Performance Evaluation of Coifman Wavelet for ECG Signal Denoising

The performance evaluation of Coifman wavelet for ECG signal denoising is presented in this paper. The Coifman wavelet family was use to evaluation its performance on the denoising of ECG signal. The denoising technique was performed by forward discrete wavelet transform up to decomposition of 5 levels, soft thresholding on the wavelet coefficients and inverse discrete wavelet transform. The Signal to Noise Ratio (SNR) in dB is used as a numerical measurement of denoised signal quality. The ECG Signal was obtained from MIT-BIH Arrhythmia reference database. White gaussian noise was added to the clean reference ECG signal to produce the noisy ECG signal with 3 noise levels of initial SNR of 6.5dB, 16.1dB and 20.5dB for denoise evaluation. The evaluation results shows that Coifman N = 5 wavelet achieves the best overall denoise performance at all 3 noise levels for ECG signal with the SNR improvement of up to 6.3dB. The evaluation results presented in this paper provide a basic reference for Coifman wavelet family selection for ECG signal denoising applications.