Abstract: According to the redundant and complementary information among polarization characteristic images, a second polarization fusion method was proposed based on gradient characteristic and support value transformation. Firstly, the images of linear degree of infrared polarization and polarization angle were fused using gradient characteristic fusion. Then the support value transform was used to fuse the images of primary polarization fusion and total intensity second，and a fusion image with enhanced edge ,clear contour and high contrast was obtained. Experiment results show that compared with the polarization image, polarization linear image and total intensity image, the fused image increases the local standard deviation by 23.02%, 176.9% and 148.2% respectively, increases the contrast between the target and the background by 67.84%, 196.5%, 49.39% respectively and the average gradient by 46.09%, 164.1%, 214.5% respectively .The validity of the algorithm proposed is proved.