I have to decompose a series of EEG signals which is sampled around (160-250 hz) to detect abnormalities in brain activities. Which mother wavelet will be better for doing that and how much levels are generally used in those cases? also why it is better in this case? Any related resources will be warmly welcomed.
You must try Daubchies' wavelets, since they are kind of bandlimited. Harr wavelet, for example, is rectangular, and hence, is not that useful. Number of decomposition levels depends upon the length of the window or analysis frame of your signal. What kind of abnormalities are you looking for ?
A.G. Ramakrishnan sir actually I am working on Epileptic Seizure Detection.
1. Simple approach: Test many general purpose (well known standar wavelets Haar, D4, etc.) and define a target function to extract the features from your signal
2. Review several of my publications about evolutionary optimization of parameterized wavelet filters. "Standard" filters are outperformed compared with optimized wavelets which concentrates the energy in the lowest number of coefficient. It achieves the best similarity of the wavelets and the signal.”
Check this paper.
Article Selection of Wavelet Basis Function for Image Compression – A Review
- Badges
- Science topic
- Similar topics
- Correction