A Pattern Recognition Algorithm For The Electromyographic Signals Of The Elbow Arm Surface Of Four Finger Movements

At the developing of the science and technology,There have many applications of the finger recongnition,It has occupied an important position in the human hand motion.The rising trend of finger movements is developing in human-machine interaction,the recognition of figer movements can also help the disabled and the elderly to realize human-machine interaction,and by using the electrostatic signal analysis of neurons to the patient's diagnosis,rehabilitation plan.It is going to becoming the Special research in the world.Gesture recognition has wide prospect in life,but the research of the figer movement is very less,figer movement is the base of the gesture.We usually used EMG signals base on the electrodes to recognize the gesture.This article puts forward using elbow sEMG(Surface-Electromyography,a kind of EMG signal)to the thumb click with the index finger,thumb click with the middle figner,thumb click with the ring finger,and click with the little finger,four finger movements recognition is studied.This experiment is based on the sEMG singles analysis,using the MYO to collect four fingers movements,replace the traditional way of electrodes deep in the muscle collection.EMGlab activity period of MYO acquisition of data for processing,to the high frequency of 1000Hz filtering of the signal,reapet peak.And then to MLT decomposition of signal,get each channel template waveform,will display in the templates,finally,the template waveform and the whole period of sEMG signal is the average absolute value(MAV),variance(VAR),and other five kinds of feature extraction,at the same time as input parameters of pattern recognition algorithm.We select the BP neural network for the target finger movement to elbow sEMG signal classification,because the BP neural network is widely used in previous studies,high universality,adaptability is strong and stable structure.The experimental results show that the BP classification equipment have higher recognition accuracy,its corresponding target finger recognition rate at 90.35%.Through the recognition of four finger movements as a result,the design information to control the human-computer interaction interface,the game of four finger movements and finger gestures combination provides feasible in the study of human-computer interaction design.