Polarization imaging technology is a method that acquires the object images by collecting the polarization information of the target radiation or reflected signals. In particular, compared with the light intensity detection, it has unique advantages in the artificial target detection and surface recognition. Due to the short range and low quality of the conventional polarization imaging in complex imaging environment, a new kind of polarization imaging technology based on compressed sensing was proposed. The basic principle of compressed sensing theory was elaborated. By constructing reasonable sampling matrix and reconstruction algorithm, the specific imaging system was designed. Besides, the feasibility of this technology was confirmed through the imaging experiment. The study results in the air show the system can reconstruct the polarization images of the pre-positioned target. Additionally, in the existing experimental conditions, some measures are investigated and proposed to improve the system imaging performance.