The Study Of Landslide Extraction Method Based On SAR Images

A landslide takes place frequently and is one of the most widely disaster among various types of the geological hazardand. How to quickly get landslide information is critical to timely rescue.With the continuous development of synthetic aperture radar technology, information extraction about landslide hazard using SAR image is concerned and achieved results.This paper analyzes the characteristics of the landslide in the SAR image. SAR images contain some deformation, so SAR images are needed to radiometric and geometric correction. SAR image is a reflection of radar scattering properties about surface features. So there is speckle noise. Image filtering technology can reduce the effects of speckle noise. In addition, the two SAR images before and after landslide disaster are needed to registration.According to the characteristics of SAR images have clutter statistical distribution, Edgeworth approximation method is used density function SAR data.Kullback-Leibler divergence(KLD) is measure of the difference between the SAR data. This method is feasibility to prove by SAR images. The change information is gotten, and the result is analyzed by optical images.In this paper, the gray and texture characteristics of SAR images after disaster and expectation maximization algorithm are used to obtain landslide information.There is great changes in SAR images after landslide disaster. Eight gray and texture features images are obtained by the gray co-occurrence matrix of SAR images. As for the gray texture features three kinds of landslide information extracted features are selected according to Bhattacharyya distance.Using the optimal threshold calculated by Expectation Maximization algorithm divides the composite image into landslide area and non-landslide area. So the image contained suspected landslide information obtained. There are some similarities between certain terrain and landslide in SAR images. Finally, it combines the suspected landslide information and change information to get the final result of landslide information. In this paper, using this method extracts landslide information by Radarsat-2 SAR images in Yingxiu country, Sichuan Provinve after the earthquake.The results are analyzed by optical images. It proves that this method can accurately extract landslide information. The method also applies to other areas images to extract landslide information.