The characteristic of landscape spectrum is the basic of application of remote sensing and plays an important role in quantitative analysis of remote sensing. However, in spectrum-based application of remote sensing, because the difference of measuring scale and instrument resolution yield serious error in spectral curve and reflectance for the same landscape, there exists difficulty in quantitative retrieval of special information extraction of remote sensing. Firstly, the imaging simulation principles of the optics image was described and proposed A method using field measured endmember spectrum with higher spectrum resolutions to simulate spectrum of Multi-spectrum images with lower spectrum resolution was proposed. In the present paper, the authors take the delta oasis of Weigan and Kuqa rivers ocated in the North of Tarim Basin as study area, and choose vegetation and soil as study object. At first, we accomplished the simulation from field measured endmember for multi-spectrum by using the spectral response function of AVNIR-2, and found the large correlation between simulated multi-spectrum and pixel spectrum of AVNIR-2 by using the statistical analyse. Finally, the authors set up the linear model to accomplish the quantitative transformation from edmember scale to pixel scale. The result of this study has the realistic meaning for the quantitative application of remote sensing.