Abstract: Infrared polarization imaging measurement can be used to obtain the polarization state in target radioactive and reflective detection. With this method, the problems that traditional photometry cannot solve can be settled, and high-precision results were obtained. Particularly, in the field of military detection, polarization measurement can be introduced to distinguish artificial target quickly from the natural background, and performed well in target identification. The intensity, degree and angle of polarization that used in polarization measurement can reflect different physical properties, and along with them, redundancy and complementarity are revealed seriously. In allusion to the character, a target detection method for infrared polarization image was proposed: firstly, Mean-Shift algorithm was employed to aggregate the infrared and polarized image. Secondly, the object information of clustered infrared and polarized image according to DC evidence theory were combined. In this case, the target can be distinguished from the background, so as to achieve the purpose of target detection. Finally, comparison of the consequence of experimental image and wavelet fusion image to demonstrate the superiority of the presented algorithm.