Ali Komaty
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Ali Komaty has a total of 19 co-authors. Here are the top co-authors by number of shared publications:
- Abdel-Ouahab Boudraa
Institut de Recherche de l'Ecole Navale , CC 600, 29240 Brest Cedex 9
6 shared publications Delphine Dare
Ecole Navale -IRENav - EA 3634 - BCRM Brest, CC 600, 29240 Brest Cedex 9, France
3 shared publicationsA. Drira
IRENav, Ecole Navale, BCRM Brest, CC 600, 29240 Brest Cedex 9, France
1 shared publicationAurelia Fraysse
Laboratoire des Signaux et Systèmes (L2S), CNRS - Supelec - Univ. Paris-Sud, 3, Rue Joliot Curie, 91192 Gif sur Yvette, France
1 shared publicationBenoit Augier
IRENav, Ecole Navale, Brest, France
1 shared publicationClaude Barras
1 shared publication
Latest Publications
- 31 March 2026
- Published by Agence Bibliographique de l'Enseignement Supérieur
No
Traitement et analyse des processus stochastiques par EMD et ses extensions: L’objectif de cette thèse est d’analyser le comportement de la décomposition modale empirique (EMD) et sa version multivariée (MEMD) dans le cas de processus stochastiques : bruit Gaussien fractionnaire (fGn) et processus symétrique alpha stable (SαS). Le fGn est un bruit large bande généralisant le cas du bruit blanc Gaussien et qui trouve des applications dans de nombreux domaines tels que le trafic internet, l’économie ou le climat. Par ailleurs, la nature «impulsive» d’un certain nombre de signaux (craquement des glaces, bruit des crevettes claqueuses, potentiel de champ local en neurosciences,…) est indéniable et le modèle Gaussien ne convient pas pour leur modélisation. La distribution SαS est une solution pour modéliser cette classe de signaux non-Gaussiens. L’EMD est un outil bien adapté au traitement et à l’analyse de ces signaux réels qui sont, en général, de nature complexe (non stationnaire,non linéaire). En effet, cette technique, pilotée par les données, permet la décomposition d’un signal en une somme réduite de composantes oscillantes, extraites de manière itérative, appelées modes empiriques ou IMFs (Intrinsic Mode Functions). Ainsi, nous avons montré que le MEMD s’organise spontanément en une structure de banc de filtres presque dyadiques. L'auto-similarité en termes de représentation spectrale des modes a aussi été établie. En outre, un estimateur de l’exposant de Hurst, caractérisant le fGn, a été construit et ses performances ont été comparées, en particulier à celles de l’approche ondelettes. Cette propriété de banc de filtres du MEMD a été vérifiée sur des données d'hydrodynamique navale (écoulement turbulent) et leur auto-similarité a été mise en évidence. De plus, l’estimation du coefficient de Hurst a mis en avant l’aspect longue dépendance (corrélation positive) des données. Enfin, l’aspect banc de filtres de l’EMD a été exploité à des fins de filtrage dans le domaine temporel en utilisant une mesure de similarité entre les densités de probabilités des modes extraits et celle du signal d’entrée. Pour éviter le problème du mode mixing de l'EMD standard, une approche de débruitage dans le domaine fréquentiel par une reconstruction complète des IMFs préalablement seuillées a été menée. L’ensemble des résultats a été validé par des simulations intensives (Monte Carlo) et sur des signaux réels.
