Hb vsEMG signal classification with time domain and Frequency domain features using LDA and ANN classifier

The surface electromyography (sEMG) signals have been widely employed for the development of the human–machine interface and have enormous bio-engineering applications. Movement identification of SEMG signal plays a significant factor in the designing of human assistive robotic devices. This work compared the time domain (TD) and frequency domain (FD) features by using linear discriminant analysis (LDA) and artificial neural network (ANN) classifiers for six different hand movements’ identification. Discrete Wavelet Transform is employed mainly for de-noising the sEMG signal before the feature extraction. Finally, a feature vector is formed which consists of all TD and FD features for classification purpose. ANN exhibited96.4% accuracy and found better as compared to the LDA classifier whereas classification accuracy of LDA classifier was found 94.5%. The resultexhibits that ANN has a greater ability for sEMG signal classification as compared to LDA classifier and further suggested to design the assistive robotic technology.