Landslide recognition in mountain image based on support vector machine

Abstract

To improve the recognition of landslides, an algorithm based on combined features and support vector machine (SVM) is proposed. The landslide image was preprocessed firstly, including size equalization and histogram equalization. Then feature extractions were done as follows: dividing the image into sub-regions vertically, extracting texture features based on gray level co-occurrence matrix (GLCM) in each sub-region, extracting segmentation feature based on RGB color space, extracting color features based on HIS color space in each sub-region, and extracting gradient features in gradient image. Based on SVM, the above extracted features were used to realize the classification as well as the disaster recognition. Experiments show that this algorithm has better recognition effect on the mountain images than the former algorithm which we have proposed before. © 2012 Springer Science+Business Media, LLC.