一种基于sEMG信号的手势识别方法研究
The Study of Hand Gesture Recognition Method Based on sEMG Signal

随着机器人技术的发展,利用表面肌电(surface electromyography,sEMG)信号进行动作识别成为研究的热点.针对sEMG与手部动作关系复杂且实际应用困难的问题,该文提出一种基于BP(back propagation)神经网络的模式识别系统,可通过指浅屈肌和肱挠肌的2路sEMG信息源,识别手部6种不同姿态.该研究采用1阶数字低通无限脉冲响应滤波器提取信号包络,并利用能量特征值进行端点检测,选取短时能量、过零率和12阶线性预测系数进行模式识别.实验结果表明:该方法可以达到90%以上的识别正确率,具有一定的实际应用前景.
The research of gesture recognition based on sEMG is becoming a hot spot in recent years with the development of robotics.In view of the complex relationship between sEMG and hand gestures and the difficulty of practical application,a pattern recognition system based on BP(back propagation)neural network is proposed which can recognize six hand gestures by the sEMG of superficial digital muscle and flex muscle.The signal envelope is extract by the first-order infinite impulse response digital low-pass filter.And the energy eigenvalues are chosen to do endpoint detection,short-time energy and zero crossing rates and 12 level linear prediction coefficient are adopted to do pattern recognition.Finally,a pattern recognition experiment has been done which recearchs the relationship between the sEMG and six different hand gestures and the accuracy is above 90%.The result shows that the method proposed in this study can achieve a high recognition rate and has a practical application prospect