I am having an ECG signal. It is contaminated with low-frequency breathing and motion artifacts. I want to use a wavelet to denoise the ECG signal. What wavelet I can choose and what level of decomposition I have to choose to accurately eliminate the artifacts.
Well reply by J. Rafiee
, You may also apply notch filter to desire cut-off frequency.
Wavelet-based denoising works well if the noise spreads to all frequencies. However, in your case, the noise signals are spread to the signal of interest. To remove such noise, one can use Wavelet denoising with a combination of statistical filters, such as LMS, RLS. For finding the wavelet level and Thersolding one needs to use the brute-force method as pointed in the previous answers. You can go through the following link:
Article ECG signal denoising by fractional wavelet transform thresholding
Article A Novel Adaptive Recursive Least Squares Filter to Remove th...
An ecg is usually a low frequency signal, to remove high frequency artifacts should be straight forward just use a good LP filter. If the artifacts are low frequency, embedded in low-frequency, then it's a bit more involved. One way is to use LMS, it's cheap and cheerful. .
However you are using the wavelet, my suggestion therefore is to use a simple undecimated wavelet, 2 coefficients to begin with (e.g. dB1) and a decomposition level at a high power of 2 first e.g. 2^11 or 2^10 this isolates the lower frequencies, then use Donoho's denoising (Matlab) function. If the output is not that which you were looking for, then repeat but work down the decomposition levels for higher frequencies. Eventually you may reach an acceptable output showing you the features you are looking for.
- Badges
- Science topic
- Similar topics
- Control Systems Engineering
- Filtering