Millimeter-wave radiative sensing has been used in several close-range applications such as security checks, military detection, terrain modeling, and so on. Obtaining target information through radiometry is the goal of above applications. However, there was no method to recovery stable material information and complete geometric information through radiometry. The common problem of physically mode1-based information recovery is the solution of underdetermined equations. In this article, we analyse the multi-polarization brightness temperature model and propose the equivalent permittivity to characterize the material information and to simplify the reflectivity model. With the simplified reflectivity model, we solve the underdetermined problem in information inversion and realize the material classification and surface normal vector reconstruction. Simulations or experiment are conducted to analyse the error and suitable range of the simplified reflectivity model, and to varify the validity and accuracy of our methods of material classification and normal vector reconstruction. The results show that, our classification method is suitable for most objects at incident angle of 20 • ∼ 60 • , and our SNV estimation method is effective at all incident angle. The possible applications include liquid composition analysis, road perception, and three-dimensional reconstruction.