Time-Frequency Analysis Based on Matching Pursuit and Its Application on Epilepsy

Since the matching pursuit algorithm has good parametric characterization in signal decomposition, this paper considers the application of matching pursuit to describing the EEG's time-frequency distribution. Compared with other methods such as short time Fourier transform and Wigner-Ville distribution, matching pursuit algorithm can get a better result. It was also found that significant difference exist between normal EEG and epilepsy EEG's time-frequency distribution.