Human‐computer Interaction System Design Based On Surface Emg Signals

The human body gestures contains a wealth of information, so gesture recognition become the hot spot in human-computer interaction researching recent years. In this paper, we designed a human-computer interaction system based on the multi-channel SEMG of hand motions. The system detects the forearm muscle multi-channel SEMG, the AR model is established to extract the signal eigenvalue and the artificial neural network classifier(BP) is utilized to distinguish the four different gestures that the results of pattern recognition are applied as the quad rotor control signals. At last, the results of pattern recognition are used as control signal of aircraft to complete real-time interactive process.First of all, this paper chooses the chip AD620 and OPA2277 with high common mode rejection ratio.This paper uses the form of differential amplifier circuit to obtain multi-channel SEMG coming from the different gestures corresponds from the medial, lateral, feet side of the forearm muscle in the human body. It guarantees the access to SNR of signal by using trap and band-pass filter to filter out noise effectively. Then, this paper compares the advantages and disadvantages of the variety of feature extraction methods about the multi-channel SEMG. The AR model and the U-C arithmetic are chosen as the final feature extraction method in this paper. and the mathematic method is used to calculate to determine the best order of AR model. AR model is set up and is used to access to get the feature extraction of multi-channel SEMG. Then, it uses the BP artificial neural network algorithm for pattern recognition about the classification of gestures in this paper. And in view of the slow convergence speed and the low shortcomings of the BP network algorithm, it puts forward the adaptive variable step improvement methods to solve the problem effectively in this paper. Gestures classification results are obtained. The classification results is used as the control signals, and they control the back-end quad rotor aircraft flying to complete the corresponding action to realize human-computer interaction. Finally, in terms of human-computer interaction, this paper takes the interaction of physical control object using quad rotor aircraft as actuators to complete the corresponding action. It changes the previous research that the classification of the gesture recognition of multi-channel SEMG results are applied to the simulation of the situation.In this paper the system gets the accurate and multi-channel SEMG successfully and gets the pattern recognition classification results well. It improves the precision of classification of gestures and the classification results of the quad rotor aircraft is used for carrying out the control effectively.