A Method for Extracting Texture Features of Landslide in High Resolution Remote Sensing Images

Abstract

High resolution remote sensing images play an important role in landslide recognition. In order to solve the problems of diverse data, complex algorithms, redundant features, and insufficient research on the specific texture characteristics of a single landslide, Gabor filtering was used for texture feature extraction and fusion of spectral features using Support Vector Machine to achieve landslide extraction. First, multi-scale and multi-direction Gabor transformation is carried out on the landslide image. Then, the Gabor features of different scales in the same direction are fused to obtain the multi-directional features. Finally, the multi-directional Gabor features are filtered at the maximum to form the multi-scale and multi-directional Gabor texture features. The results show that, compared with the local binary model and local spatial autocorrelation, the texture analysis method in this paper has the best effect on landslide extraction and a high extraction accuracy.

Keywords

• high-resolution remote sensing image
• multi-scale and multi-direction gabor texture features
• landslide extraction