摘要
表面肌电(surface electromyography,s EMG)信号被广泛应用于临床诊断、康复工程和人机交互等领域中.针对目前控制肌电假肢手的电极成本高、电极佩戴困难以及操作灵活性差等问题,设计一种基于MYO的肌电假肢手手势在线识别系统.通过采集人体上肢前臂的表面肌电信号,在时域上分别提取5种特征值,利用反向传播(back propagation,BP)神经网络分类算法实现对8种手势动作意图的在线实时识别.实验结果证明,利用MYO进行手势识别可以获得较好的识别结果,该系统能够准确识别8种手部动作,平均在线识别率达到92%.
Surface electromyography(s EMG) is widely used in clinical diagnosis,rehabilitation engineering and human-computer interaction,etc. Aming at the problems of high cost of electrodes to control electromyography prosthetic hands,the difficulty in electrodes wear and poor operation flexibility,a MYO-based hand mode online identification system of electromyography prosthetic hands is designed. By collecting the s EMG of the human upper-limb-forearm and extracting 5 characteristic values in the time domain,8 real-time gesture recognition strategies are realized through the back propagation neural network. Experimental results show that the MYO-based gesture recognition canproduce bettergesture recognition results. The system can accurately identify the eight kinds of hand movements,and the average online recognition rate reaches 92%.
出处
《上海师范大学学报(自然科学版)》
2018年第1期43-48,共6页
Journal of Shanghai Normal University(Natural Sciences)
关键词
表面肌电信号
MYO
特征提取
BP神经网络
在线识别
surface electromyography
MYO
feature extraction
BP neural network
online recognition