For the sake of retrieving soil moisture by the polarized information, a polarized reflection model of soil surface was established, and an experimental study of soil moisture and reflected light′s polarized information was carried out in the visible wavebands. Firstly, the polarized reflection model of soil surface was established based on the geometrical optics theory and the measured data. Secondly, the soil samples of different moisture contents were made, the experimental platform was set up, the soil samples polarized information were obtained by a polarized imaging detection system with high precision in a variety of incident and observation conditions, the model was verified by analyzing the experimental data. Finally, a quantitative relationship among the micro-surface slope variance, the diffuse reflection coefficient, the equivalent refractive index and the soil moisture was established in the model and the soil moisture was retrieved according to the quantitative relationship. The research results show that: on 476 nm, the polarized information of a soil sample with a moisture content of 26% is measured by a polarized imaging detection system in the indoor environment, the soil moisture is calculated as 24.73% according to the proposed retrieval method in an error of 4.88%. Therefore, it is proved that the proposed method of retrieving soil moisture by the polarized information is correct and feasible.