With the rapid development of science and technology in the information age, video surveillance plays an important role in banks, hospitals, hotels and other public places, providing an important way for the application of public security investigation and judicial evidence collection. However, due to the influence of illumination and environment, the image quality is not ideal in some cases. Therefore, how to achieve video image enhancement under the existing conditions has been a hot research topic. Traditional image enhancement algorithms only deal with a single frame image, but the image itself contains less information and the processing effect is not ideal. In this paper, an improved super-resolution reconstruction algorithm based on POCS is proposed, which recon- structs multiple degraded YUV video images with low resolution into one frame with high resolution. Compared with the traditional gray transformation and histogram equalization algorithm, the effect of video image enhancement proposed in this paper is obvious- ly improved.