- 31 March 2026
- Published by Agence Bibliographique de l'Enseignement Supérieur
No
Traitement et analyse des processus stochastiques par EMD et ses extensions: L’objectif de cette thèse est d’analyser le comportement de la décomposition modale empirique (EMD) et sa version multivariée (MEMD) dans le cas de processus stochastiques : bruit Gaussien fractionnaire (fGn) et processus symétrique alpha stable (SαS). Le fGn est un bruit large bande généralisant le cas du bruit blanc Gaussien et qui trouve des applications dans de nombreux domaines tels que le trafic internet, l’économie ou le climat. Par ailleurs, la nature «impulsive» d’un certain nombre de signaux (craquement des glaces, bruit des crevettes claqueuses, potentiel de champ local en neurosciences,…) est indéniable et le modèle Gaussien ne convient pas pour leur modélisation. La distribution SαS est une solution pour modéliser cette classe de signaux non-Gaussiens. L’EMD est un outil bien adapté au traitement et à l’analyse de ces signaux réels qui sont, en général, de nature complexe (non stationnaire,non linéaire). En effet, cette technique, pilotée par les données, permet la décomposition d’un signal en une somme réduite de composantes oscillantes, extraites de manière itérative, appelées modes empiriques ou IMFs (Intrinsic Mode Functions). Ainsi, nous avons montré que le MEMD s’organise spontanément en une structure de banc de filtres presque dyadiques. L'auto-similarité en termes de représentation spectrale des modes a aussi été établie. En outre, un estimateur de l’exposant de Hurst, caractérisant le fGn, a été construit et ses performances ont été comparées, en particulier à celles de l’approche ondelettes. Cette propriété de banc de filtres du MEMD a été vérifiée sur des données d'hydrodynamique navale (écoulement turbulent) et leur auto-similarité a été mise en évidence. De plus, l’estimation du coefficient de Hurst a mis en avant l’aspect longue dépendance (corrélation positive) des données. Enfin, l’aspect banc de filtres de l’EMD a été exploité à des fins de filtrage dans le domaine temporel en utilisant une mesure de similarité entre les densités de probabilités des modes extraits et celle du signal d’entrée. Pour éviter le problème du mode mixing de l'EMD standard, une approche de débruitage dans le domaine fréquentiel par une reconstruction complète des IMFs préalablement seuillées a été menée. L’ensemble des résultats a été validé par des simulations intensives (Monte Carlo) et sur des signaux réels.
- 26 June 2018
- Published by International Speech Communication Association
- p. 140-146
Low-latency speaker spotting with online diarization and detection
- 26 June 2018
- Published by International Speech Communication Association
- p. 140-146
Low-latency speaker spotting with online diarization and detection
- 1 August 2015
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 280-284
On the estimation of time-delays for image source method
The image source method is based on the study of signals reflected by a sedimentary seabed. It produces an estimation of the geoacoustic parameters of the stratified sediments. The key step of this method is the detection and estimation of the time delays of the signals relected by the interfaces between the sedimentary strata. The Teager-Kaiser (TK) operator is known as a reliable way to extract from the recorded signal the time delays of the reflected signals. Due to its sensitivity to discontinuities and excellent time resolution, this operator is a good tool to detect and estimate the time delays. We investigate whether the TK operator associated with time-frequency approaches (spectrogram, wavelets) could be of interest to the detection of time delays and perform better than the TK operator applied directly on the relected signals. The different methods are presented and compared first on synthetic signals. The results show that the use of wavelet transform improve the quality of the time-delays estimation. The algorithm is finally applied to real signals acquired in an acoustic tank experiment and the estimated parameters are very close to the ground-truth.
- 1 August 2015
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 280-284
On the estimation of time-delays for image source method
The image source method is based on the study of signals reflected by a sedimentary seabed. It produces an estimation of the geoacoustic parameters of the stratified sediments. The key step of this method is the detection and estimation of the time delays of the signals relected by the interfaces between the sedimentary strata. The Teager-Kaiser (TK) operator is known as a reliable way to extract from the recorded signal the time delays of the reflected signals. Due to its sensitivity to discontinuities and excellent time resolution, this operator is a good tool to detect and estimate the time delays. We investigate whether the TK operator associated with time-frequency approaches (spectrogram, wavelets) could be of interest to the detection of time delays and perform better than the TK operator applied directly on the relected signals. The different methods are presented and compared first on synthetic signals. The results show that the use of wavelet transform improve the quality of the time-delays estimation. The algorithm is finally applied to real signals acquired in an acoustic tank experiment and the estimated parameters are very close to the ground-truth.
