I'm working on wavelet packet denoising of EMG signal. The dominant frequency of the signal ranges from 10 - 200 Hz. Should only the dominant frequencies be considered for selecting the level? Pls provide any suggestions on implementing wavelet packets for denoising.
In continuation with Mr Rafiee's answer, the decomposition level is depends on both dominant frequency as well as sampling frequency. As a thumb rule, the decomposition level is more as the sampling frequency is high. The decomposition level should be less to reduce computational load. The frequency band of each packet will be reduced half with each decomposition. The sampling frequency is required to calculate the decomposition level for desired dominant frequency.
I think you can use the matching pursuit algorithm.
For example:
Conference Paper Application of adaptive matching based on wavelets for spect...
As you have a particular type of signal viz EMG ,suggest ( a conjecture) to analyze first healthy responses at all levels and compare search top-down algorithm to detect anamolies.
This website was particularly interesting and may be useful :
https://www.bing.com/videos/search?q=EMG&view=detail&mid=A9351B618A8977162CA3A9351B618A8977162CA3&FORM=VIRE
Good Luck
Cheers
- Badges
- Science topic
- Similar topics
- Phytochemistry
- Extracts