Comparison of PCA and IHS Based Methods for Fusing Hyperion and Cartosat-1 Images

To use the information of multi-source data, numerical or analytical fusion techniques are used in combination. The fusion makes for better interpretation and reduce the uncertainty and improve the classification. Several methods have been developed for fusion, including methods based on IHS, wavelet and PCA. In this study, the capabilities of fusion methods based on the IHS and PCA have been examined. In this regard, the methods FFT-PCA, Wavelet-PCA, FFT-IHS and Wavelet-IHS are used in order to fusion on Hyperion L1R data with spatial resolution of 30 meters and panchromatic image Cartosat-1 of IRS satellite with spatial resolution of 2.5 meters. To evaluate the results of the methods some criteria such as correlation coefficient, filtered correlation coefficient, error of least squares, ERGAS and SAM index were used. results of least square error criteria, respectively for, FFT-PCA, Wavelet-PCA, FFT-HIS and Wavelet-IHS for band 15 is 0/020, 0/022, 0/025 and 0/024 and for band 25 is 0/032, 0/035, 0/038 and 0/036 and for band 34 is 0/040, 0/043, 0/044 and 0/041.The results of this criteria and other criterias showed high accuracy of PCA-based methods to fusion of images. The PCA-based methods has advantages over methods based on IHS, such as no band limitation. Of all these methods, the FFT-PCA method provides more accurate results and preserve the spectral and spatial information better than the other methods.