Study on the Accuracy of Land Cover Classification of ETM + Remote Sensing Image Fusion Method

张海霞,卞正富,陈秋成
DOI: https://doi.org/10.3969/j.issn.0517-6611.2008.01.015
2008-01-01
Abstract:The remote sensing image fusion methods of wavelet transform,combination of wavelet and PCA,and wavelet and HIS etc were used to fuse Landsat7 ETM+ multi-spectral and panchromatic images and study the land cover classification.And these methods were evaluated by combining the spectral static parameter of images with the classification accuracy of fused images.It was found in the results that all these 3 fusion methods could enhance the classification accuracy of images to different degrees,in which,the correlation coefficient between the fused images obtained by the 3rd method and original multi-spectral images was biggest,with the mean square deviation and information entropy being biggest,the information contained in images being most and spectrum characters being kept better.The fused images of the 3rd method had higher definition and superior capability on showing detail spatial information,so the classification accuracy of fused images was highest.Therefore,the fusion method of combining wavelet with HIS was more suitable for studying land cover classification of ETM+ fused images.
What problem does this paper attempt to address?