Abstract: The image of an airborne multispectral seanner may distort due to the large attitude variation with time and the effect of the atmospheric interference. To make full use of the image data for analysis and interpretation, the raw data must be carefully processed by a series of methods. The purpose of this paper is to suggest a complete series of preprocessing procedures. To evaluate the proposed overall methods, the data of Beijing and Xinjiang areaes obtained by an 8-channel airborne multispectral-scanner is used. The test result shows that both the image quality and accuracy are obviously improved.