Comparative study of classification algorithms with modified multivariate local binary pattern texture model on remotely sensed images

Texture analysis plays a vital role in remotely sensed image classification as every pixel is going to be classified based on the collective pixel values of neighborhood. The result thus obtained gives increased classification accuracy. In this paper, a modified texture model obtained by modifying Multivariate Local Binary Pattern (MLBP) texture model is used for classification in remotely sensed images together with Self organizing map, Support vector machine and Fuzzy KNN. The results are evaluated based on classification accuracy. After the study, it was found that support vector machine outperformed other classification algorithms in getting high classification accuracy.