Bare land extraction method of rain-induced shallow landslides based on SPOT5 images

Abstract

For the image characteristics of more serious damage to vegetation, geological characteristics and mechanism of remote sensing, the paper put forward some remote sensing indices based on background value of remote sensing images, such as vegetation index, soil brightness index, PC1 of principal components transformation and terrain data, which could reflect background information of landslides and collapses. Meanwhile, integrated calculation method of remote sensing index was proposed, which could effectively increase spectral differences of bare land with vegetation, terrain shadows, rivers and other surface features. Gray value of bare land was the lowest in the image so as to obtain segmenting value to accurately extract polygons of bare land based on automatic selection algorithm of threshold.