Vegetations in desert play key a role in protecting eco-environment, especially in desert-oasis crisscross zone. Vegetations are of great significance in soil conservation and improving the shear resistance of land. Therefore, they can help to prevent soil from wind erosion and keep land from desertification. Analyzing the spectral data of typical vegetation in desert-oasis crisscross zone by using the hyper spectral technology can be a guidance for remote sensing vegetation classification and serve as the basis for remote vegetation monitoring. In our research, four kinds of typical vegetations have been selected: cotton, tamarix chinensis, Haloxylon ammodendron and Halostachys caspica. The researcher collected series of spectral data of different typical vegetations under different conditions with the help of Field Spec 4 high resolution spectrometer. The collected data was classified, flitted and synthesized with two kinds of transform methods: FDR (First Order Derivative Reflectance) and RLR (Reciprocal Logarithmic Reflectance) transformation. And then three kinds of data were used in further research to analyze the sensitive spectrum band and expression of different vegetation. The result shows that the spectral curve of different vegetations show the same changing trend; different vegetation show different expression in “red edge” with the near infrared band of 780~1 260 nm. The visible light absorption of vegetation is very strong, and the difference of absorption extent causes peaks and troughs. The “red edge” characteristic is unique, which will carry the proper information of certain vegetation, and the result of different kinds of transformation show that FDR can express the red edge characteristic much better than other ways. At last, three ways were used to calculate NDVI, using the original spectral data, transforming spectral data with FDR and transforming spectral data with RLR. Result shows that the NDVI , which calculated by RLR can help to distinguish the type of vegetation with higher accuracy is.