THE USAGE OF STATISTICAL FEATURES IN THE APPROXIMATION COMPONENTS OF WAVELET DECOMPOSITION FOR ECG CLASSIFICATION: A CASE STUDY FOR STANDING, WALKING AND SINGLE JUMP CONDITIONS

Abstract

The purpose of this study is to classifyelectrocardiogram (ECG) signals with a high accuracy rate. The ECG signals usedare obtained from the Physiobank archive. These signals are preprocessed toremove noise. Features with distinctiveness in classification are obtained bothin the time domain and the frequency domain. The Discrete Wavelet Transformmethod is used for feature extraction in frequency domain. ECG signals areclassified by the Naive Bayes method after the required features are extracted.