Abstract: In order to improve the target fuzzy recognition accuracy and computational performance of human motion images, a target fuzzy pattern recognition algorithm based on Fourier transform is proposed. According to the Hu-Fourier feature descriptor, the contour features of the target region of the human motion image can be accurately extracted. On the basis of the above, the sliding window detection is performed on the target area to match the model of different parts of the human body, and the obtained feedback information is modeled by the tree structure to realize the target fuzzy pattern recognition of the human motion image. The experimental results show that the proposed algorithm can improve the recognition speed of the proposed algorithm and improve the real-time monitoring requirements under the condition of high accuracy.