OPEN ACCESS
We deal in this paper with the extraction of multiresolution statistical signatures for the characterization of transient signals in strongly noisy contexts. These short-time signals have sharp and highly variable frequency components. The Time-Frequency analysis window to adopt is then a major issue. Thus we have chosen the wavelet packet domain due to its natural ability to provide multiple time-frequency resolutions. We propose a new oriented Markov model dedicated to the wavelet packet transform, which offers sharp analysis of frequency variations in a signal, locally in time and at several resolutions. We show its efficiency on synthetic signals and we then illustrate its applicative relevance in a biomedical context related to the detection of transient signals in pulmonary sounds.
RÉSUMÉ
Nous nous intéressons dans cet article à l’extraction de comportements statistiques multirésolutions pour la caractérisation et la segmentation de signaux transitoires dans un contexte fortement bruité. Ces signaux de courte durée possèdent des composantes fréquentielles très localisées et fortement variables. Le choix du compromis temps/fréquence pour l’étude de ces signaux est donc crucial. Nous nous plaçons de ce fait dans le domaine transformé en paquets d’ondelettes, permettant une analyse fine des variations fréquentielles du signal. Nous proposons un modèle d’arbre de Markov original adapté à la décomposition en paquets d’ondelettes afin d’intégrer l’information multirésolution d’échelle en échelle dans un objectif de segmentation. Nous validons l’approche sur des signaux synthétiques, puis nous illustrons son intérêt applicatif dans un contexte biomédical liée à la détection de signaux transitoires dans les signaux pulmonaires.
Hidden Markov Model, Signatures Extraction, Transient Signals Detection, Generalized Gaussian Mixtures, ICE
MOTS-CLÉS
Modèles de Markov Cachés, Extraction de Signature, Détection de Signaux Transitoires, Mélanges de Gaussienne Généralisée, ICE
Akay M., Mello C. (1997), « Wavelets for biomedical signal processing », Engineering in Medicine and Biology society, 1997. Proceedings of the 19th Annual International Conference of the IEEE, vol. 6.
Bahoura M., Pelletier C. (2004a), « Respiratory sounds classification using cepstral analysis and Gaus-sian mixture models », Engineering in Medicine and Biology Society, 2004. EMBC 2004. Conference Proceedings. 26th Annual International Conference of the.
Bahoura M., Pelletier C. (2004b), « Respiratory sounds classification using Gaussian mixture models », Electrical and Computer Engineering, 2004. Canadian Conference on.
Buccigrossi R., Simoncelli E. (1999), « Image compression via joint statistical characterization in the wavelet domain », Image Processing, IEEE Transactions on, vol. 8, n° 12, p. 1688-1701.
Choi H., Baraniuk R. (2001), « Multiscale image segmentation using wavelet-domain hidden markov models », Image Processing, IEEE Transactions on, vol. 10, n° 9, p. 1309-1321.
Choi H., Hendricks B., Baraniuk R. (1999), « Analysis of multiscale texture segmentation using wavelet-domain hidden Markov models », Signals, Systems, and Computers, 1999. Conference Record of the Thirty-Third Asilomar Conference on, vol. 2.
Chou K., Golden S., Willsky A. (1993), « Multiresolution stochastic models, data fusion, and wavelet transforms », Signal Processing, vol. 34, n° 3, p. 257-282.
Coifman R. R., Wickerhauser M. V.(1992), « Entropy-Based Algorithms for Best Basis Selection », IEEE Transactions on Information Theory, vol. 38, n° 2, p. 713-718.
Crouse M., Baraniuk R. (1997), « Contextual hidden Markov models for wavelet-domain signal processing », Signals, Systems & Computers, 1997. Conference Record of the Thirty-First Asi-lomar Conference on, vol. 1.
Crouse M., Nowak R., Baraniuk R. (1998), « Wavelet-based statistical signal processing using hidden markov models », Signal Processing, IEEE Transactions on, vol. 46, n° 4, p. 886-902.
Dasgupta N., Runkle P., Couchman L., Carin L. (2001), « Dual hidden Markov model for characterizing wavelet coefficients from multi-aspect scattering data », Signal Processing, vol. 81, n° 6, p. 1303-1316.
Daubechies I. (1992), « Ten lectures on wavelets », Society for Industrial and Applied Mathematics Philadelphia, PA, USA, vol. , p. 357.
