Remote sensing image fusion can be use to highlight the thematic information, to eliminate or suppress irrelevant information, to improve the quality of image for target recognition, thereby increasing the reliability of interpretation and reduce ambiguity and improve the classification, to expand its application and effectiveness. Landsat-7 ETM+ image is applied aboard in various departments related to earth observation. It provides the 15m high special pan and 30m and 10m multi-spectral data, which can be fused to achieve both high spatial and spectral resolution in a single image. Seven pixel-level fusion algorithms of remote sensing images, Brovey Transform, Intensity-Hue-Saturation Transform(HIS), Principal Components Transform (PCA), Multiplication Transform (MLT), Smoothing Filer-Based Intensity Modulation Transform (SFIM), High-Pass Filer Transform (HPF), Wavelet-HIS Transform, have been used to fusion Landsat-7 ETM+ multi-spectral image and panchromatic image. And the seven fused images have been analyzed and evaluated by the amount of spectral information (mainly deviation index, spectral distortion and the correlation coefficient) and spatial information (mean, standard deviation and the information entropy) maintained in fused images. The results shown that when SFIM fusion model was used, both spectral distortion degree of fused images (8.9182) and deviation index (0.0004) were much smaller than that of the others, which indicated that maintenance of the spectral characteristics of multi-spectral images was good. And the correlation coefficient is the second high (0.6376), indicating that there is big similarity between the fused images and original multi-spectral images. The standard deviation (19.0427) and information entropy (6.1297) of the fused images is big which indicated there is a larger amount of information in them. Overall, SFIM fusion model is simple and easy to implement, with low time-consuming, small distortion and rich spatial informati...