Abstract: Polarization imaging has unique advantages in target detection, recognition and processing compared with traditional photoelectric detection technology. To overcome degraged images taken in haze weather, an image defogging method based on polarization information is proposed. By obtaining target polarized image at three angles, Stokes vector of scene object is solved. From relationship between Stokes vector and Mueller matrix, variation law of polarized image intensity with polarization angle is analyzed, orthogonal polarization images under maximum and minimum light intensity are obtained automatically and accurately. Polarization degree of atmosphere and its infinity atmospheric light intensity value are estimated using polarization filtering and bright channel priori method, and fog-free images are reconstructed. Experimental results show that clear images can be reconstructed in haze weather by using orthogonal polarized images obtained, average gradient and edge intensity of reconstructed image are improved by about 3 times and grey standard deviation is improved by about 88%.