Adaptive wavelet transform-based method for recognizing characteristic oscillatory patterns

Abstract

The problem concerning the automatic recognition of characteristic oscillatory patterns in multicomponent signals is investigated using the brain’s electric activity records, electroencephalograms (EEGs), as an example. It has been ascertained that recognition errors can be decreased by optimally selecting continuous wavelet transform (CWT) parameters to obtain characteristics describing the most important information on analyzed patterns. The adaptive CWT-based method for identifying the characteristic types of EEG rhythmic activity is proposed.