By taking the 85 black soil samples in north of Jilin province as the examples of analyzing organic matter,a optimum choosing method of combination wavelengths for calibration was presented by the ways of Multiplicative Scatter Correction(MSC) and correlation spectrum,and the model was optimized progressively based on experience and reference wavelength theory.

In this paper,according to hyperspectral RS image data characteristics,it is carried on to select feature by spectral relevance,and then the SVM is used for analysis and interpretation of hyperspectral RS image,finally,test is done on the AVIRIS image and the results show that classification accuracy is better than conventional methods.