Compared with traditional infrared imaging, infrared polarization imaging system can detect and identify the man-made or camouflaged target more efficiently by using the difference in the degree of polarization (DoP) between the target and background. The scene’s radiation is attenuated by the path atmosphere firstly, and then modulated by the polarizer and the optical system. Because of the effect of the atmosphere (such as absorption, radiation, diffusion etc.), the final radiation intensity the sensor received changes, which affects the result of detection and identification. In this paper, the component characteristic of particles in atmosphere was discussed particularly. And the propagation of signal was described by analyzing the scattering effect between atmospheric particles and photons. After the process of free path sampling, selecting the radius of the colliding particles, the scattering angle and azimuth sampling, and particle collision and extinction judgment, a Monte Carlo model of polarized light propagation in atmosphere was present by use of the Stokes/Mueller formalism and Meridian planes method. Then two different methods (the radiation intensity and the DoP) used for target recognition in atmosphere were simulated. The relationship between the received radiation intensity, the DoP and the distance was developed. The contrast showed that the DoP had a better performance than the intensity measurements on the whole. However, there was a maximum distance for polarization imaging system using short wavelength to make the most of the advantage. When beyond this distance, the polarization imaging advantage will disappear. Polarized light with longer wavelengths had a better ability to maintain the state of polarization after propagation in the atmosphere.