Classification of Motor Imagery EEG Based on Sparsification and Non-negative Matrix Factorization

Abstract

The analysis of EEG is a hot topic in the area of biomedical signal processing. In this paper, the EEG signals with Mu (Μ) rhythm and Beta (Β) rhythm are used to solve the motor imagery problem, i.e., the imagery of the left hand and the right hand. The collected raw data is first filtered by FIR band-pass filter, followed by using the maximization of feature difference to increase the sparsity of the matrix. Then, to reduce the redundant information of these features, a non-negative matrix factorization (NMF) method is employed. Due to the usage of the NMF scheme, the obtained factorizations has been better class property. Simulations show that our method achieves higher classification accuracy (more than 91%) than existing results (about 86%).