Fan G., Xia X. (2000), « Improved hidden Markov models in the wavelet-domain », IEEE transactions on signal processing, vol. 49, n° 1, p. 115-120.
Fan G., Xia X. (2001), « Image Denoising Using a Local Contextual Hidden Markov Model in the Wavelet Domain », Signal Processing Letters, IEEE, vol. 8, n° 5, p. 125-128.
Fan G., Xia X. (2003), « Wavelet-based texture analysis and synthesis using hidden Markov models », Circuits and Systems I : Fundamental Theory and Applications, IEEE Transactions on [see also Circuits and Systems I : Regular Papers, IEEE Transactions on], vol. 50, n° 1, p. 106-120.
Fieguth P., Karl W., Willsky A., Wunsch C. (1995), « Multiresolution optimal interpolation and statistical analysis of TOPEX/POSEIDONsatellite altimetry », Geoscience and Remote Sensing, IEEE Transactions on, vol. 33, n° 2, p. 280-292.
Graffigne C., Heitz F., Perez P., Preteux F., Sigelle M., Zerubia J. (1995), « Hierarchical Markov random field models applied to image analysis : a review », Proceedings of SPIE, vol. 2568, SPIE, p. 2.
Homs-Corbera A., Fiz J., Morera J., Jane R. (2004), « Time-frequency detection and analysis of wheezes during forced exhalation », Biomedical Engineering, IEEE Transactions on, vol. 51, n° 1, p. 182-186.
Homs-Corbera A., Jane R., Fiz J., Morera J. (2000), « Algorithm for time-frequency detection and analysis of wheezes », Engineering in Medicine and Biology Society, 2000. Proceedings of the 22nd Annual International Conference of the IEEE.
Huang J., Box F., Providence R., Mumford D. (1999), « Statistics of Natural Images and Models », 1999 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, Institute of Electrical & Electronics Engineers (IEEE).
Ichir M., Mohammad-Djafari A. (2006), « Hidden Markov Models for Wavelet-Based Blind Source Separation », IEEE Transactions On Image Processing, vol. 15, n° 7, p. 1887.
Kahya Y., Yerer S., Cerid O (2001)., « A wavelet-based instrument for detection of crackles in pulmonary sounds », Engineering in Medicine and Biology Society, 2001. IEEE-EMBS., vol. 4, p. 3175-3178.
Lauritzen S. (1996), Graphical Models, Oxford University Press, USA.
Le Cam S., Belghith A., Collet C., Salzenstein F. (2009), « Wheezing Sounds Detection Using Multivariate Gaussian Distributions », Acoustics, Speech and Signal Processing, 2009. ICASSP 2009. IEEE International Conference on.
Le Cam S., Collet C., Salzenstein F. (2008), « Acoustical respiratory signal analysis and phase detec-tion », Acoustics, Speech and Signal Processing, 2008. ICASSP 2008. IEEE International Conference on, vol. , p. 3629-3632.
Lovejoy S., Schertzer D. (2006), « Multifractals, cloud radiances and rain », Journal of Hydrology, vol. 322, n° 1-4, p. 59-88.
Lucke H. (1996), « Which stochastic models allow Baum-Welch training ? », IEEE transactions on signal processing, vol. 44, n° 11, p. 2746-2756.
Ma Y., Tian Y., Zhang J. (2007), « Wavelet-Domain Image Denoising Using Contextual Hidden Markov Tree Model », Automation and Logistics, 2007 IEEE International Conference on, p. 2617-2621.
Mallat S. (1998), A wavelet tour of signal processing, Academic Press.
Molla S., Torresani B. (2005), « An hybrid audio scheme using hidden Markov models of waveforms », Applied and Computational Harmonic Analysis, vol. 18, n° 2, p. 137-166.
Monfrini E., Lecomte J., Desbouvries F., Pieczynski W. (2003), « Image and signal restoration using pairwise Markov trees », Statistical Signal Processing, 2003 IEEE Workshop on, p. 174-177.
Pieczynski W. (1994), « Champs de Markov cachés et estimation conditionnelle itérative », Traitement du Signal, vol. 11, n° 2, p. 141-153.
Pieczynski W. (2002), « Arbres de Markov Couple - Pairwise Markov Trees », CR Acad. Sci. Paris, Ser. I, vol. 335, p. 79-82.
