The existing polarization imaging systems are hard to meet the requirements of target detection and tracking in real-time image processing field. A three-channel polarization imaging system was put forward to solve above problems, and correction methods were proposed to solve the field differences and non-uniformity errors of the three-channel polarization imaging system. This system used three separate channels to acquire the polarization images from the direction of 0°, 60° and 120°. SIFT algorithm was used to solve the field differences among three independent channels. According to depolarization characteristics of the integrating sphere and linear response model of the detector, integrating sphere was used to produce unpolarized light to correcte the non-uniformity errors. Experimental results show that: 1) the imaging effect of the corrected three-channel polarization imaging system can be close to the level of single-polarization imaging system′s; 2) the system can capture images real-time meanwhile overcome the quality-descent problem of the polarization images caused by multi-channel, improving the practical applicability.