Constraining noise in ECG signals

In the paper 11 methods are presented to constrain and eliminate the main noise in ECG signal including disturbance of power frequency and its harmonic, baseline drift and noise of contacted electrodes. From the viewpoint of the advanced localized time-frequency feature, the principium of using wavelet transform to constrain the noise in ECG signal as well as the feature scales where various noise exist and how to constrain it is discussed. With combined wavelet transform and adaptive filtering in which the problem of selecting threshold in wavelet filtering is avoided, thus the noise constraining becomes more effectively. There exist many methods to constrain the noise in ECG signal, but various method of noise eliminating should be used in eliminating various noise in ECG signal, and no one can constrain effectively all kinds of noise disturbance in ECG signal. Comparatively since wavelet transform can make multi-scale decomposition of ECG signal, and then process a specific noise, which leads to get better noise-eliminating result. Whereas the wavelet adaptive filtering enables avoiding the problem of threshold selection in noise-eliminated by wavelet transform and obtaining better filtering result as well.