ECG denoise method based on wavelet function learning

In this paper, we propose a new denoise method for noisy electrocardiogram (ECG) signals. We employ an n-gram-based wavelet learning in order to investigate optimal classical wavelet sequences for ECG signals denoise. Our main approach separates the ECG signal of the interest into multi-windows then assigns the optimal wavelet to each window. The wavelet learning approach uses the mean square error(MSE) as a feature to generate an n-gram table. Also, we selected MSE and the signal-to-noise ratio(SNR) for evaluation factors. As a result of simulation, we confirmed that the performance become more precise than the previous approaches.