Human Gait Analysis Method Based on Sample Entropy Fusion AlphaPose Algorithm
Xinyu Lv, Shengying Wang, Tao Chen, Jing Zhao, Desheng Chen, Mingxia Xiao, Xiaoye Zhao, Haicheng Wei · 2021
Aiming at the problem of difficulty to evaluate the recovery of patients after their joint replacement accurately, a new gait analysis method was proposed on the basis of Sample Entropy fusion AlphaPose. In the algorithm, AlphaPose was encouraged to extract the trajectory of the key points from healthy people and patients, and convert the transform trajectories into feature vectors. After normalized, feature vectors were calculated and analyzed by Sample Entropy algorithm. The results demonstrated that there was significant difference in the entropy value of heel key point waveform between two groups, the entropy value of patients with joint disease was lower than that of healthy group. Comparing with the traditional gait analysis method, the proposed algorithm performs better on human pose recognition, because of its strong robustness and efficiency.