Abstract:
The paper presents a denoising algorithm using which is particularly suited to ECG signals. The main stage of this algorithm consists in a MAP filtering in wavelet domain. Its effectiveness relies on the diversity enhancement of the signal to be processed and on realistic a-priori assumptions regarding statistical properties of the wavelet coefficients. Tests made on a big number ECG signals, in realistic conditions, showed very promising results. The noise is removed, while the useful waveforms are preserved.