L. G. Akulov, I. A. Tarasova, Yu. P. Mukha

Article describes algorithmic features of of Electrophysiologic Time Series Representation by Gabor Functions. Electrophysiologic signals such as EEG, EMG, EKG, EOG, etc. are having finite spectrum and localized at time. For discrete signals accuracy of Gabor function representation depend on analog to digital converting parameters. It is given proof of decomposition accuracy with steps of variable parameters and its possible intervals. It is describes base Matching Pursuit algorithm what have huge working time (more bigger than needed real time). Fast correlation method is a key for complexity reducing of decomposition. Next improvement is taking into account windowing features of Gabor functions (fast descent of curve in out of window bounds). Finite spectrum of signal is feature what reduced intervals of frequency variation. As an example of algorithm working are given real EEG signals and manual generated signals with time localized activity. It is shown that time of algorithm working may be real. It is a reason for apply method in biomedical measurement practice