How to fuse all kinds of images with various spatial resolution, to retain its spectral character, and to improve classification accuracy and spatial resolution is a hotspot in data fusion on remote sensing images. Based on wavelet transformation, the paper has extracted edge information of high spatial resolution image, and fused it with multi-spectral images. We chose IRS, SPOT and ETM images to do this experiment. The experimental results showed that it could improve spatial resolution of the multi-spectral images, while at the same time it also could retain spectral character very well with the method. Furthermore it is superior to other methods based on wavelet transformation in data fusion on remote sensing images with high spatial resolution difference.