Application of non-linear SVM and LDA in sEMG gesture recognition
L. Lu · 2014
For the rationality problem of nonlinear Support Vector Machine(SVM) and LDA in the application of gesture recognition on EMG signal, an experiment to compare the discriminant ability between the non-linear classifier SVM and a conventional used linear discriminant analysis(LDA) has been done. Firstly, data of 3 groups with different hand motion are collected and recorded by using 1 to 6 channels forearm EMG signals. Then, the accurate rate of EMG hand motion recognition between SVM and LDA is compared. Finally, the 2 algorithms gesture recognition rate related to the EMG electrodes number, according to the number of electrodes for selecting the appropriate classification algorithm. Analysis results show that this experiment is of great importance for the choice of hand motion recognition algorithm in the case of fewer electrodes.