UNSUPERVISED OPTIMIZATION FOR BRAIN EEG SIGNAL ARTIFACT ERADICATION USING LOWER ORDER IIR FILTER DESIGN

While being acquired, the brian electroencephalography (EEG) signals are subject to a variety of motion artifacts. Therefore, it is crucial to get rid of these artifacts at the beginning of the investigation of human diseases. The removal of the motions using current artifact removal techniques is either too complicated or ineffective. The goal is to create a filter that simultaneously removes all motion artifacts. The idea presented in this study is to create the best lower order IIR filter possible to eliminate noise, eye blinks, and muscular artifacts from EEG signals. An IIR filter is initially designed using a hybrid combination of pass and cutoff band filters. After that, the approach of unsupervised transfer function (TF) optimization is used to replicate the ideal filter coefficients. For the design of lower order filters, the optimum coefficients are applied. Performance of the filter is evaluated on multichannel real EEG signal database. The parametric evaluation is presented using order of TF for fitters designs and qualitative evaluation based o the removal of eye blink peaks for the EEG signal databases