A method aimed at automatic landslide extraction based on background values of satellite imagery

作者:Yang Shuwen*; Li Yikun; Feng Guangsheng; Zhang Lifeng
来源:International Journal of Remote Sensing, 2014, 35(6): 2247-2266.
DOI:10.1080/01431161.2014.890760

摘要

The vegetation and soil layers of land surfaces become damaged during landslides. When viewed in remote-sensing images, these post-landslide vegetation and soil layers appear significantly different from the surrounding background. This article proposes an integrated method for calculating remote-sensing indices to extract background information of areas that have undergone landslides by increasing the spectral difference between bare land and other surface features, such as vegetation, areas of shadow, settlements, and waterbodies. Based on the background values of satellite images, the indices include vegetation index, soil brightness index, the first component (PC1) of the principal component transformation of the image, and also terrain data. An automatic extraction model of landslides is proposed based on an improved multi-peak histogram threshold algorithm and mathematical morphology filtering algorithm. Landslide information is extracted and verified using an experimental system built by the Environment for Visualizing Images (ENVI) and Interface Description Language (IDL). The experiment includes manually and automatically recognized landslides. The experimental results show that the proposed method not only effectively removes non-landslide and other interference information, but also significantly improves the efficiency and accuracy of the automated landslide extraction process.

全文