Fusion of high-resolution remote sensing images based on a trous wavelet algorithm

There are some problems not to be resolved very well, when we merge images by the wavelet transform method. One problem is how many levels the original images should be decomposed to, in order to make fusion images obtain more information. Other problem is how to reduce the spectral distortions of the fusion images more efficiently, when we enhance the spatial resolution of the low-resolution images. In this paper we rightly select the number of wavelet decomposition levels by computing entropy of the fusion images. When performing wavelet reconstruction, we introduce the local correlation coefficient and set up the different thresholds at different levels of wavelet decomposition, in order to reduce the spectral distortions of the fusion images. In our experiment we merge three multi-spectral images with a panchromatic image of Quickbird data by our method. The results demonstrate that our method is a good fusion method to increase information and reduce spectral distortions.