KLASIFIKASI SPASIAL PENUTUP LAHAN DENGAN DATA SAR DUAL-POLARISASI MENGGUNAKAN NORMALIZED DIFFERENCE POLARIZATION INDEX DAN FITUR KERUANGAN DARI MATRIK KOOKURENSI

Abstract

In this study, the land cover classification method using the spatial information features of co-occurrence matrix and Normalized Difference Polarization Index (NDPI) data from dual polarization SAR Data was proposed. The spatial information features are used as input of supervised classification, and to get the performance of the proposed method, land cover classification was conducted with SAR C-band and L-band satellite data of Envisat ASAR and ALOS PALSAR. The results of the study are, the size of window on the SAR image to get the spatial information features of co-occurrence matrix and the use of additional NDPI data are giving effect to the accuracy of classification results. At the test area in Siak Riau Province which have 7 classes of land use, the optimum window size for co-occurrence matrix is 7 pixel x 7 pixel for ASAR data which has 75m spatial resolution, and more than 9 pixel x 9 pixel for PALSAR data which has 10m spatial resolution. The addition of the co-occurrence matrix information of NDPI data can improve the classification of accuracy up to 2%.