Target detection in smoke environment using laser active imaging is of great research significance. According to the characteristics of light propagating in smoke environment, the method of laser transmission imaging is adopted and polarization-difference (PD) imaging technology is used to suppress the effect of scattering light on imaging quality. PD imaging method needs to record two target images of different polarization directions and then the two images are subtracted. A correction factor α is used to match the scattering light intensity of the two images, so scattering light can be mostly filtered out. Experiments show that the quality of images obtained by PD imaging method is better than intensity ones. The PD images have big noisy when smoke concentration is high. A method which combines median filtering with multi-scale morphological edge detection is proposed to process images. It can suppress image noise, improve image contrast and edge sharpness effectively.