Extraction and Network Sharing of Forest Vegetation Information based on SVM

Abstract

The support vector machine (SVM) is a new method of data mining, which can deal with regression problems (time series analysis), pattern recognition (classification, discriminant analysis) and many other issues very well. In recent years, SVM has been widely used in computer classification and recognition of remote sensing images. This paper is based on Landsat TM image data, using a classification method which is based on support vector machine to extract the forest cover information of Dahuanggou tree farm of Changbai Mountain area, and compare with the conventional maximum likelihood classification. The results show that extraction accuracy of forest information based on support vector machine, Kappa values are 0.9810, 0.9716, 0.9753, which are exceeding the extraction accuracy of maximum likelihood method (MLC) and Kappa value of 0.9634, the method has good maneuverability and practicality