Polarization ghost imaging system combines the strength and polarization detection, and expands the obtained information from the ghost imaging system. It can carry out effective target detection and identification. The conventional correlation polarization ghost imaging system requires a large amount of sampling data, and signal-to-noise ratio of the restored result is low. Polarization ghost imaging system based on compressed sensing was proposed to obtain the intensity and polarization information of the object using the detected information from the point detector. In the simulation experiment, the same reflectivity but different polarization characteristics of objects were employed. The results show that the high quality of the object intensity and polarization information can be obtained with fewer sample number compared with the correlation method, and it can improve the real-time performance. Finally, using image fusion algorithm to merge the intensity and polarization information, recognition by fusing information can be effectively carried out for a variety of objects.