Pieczynski W (2008)., « Sur la convergence de l’estimation conditionnelle itérative », Comptes Rendus Mathematique, vol. 346, n° 7-8, p. 457-460.
Provost J., Collet C., Rostaing P., Pérez P., Bouthemy P. (2004), « Hierarchical Markovian segmentation of multispectral images for the reconstruction of water depth maps », Computer Vision and Image Understanding, vol. 93, n° 2, p. 155-174.
Qiu Y., Whittaker A., Lucas M., Anderson K. (2005), « Automatic wheeze detection based on auditory modelling », Proceedings of the Institution of Mechanical Engineers, Part H : Journal of Engineering in Medicine, vol. 219, n° 3, p. 219-227.
Rajpoot N. (2002), « Texture Classification Using Discriminant Wavelet Packet Subbands », Proceedings of the IEEE Midwest Symposium On Circuits And Systems, vol. 45, Lida Ray Technologies inc.
Reynolds D., Rose R. (1995), « Robust text-independent speaker identification using Gaussian mixture speaker models », Speech and Audio Processing, IEEE Transactions on, vol. 3, n° 1, p. 72-83.
Sarikaya R., Pellom B., Hansen J. (1998), « Wavelet Packet Transform Features with Application to Speaker Identification », Third IEEE Nordic Signal Processing Symposium, ISCA.
Singh B., Tiwari A. (2006), « Optimal selection of wavelet basis function applied to ECG signal denoising », Digital Signal Processing, vol. 16, n° 3, p. 275-287.
Sovijarvi A., Malmberg L., Charbonneau G., Vanderschoot J., Dalmasso F., Sacco C., Rossi M., Earis J. (2000), « Characteristics of breath sounds and adventitious respiratory sounds », Eur Respir Rev, vol. 10, n° 77, p. 591-596.
Srinivasan P., Jamieson L. (1998), « High-quality audio compression using an adaptive wavelet packet decomposition and psychoacoustic modeling », Signal Processing, IEEE Transactions on [see also Acoustics, Speech, and Signal Processing, IEEE Transactions on], vol. 46, n° 4, p. 1085-1093.
Sun Q., Gou S., Jiao L. (2004), « A New Approach to Unsupervised Image Segmentation Based on Wavelet-Domain Hidden Markov Tree Models », Lecture Notes In Computer Science, vol., p. 41-48.
Sun Q., Hou B., Jiao L. (2005a), « Automatic Texture Segmentation Based on WaveletDomain Hidden Markov Tree », Lecture Notes In Computer Science, vol. 3773, p. 470.
Sun Q., Sha Y., Gao X., Hou B., Jiao L. (2005b), « A Fully Unsupervised Image Segmentation Algorithm Based on Wavelet-Domain Hidden Markov Tree Models », Lecture Notes In Computer Science, vol. 3708, p. 507.
Sun W., Palazoglu A., Romagnoli J. (2003), « Detecting abnormal process trends by waveletdomain hidden Markov models », AIChE Journal, vol. 49, n° 1, p. 140-150.
Taplidou S., Hadjileontiadis L., Kitsas I., Panoulas K., Penzel T., Gross V., Panas S. (2004), « On Applying Continuous Wavelet Transform in Wheeze Analysis », Engineering in Medicine and Biology Society, 2004. EMBC 2004. Conference Proceedings. 26th Annual International Conference of the.
Wainwright M., Simoncelli E., Willsky A. (2001), « Random Cascades on Wavelet Trees and Their Use in Analyzing and Modeling Natural Images », Applied and Computational Harmonic Analysis, vol. 11, n° 1, p. 89-123.
Waris M., Helistöa P., Haltsonena S., Saarinenb A., Sovijärvib A. (1998), « A new method for automatic wheeze detection », Technology and Health Care, vol. 6, n° 1, p. 33-40.
Wickerhauser M. (1994), Adapted Wavelet Analysis : From Theory to Software, Wellesley (Mass.).
Willsky A. (2002), « Multiresolution Markov models for signal and image processing », Proceedings of the IEEE, vol. 90, n° 8, p. 1396-1458.
Wunsch C., Stammer D. (1995), « The global frequency-wavenumber spectrum of oceanic variability estimated from TOPEX/POSEIDON altimetric measurements », Journal of Geophysical Research, vol. 100, n° C12, p. 24895-24910.