基于慢特征分析的智能护理床人体动作识别方法研究  

Research on human motion recognition method of intelligent nursing bed based on slow feature analysis

作　　者：冯秀娟 郑文凯[2] 刘玉凡 贺旭东 衡春妮 FENG Xiujuan;ZHENG Wenkai;LIU Yufan;HE Xudong;HENG Chunni(Xi’an Jiaotong University City College,Xi’an 710018,China;Inner Mongolia Medical University,Hohhot Inner Mongolia,010110,China;The Second Affiliated of Xi’an Jiaotong University,Xi’an 710004,China;Shangnan County Hospital,Shangluo,Shaanxi 726300,China;Tangdu Hospital,Fourth Military Medical University,Xi’an 710038,China)

机构地区：[1]西安交通大学城市学院,西安710018 [2]内蒙古医科大学,呼和浩特010110 [3]西安交通大学第二附属医院,西安710004 [4]商南县医院,陕西商洛726300 [5]空军军医大学唐都医院,西安710038

出　　处：《自动化与仪器仪表》2023年第8期14-17,21,共5页Automation & Instrumentation

基　　金：陕西省教育科学“十四五”规划2021年度一般课题《网络式正念训练对大学生心理健康的影响研究》(SGH21Y0409)。

摘　　要：智能护理床是针对卧床病人的护理困难问题提出的一项对病人进行实时检测的自动化护理设备。已有的智能护理床在对卧床病人实行监测时,对压力信号的识别效果不佳,无法准确判断卧床病人的实际动作。据此,研究针对已有智能护理床在人体动作识别中存在的问题,提出利用慢特征分析方法对智能护理床的信号压力进行处理后再构建人体动作的识别模型。利用从医院采集的卧床病人动作对模型进行性能测试。结果显示,研究提出的基于慢特征分析的人体动作识别模型的平均准确率达到87.03%。因此,研究提出的基于慢特征分析的智能护理床人体动作识别模型具有相对理想的准确率和效率,可以有效判断卧床病人发出的不同动作,改善卧床病人的护理质量。Intelligent nursing bed is an automatic nursing equipment for real-time detection of patients,which is put forward to solve the nursing difficulties of bedridden patients.When the existing intelligent nursing bed monitors the bedridden patients,the recognition effect of the pressure signal is not good,and the actual movement of the bedridden patients cannot be accurately judged.According to this,aiming at the problems existing in the intelligent nursing bed in human motion recognition,this paper proposes to use the slow feature analysis method to process the signal pressure of the intelligent nursing bed and then constructs the recognition model of human motion.The performance of the model is tested by the motion of bedridden patients collected from the hospital.The results show that the average accuracy of the proposed human motion recognition model based on slow feature analysis is 87.03%.Therefore,the intelligent nursing bed human motion recognition model based on slow feature analysis proposed in the study has relatively ideal accuracy and efficiency,which can effectively judge the different movements of bedridden patients and improve the nursing quality of bedridden patients.

关 键 词：慢特征分析 智能护理床 人体动作识别 混合高斯模型 卧床病人 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]