Segmentation of oil palm area based on GLCM-SVM and NDVI

Abstract

This paper presents application of texture analysis using gray-level co-occurrence matrix (GLCM) for segmentation of oil palm area based on WorldView-2 imagery data. Different parameters of GLCM consisting of five distance spacing and three directions will be calculated, where eight texture features will be extracted. Based on land-use categories determined in WorldView-2 image, the features for oil palm and non-oil palm will be trained and classified using support vector machine (SVM). Segmentation based on 10×10, 20×20, 40×40 and 80×80 window will be determined by using the resulting output of SVM classification. Then, the normalized difference vegetation index (NDVI) of segmentation area will be calculated. Accuracy of oil palm area segmentation will be determined. The resulting segmentation of oil palm area shows a promising result that it can be used for intention of developing automatic oil palm tree counting.