Due to the heterogeneous characteristics of the infrared polarization focal plane， it is necessary to take into consideration the impact of the response differences in different bias channels on the overall correction effect in the process of non-uniform correction. The non-uniform correction problem of the infrared polarization focal plane is more complex than that of the ordinary homogeneous infrared focal plane. For non-uniformity correction of polarization focal plane， an algorithm based on scene-based polarization redundancy estimation is proposed. The statistical information between acquired images and estimated images from polarization redundancy is acquired， then the differences in the response of all pixels in the whole focal plane are obtained. By comparing and analyzing the differences from two directions in each polarization channel， we can obtain the updated gain correction coefficients. Then radiometric recalibration is used to suppress the ghost image caused by the static scene， and the gain correction coefficients in the current state are obtained. In this process， the gain correction coefficients are updated adaptively by evaluating the previous correction coefficients through polarization redundancy estimation. Finally， experiments on the real scene data demonstrate the effectiveness of the proposed non-uniformity correction algorithm for polarization images.