In some complex scenes, infrared targets are difficult to be detected and recognized due to the interference of clusters and noise. A moving infrared target enhancement algorithm based on optical flow is proposed. In the algorithm, a moving target is enhanced by using the velocity field difference between the target and the background. The algorithm is optimized on the basis of the parallel processing mode of a Graphic Processor Unit (GPU). It can be implemented on line in real time. Compared with the frame difference method and background subtraction method commonly used in moving target detection, this algorithm is more robust. Because the moving target in the actual infrared video is enhanced, better real-time target enhancement effectiveness of the algorithm is revealed.