A Comparative Evaluation on the Performance of Coifman Discrete and Stationary Wavelet Transform in ECG Signal Denoise Application

Abstract

A Comparative Evaluation on the ECG Signal Denoising performance using Coifman Discrete and Stationary wavelet transform is presented in this paper. The denoise approach used for the performance evaluation is known as wavelet threasholding denoise algorithm proposed by Donoho. The denoise approach was performed by forward wavelet transform, fixed form universal soft thresholding on the detail wavelet coefficients and inverse wavelet transform. The output Signal to Noise Ratio (SNR) in dB is used as a numerical measurement to evaluate the denoised signal quality. ECG signals contaminate with Gaussian noise of initial SNR of 10dB, 15dB and 20dB were produce for denoise evaluation. The evaluation results shows that stationary wavelet transform out performed discrete wavelet transform at all 5 decomposition levels up to 13.6dB of improvement. In term of wavelet family, the Coifman N = 1 with stationary wavelet achieves the best overall denoise performance with the output SNR improvement of up to 6.6dB. The evaluation results presented in this paper provide an insight of how the translation invariant property of stationary wavelet transform contribute to the improvement of the denoise signal quality as well as provide a reference selection guide for Coifman wavelet family and the number of decomposition level to achieve optimum denoise performance for ECG signal.