Identifikasi dan Klasifikasi Sinyal EEG terhadap Rangsangan Suara dengan Ekstraksi Wavelet dan Spektral Daya
AbstractIn this research the development of identification and classification technique of three wave components of EEG signal, named alpha, beta and theta, is considered. The technique is combination of wavelet transform and power spectral analysis. Wavelet transform was used to extract the wave components so it reduces the data without loss of the information. The wavelet transform also reduces the aspects of non-stationary of the EEG signal. The EEG's wave classification was based on the appearance of the wave, synchronization between symmetric hemispheres, and the wave energy dominance, in its frequency region. The EEG signals used in this research were obtained from 5 individually-independent subjects after 2.5 minutes sound stimulation. 10 sounds of music and 2 natural sounds were used as sound stimulation in this research. Then, 16 channels of EEG signals, obtained from every individual subject after a sound stimulation, were analyzed. The technique shows that the sound stimulation increases the appearance of the alpha wave by 75% and decreases beta and theta waves by 48% and 56%, respectively. Furthermore, the sound stimulations were used in the research to increase the synchronization balance between symmetric channels by 75%. In addition, this research shows that the signal extraction using wavelet packet provided small deviation and reduced non-stationary aspects, so that it improves the power spectral analysis used in the technique.