Fusion of RADARSAT-2 imagery with LANDSAT-8 multispectral data for improving land cover classification performance using SVM

Study of the land cover classification using multi-source data are very important for eco-environment monitoring, land use planning and climatic change detection. In this study, the utility of multi-source RADARSAT-2 and LANDSAT-8 multi-spectral images for improving land cover classification performance using Support Vector Machine (SVM) classifier. HH polarized C band RADARSAT-2 images were fused with the three band (6, 5, and 4) LANDSAT-8 multispectral image for land cover classification. Wavelet-based fusion (WT) techniques are implemented in the data fusion process. The Radial Basic Function (RBF) kernel function were used for SVM classifier in order to classify land cover types in the study area. The results of the SVM classification were compared with those using standard method Maximum Likelihood (ML) classifier, and it demonstrates a higher accuracy. Finally, it was indicated by the study that the fusion of SAR and optical images can significantly improve the classification accuracy with respect to use single dataset, and the SVM classifier could clearly outperform the standard method the ML classifier.