real date experiments have been done. The experimental results show that, compared to Function of mask (Fmask), the proposed approach has higher accuracy.