In order to characterize the defects of PbS thin film photoconductive infrared detector materials, The physical model of 1/f noise and g-r noise are deduced and verified. The surface trap densities under different voltages are calculated by the relation between 1/f noise and surface trap in this model. The phenomenon that the surface trap density increases with bias voltage is observed. Therefore, the conclusion that 1/f noise is proportional to bias voltage is drawn, and it is consistent with the experimental measurements. In addition, the relation between g-r noise and deep-level defect characterization parameters is investigated based on this model, and the method of using low frequency noise to characterize defect parameters including defect activate energy, degeneracy factor and capture section is presented.