The scene of human motion posture recognition is relatively complex, making it difficult to extract feature information with good separability between multiple categories, resulting in a decrease in the accuracy of human motion posture recognition. Therefore, a human motion posture recognition method based on the EMD-PSO-LSTM combination model is proposed. Using EMD algorithm to extract subtle features of human motion posture in different time dimensions, ensuring that the extracted features have multi class separability. Build an LSTM model, introduce PSO algorithm to optimize and adjust the parameters in the LSTM model, and combine the optimized model to build an EMD-PSO-LSTM combination model. Input the extracted features into the model, and the output of the model is the human motion posture recognition result. After experimental verification, the algorithm has demonstrated excellent performance in joint recognition and overall posture recognition. At the same time, it can still maintain high recognition accuracy when facing large-scale human posture data samples, and can be widely applied in practice.
ZhuLi-ping, TangLiang, ZhuKai-jie, et al. Multi-target human pose estimation in sports scenes[J]. Computer Engineering and Design, 2023, 44(7): 2156-2162.
FuHui-chen, GaoJun-wei, CheLu-yang. Human posture estimation and movement recognition in fitness behavior[J]. Chinese Journal of Liquid Crystals and Displays, 2024, 39(2): 217-227.
DengPing, WuMing-hui. Human motion attitude recognition method based on machine learning[J]. Journal of Chinese Inertial Technology, 2022, 30(1):37-43.
HengLyu, YangHong-yu. A 3D human pose estimation approach based on spatio-temporal motion interaction modeling[J]. Journal of Graphics, 2024, 45(1): 159-168.
YangGang, WangChao, WangLi-fang, et al. Human motion monitoring based on TENG and its application in fall detection[J]. Micronanoelectronic Technology, 2023, 60(9): 1464-1472.
KongFan-miao, GaoLu, LiPeng-cheng, et al. EMD-LSTM-LB frequency-divided time series prediction algorithm[J]. Computer Engineering and Design, 2023, 44(10): 3021-3030.
LiXia, LiShou-wei. Non-linear time series prediction model based on EMD and DVG and its application[J]. Chinese Journal of Management Science, 2022, 30(9): 275-286.
SunYan-xi, ChenJi-bin, WuDong-hui. Human activity recognition method based on convolutional neural network-BiLSTM[J]. Science Technology and Engineering, 2022, 22(4): 1517-1525.
LiXin-chun, ZengShi-hao. UWB radar human action recognition based on improved particle swarm optimization[J]. Journal of Chongqing University of Posts and Telecommunications (Natural Science Edition), 2024, 36(2): 268-276.