Novel Landslide Extraction Method of SAR Image Based on Change Detection and Polarization Decomposition

This paper proposes a landslide detection method combining gray level co-occurrence matrix (GLCM) and polarization decomposition technique. The proposed method makes full use of the characteristics of the SAR image and combines the polarization information with the texture feature. Landslide candidate areas were identified using GLCM and final landslide areas were determined using polarization decomposition techniques. The data used for evaluation is taken from Guangyuan City, Sichuan, China. The proposed method successfully detects the changes of two SLC sentinel-1A C-band VV and VH dual polarization SAR images before and after the landslide, which proves the effectiveness of this method.