Human Posture Recognition System Based on Dynamic Time Warping Algorithm

In response to the limitations of vision-based human posture recognition in low-light or occluded environments, this article innovatively proposes a flexible pressure sensor array system to monitor the pressure distribution of synovial joints in real time and accurately capture postures. Through a weighted acquisition strategy, the accuracy and computational efficiency are optimized, and the dynamic time warping algorithm is improved to adapt to sensor data. Experiments have confirmed that under various wearing and assignment strategies, the average accuracy rate of the system exceeds 75%. Wearing on the inside is better than on the outside. It does not rely on image information and shows broad application potential.