Abstract: In the present study, Pingluo of Ningxia Province in China was taken as the study area, and spectral data obtained by Unispec-SC, the value of soil salt content measured by experiment were taken as the basic data. Hyper-spectral data processing method was used to analyze spectral characteristics of different levels of salinization area vegetation. Spectral data were transformed in 16 different approaches, including logarithm, root mean squares, and first order differentiation. Correlation analysis was carried out between the obtained spectra and soil salinity. The most sensitive bands was selected, soil index and vegetation index were built. Nonlinear regression was employed to establish soil salinization remote sensing monitoring model. The results show that by comparing various spectral transformations, the first order differential of soil spectral was the most sensitive to soil salinization degrees. The model was based on the spectral index, including SI and MSAVI, and it could monitor soil salinization accurately. The correlation between simulated values and measured values was 0.758 9. The soil salinization could be achieved rapidly in the area.