Abstract | Full Paper | PDF | Printer friendly format

Abstract :
There are different cases that high spectral and spatial resolution are needed both together, but the instruments presented now can't produce such data because of several limitations. One solution to get such data is data fusion .There are some methods to merge panchromatic (high spatial resolution) and multispectral data (high spectral resolution). Some of more common methods are based on getting Principle Component (PC), Brovey method, IHS (Intensity-Hue-Saturation) transform ,Hexcone ,….These methods change the spectral characteristics very much ,so aren't appropriate to merge the images. Here we use a method based on wavelet multiresolution decomposition to merge the panchromatic band of ETM sensor (15m) and multispectral bands of the same sensor (30m).

The correlations between the merged images and the original ones show that the proposed method can preserve the spectral characteristics more than the others. The computed correlations in this method is more than 0.8 while they are about 0.6 or less than it in the other methods.

Using wavelet_based image fusion leads to that spectral postprocessing such as classification is done more precisely. So this method is perfered.