Automatic Snow Cover Detection from Single Cloud-interrupted Remote Sensing Image by a Two-stage Approach

Abstract

Snow cover detection (SCD) from single cloud-interrupted remote sensing image is an important task in many applications like avalanche forecast, real-time hazard monitoring and loss assessment. Aimed at development of an automatic procedure for SCD, a two-stage approach is proposed in this study using single remote sensing image. Due to the interruption of cloud in the detection process, the two-stage approach includes crude snow detection and accurate snow detection. In the first stage, weighted fuzzy C-means (WFCM) is introduced for crude snow detection to extract snow and cloud. In the second stage, gradient optimization and morphological closed operation (GOMCO) are combined for accurate snow detection to remove the cloud. By iteratively using GOMCO, the result of snow is obtained. The experiment shows that the interference of the others can be effectively removed and the snow area is obtained accurately by using the proposed method.