Fusion of multi-frequency SAR data with THAICHOTE optical imagery for maize classification in Thailand

Remote sensing data have been commonly used for agricultural crop monitoring. This paper was assessed the quality of using SAR and optical data fusion for maize classification. Two different SAR data sets from different sensors including dual polarization (HH and VV) X-band COSMO-SkyMed (CSK) and quad polarization (HH, HV, VH and VV) C-band RADARSAT-2 images were fused with THAICHOTE (namely, THEOS, an Earth observation mission of Thailand) optical data. This paper describes a comparative study of multi-sensor image fusion techniques in order to maintain spectral quality of the fused images. Principal Component Analysis (PCA), Intensity-Hue-Saturation (IHS), Brovey Transform (BT) and High-pass filter (HPF) techniques are implemented for image fusion. For the supervised classification, maximum likelihood was applied to the fused images to identify maize crop. Finally, the accuracy assessment was done by comparing maize maps generated from fused images and THAICHOTE classification. The PCA fused RADARSAT-2 with THAICHOTE images consistently provide excellent classification accuracies, well over 85%. The results obtained not only improving of the classification accuracy, but also can be identified the growing cycle of maize crop.