Land Cover Classification Based Multiclass Support Vector Machine

Abstract

Remote sensing data are attractive for deriving land cover information through the image classification. Most of the practical application involves multiclass classification, especially in remote sensing land cover classification. Support Vector Machine (SVMs) are based on statically learning methods and originally designed for the binary classification. A number of methods have been purposed to implement SVMs to produce multiclass classification. The main aim of this paper is to implement the supervised classification method for multispectral remote sensing images by using multiclass SVMs. This paper emphasizes on Directed Acyclic Graph (DAG), multiclass classification method on SVM. This system is implemented by JAVA Netbeans5.5 language.