2014 | OriginalPaper | Buchkapitel
Matching Pursuit with Asymmetric Functions for Signal Decomposition and Parameterization
verfasst von : K. J. Blinowska, W. W. Jedrzejczak, K. Kwaskiewicz
Erschienen in: XIII Mediterranean Conference on Medical and Biological Engineering and Computing 2013
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by (Link öffnet in neuem Fenster)
The high resolution time-frequency method - adaptive approximations by matching pursuit relies on the decomposition of signals into basic waveforms taken from a very large dictionary. Usually Gabor functions are used, however they are often not optimal in describing waveforms present in biological and medical signals. This especially concerns the signals of steep rise and slower decay. For the decomposition of this kind of signals we introduced a dictionary of functions of various degrees of asymmetry. The application of this dictionary to otoacoustic emissions and speech demonstrated the advantages of the proposed method. The approach allows for correct determination of the latencies of the components, removes the “pre-echo” effect generated by symmetric waveforms that do not match the structures of the analyzed signal sufficiently and provides more sparse representation. Additionally, we introduced a time-frequency-amplitude map, which is more adequate for representation of asymmetric waveforms than the conventional time-frequency-energy distribution.