The soil in midwest of Songnen Plain is becoming increasingly more salinized, which highlights the importance of rapid and precise monitoring and evaluation on salinization of soil. With great advantages, the microwave remote sensing becomes an emerging method with huge potential in detecting composition of soil. With the Sentinel-1 image covering the region with salinized soil in Songnen Plain as data source, combined with the assay data of total salt content in the sampled soil from the region of interest, the technology and method of soil salinization information extraction are investigated based on dual polarization radar image. Firstly, 64 soil samples are collected in the study area, and the total salt content of soil samples is tested in the laboratory. Fifty-two soil samples are taken as modeling samples, and 12 samples are taken as test samples. Saline-alkaline soil is divided into light salinization soil (with salt content of 1-3 g/kg), medium salinization soil (with salt content of 3-5 g/kg), heavy salinization soil (with salt content of 5-7 g/kg) and saline-alkaline soil (with salt content > 7 g/kg). The Speckle Filtering of S1TBX software is used to filter Sentinel-1 image to eliminate the influence of noise in the image on information extraction. Radiometric calibration is made for image using Radiometric-Calibrate tool to eliminate the absorption and scattering of atmospheric aerosol for imaging process so as to obtain the true back scattering coefficient of topographical surface feature. The image is subjected to geometrical correction by Terrain Correction tool. Then, by analyzing the quantitative relations between radar image VH and VV polarization modes, back scattering coefficient of polarization combination of VV+VH, VV/VH, (VV+VH)/(VV-VH), and (VV2+VH2)/(VV2-VH2), and soil salt content, the optimized polarization combination mode of the inversion model is determined. Lastly, the prediction model for soil salinity in the region of interest is established using multiple regression technique, and the relative error and root mean square error (RMSE) between predicted value and actual value of salinity in test sample of soil are compared to evaluate the precision of inversion model. The inversion model of soil salinity is used to inverse topsoil salinity in region of interest and the inversion result chart of soil salinity is drawn for the region of interest. The findings are: The back scattering coefficient of Sentinel-1 image VH polarization mode is strongly responsive to medium salinization soil, heavy salinization soil and saline-alkaline soil, and the back scattering coefficient of VV polarization mode is strongly responsive to all degrees of saline-alkaline soil; the back scattering coefficient of polarization mode of (VV2+VH2)/(VV2-VH2) can well separate non-salinized soil, light salinization soil, medium salinization soil, heavy salinization soil and saline-alkaline soil. The coefficient of determination R2 for the established model reaches 0.872, and the RMSE is 0.988. Model checking results show that, the maximal relative error between predicted value and actual value of sample’s salinity is 4.87%, and the minimum relative error is only 0.91%. In the scatter plot, inverse value and measured value of sample salt content after checking are evenly distributed on both sides of 1:1 straight line, and coefficient of determination R2 is up to 0.98, and RMSE is 0.412, showing that the inversion model has a high precision in prediction of soil salt content in the research area. The graphical result of the inversion shows that: light salinization soil is widely distributed in the research area; medium salinization soil is mainly distributed in the western and southern Daqing City and western and northern Datong District, and concentrates around the rivers and lakes in the research area; heavy salinization soil, saline-alkaline soil and medium salinization soil are incidentally distributed, and mainly distributed, and mainly distributed in western and southern Datong District. This method can meet the need for monitoring soil salinization in large region, and provide reference for research on extraction of composition of soil based on Sentinel-1 radar data. [ABSTRACT FROM AUTHOR]