Feature Extraction Method of Epileptic EEG Signal based on Wavelet Packet and Improved Fuzzy Entropy

Abstract

Epileptic eeg signal has obvious characteristic information, which can be used as an important basis to judge whether epileptic seizure occurs. Because of the low recognition rate of single feature extraction method, a method of eeg feature extraction based on wavelet packet transform and improved fuzzy entropy was proposed. In view of the characteristics of eeg signal with large noise and weak signal, the Wavelet packet Transform (WPT) is used to decompose the EEG signal with multi-resolution and make it into the signal with different characteristics. The original Fuzzy entropy (Fuzzy EN) algorithm was improved to improve its ability of reflecting the degree of irregularity and complexity of time series. Finally, the feature extraction of epileptic EEG signal was completed by combining the wavelet packet transform method.