Ocular artifact removal from EEG signals using discrete orthonormal stockwell transform

Electroencephalographic data is often contaminated by artifacts generated from non-cerebral activities. Elimination of such artifacts is necessary prior to information extraction from electroencephalogram data to uphold classification highly accurate. In this paper, a novel method of ocular artifact removal of scalp electroencephalogram data is proposed for error-free feature extraction and classification of electroencephalogram signals for successful implementation of brain-computer interface and disease diagnosis. This method is based on joint use of independent component analysis and discrete orthonormal stockwell transform. It involves three corrective steps viz. blind source separation of contaminated electroencephalogram data, identification of ocular artifact components and discrete orthonormal stockwell transform denoising of identified components. The results are validated on simulated electroencephalogram signals by quantitative evaluation analysis. The results validate excellence of proposed ocular artifact removal technique to attain higher accuracy in electroencephalogram based applications.