Design of rehabilitation system based on multi-channel EMG signal recognition and virtual reality technology
Yang, Xuerui (2018-02-14)
Design of rehabilitation system based on multi-channel EMG signal recognition and virtual reality technology
Yang, Xuerui
(14.02.2018)
Tätä artikkelia/julkaisua ei ole tallennettu UTUPubiin. Julkaisun tiedoissa voi kuitenkin olla linkki toisaalle tallennettuun artikkeliin / julkaisuun.
Turun yliopisto
Tiivistelmä
Electromyogram (EMG) signal is a bioelectrical signal generated in muscles and can be detected on the surface of the human skin. Virtual reality has developed rapidly in the field of rehabilitation and help patients recover by natural interaction. At present, the virtual motion capture and recognition is mainly based on computer vision technology. Due to the visual masking effect, it has the disadvantages of complex algorithm, low precision and high cost. This thesis developed indoor illumination system and hand function rehabilitation training system based on gesture interaction and virtual reality through the acquisition, processing and recognition of surface EMG (sEMG) signal.
The method of this research and the specific work is as follows. First, we used Myo armbands to conduct sEMG signal acquisition, and through the comparison of different wavelet functions and decomposition levels, an optimal wavelet processing method is found. Then, the feature vectors of sEMG signals in time domain, frequency domain and time-frequency domain are analyzed. Finally, support vector machine algorithm is used to identify the corresponding gestures. Meanwhile, an indoor lighting control system is designed in which users do not need to switch on the light but complete by hand gestures. Another system is virtual display rehabilitation training system based on gesture recognition. Unity engine and virtual reality technology are applied to develop the task of hand rehabilitation training in virtual environment, in which users complete various tasks with their gestures.
In this thesis, the noise reduction, feature extraction and gesture recognition are realized by MATLAB. The development of the lighting control system based on the development board has been carried out, as well as the virtual reality hand function rehabilitation training task system. This study verified the feasibility of hand gesture recognition based on sEMG signals in the Internet of things control and virtual rehabilitation.
The method of this research and the specific work is as follows. First, we used Myo armbands to conduct sEMG signal acquisition, and through the comparison of different wavelet functions and decomposition levels, an optimal wavelet processing method is found. Then, the feature vectors of sEMG signals in time domain, frequency domain and time-frequency domain are analyzed. Finally, support vector machine algorithm is used to identify the corresponding gestures. Meanwhile, an indoor lighting control system is designed in which users do not need to switch on the light but complete by hand gestures. Another system is virtual display rehabilitation training system based on gesture recognition. Unity engine and virtual reality technology are applied to develop the task of hand rehabilitation training in virtual environment, in which users complete various tasks with their gestures.
In this thesis, the noise reduction, feature extraction and gesture recognition are realized by MATLAB. The development of the lighting control system based on the development board has been carried out, as well as the virtual reality hand function rehabilitation training task system. This study verified the feasibility of hand gesture recognition based on sEMG signals in the Internet of things control and virtual rehabilitation.