With the increasing development of computer vision technology in the application field of image processing and pattern recognition, the detection of moving human targets has become a hot focus attracting more attention gradually. An improved moving human target detection algorithm is proposed. The background is extracted in an image by a median method. Then, the moving target region is obtained by combining the background subtraction with the frame difference in the processing of the image. The experimental results show that the algorithm can overcome the defect brought by a single algorithm. It has a higher accuracy and meets the real-time requirements in engineering application. Since the algorithm is simple and effective, it is suitable for embedded platforms.