- 1 August 2015
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 2656-2660
On signal denoising by EMD in the frequency domain
In this work a new denoising scheme based on the empirical mode decomposition associated with a frequency analysis is introduced. Compared to classical approaches where the extracted modes are thresholded in time domain, in the proposed strategy the thresholding is done in the frequency domain. Each mode is divided into blocks of equal length where the frequency content of each one is analyzed. Relevant modes are identified using an energy and a frequency thresholds obtained by training. The denoised signal is obtained by the superposition of the thresholded modes. The effectiveness of the proposed scheme is illustrated on synthetic and real signals and the results compared to those of methods reported recently.
- 1 August 2015
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 2656-2660
On signal denoising by EMD in the frequency domain
In this work a new denoising scheme based on the empirical mode decomposition associated with a frequency analysis is introduced. Compared to classical approaches where the extracted modes are thresholded in time domain, in the proposed strategy the thresholding is done in the frequency domain. Each mode is divided into blocks of equal length where the frequency content of each one is analyzed. Relevant modes are identified using an energy and a frequency thresholds obtained by training. The denoised signal is obtained by the superposition of the thresholded modes. The effectiveness of the proposed scheme is illustrated on synthetic and real signals and the results compared to those of methods reported recently.
- 14 November 2014
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Signal Processing Letters
- Vol. 22 (10709908) , 1
On the behavior of EMD and MEMD in presence of symmetric $\alpha$-stable noise
Empirical Mode Decomposition (EMD) and its extended versions such as Multivariate EMD (MEMD) are data-driven techniques that represent nonlinear and non-stationary data as a sum of a finite zero-mean AM-FM components referred to as Intrinsic Mode Functions (IMFs). The aim of this work is to analyze the behavior of EMD and MEMD in stochastic situations involving non-Gaussian noise, more precisely, we examine the case of Symmetric α-Stable (SαS) noise. We report numerical experiments supporting the claim that both EMD and MEMD act, essentially, as filter banks on each channel of the input signal in the case of SαS noise. Reported results show that, unlike EMD, MEMD has the ability to align common frequency modes across multiple channels in same index IMFs. Further, simulations show that, contrary to EMD, for MEMD the stability property is well satisfied for the modes of lower indices and this result is exploited for the estimation of the stability index of the SαS input signal.
- 14 November 2014
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Signal Processing Letters
- Vol. 22 (10709908) , 1
On the behavior of EMD and MEMD in presence of symmetric $\alpha$-stable noise
Empirical Mode Decomposition (EMD) and its extended versions such as Multivariate EMD (MEMD) are data-driven techniques that represent nonlinear and non-stationary data as a sum of a finite zero-mean AM-FM components referred to as Intrinsic Mode Functions (IMFs). The aim of this work is to analyze the behavior of EMD and MEMD in stochastic situations involving non-Gaussian noise, more precisely, we examine the case of Symmetric α-Stable (SαS) noise. We report numerical experiments supporting the claim that both EMD and MEMD act, essentially, as filter banks on each channel of the input signal in the case of SαS noise. Reported results show that, unlike EMD, MEMD has the ability to align common frequency modes across multiple channels in same index IMFs. Further, simulations show that, contrary to EMD, for MEMD the stability property is well satisfied for the modes of lower indices and this result is exploited for the estimation of the stability index of the SαS input signal.
- 1 January 2014
- Published by Acoustical Society of America (ASA) in The Journal of the Acoustical Society of America
- Vol. 135 (1) , 451-459
Speech enhancement using empirical mode decomposition and the Teager–Kaiser energy operator
In this paper a speech denoising strategy based on time adaptive thresholding of intrinsic modes functions (IMFs) of the signal, extracted by empirical mode decomposition (EMD), is introduced. The denoised signal is reconstructed by the superposition of its adaptive thresholded IMFs. Adaptive thresholds are estimated using the Teager–Kaiser energy operator (TKEO) of signal IMFs. More precisely, TKEO identifies the type of frame by expanding differences between speech and non-speech frames in each IMF. Based on the EMD, the proposed speech denoising scheme is a fully data-driven approach. The method is tested on speech signals with different noise levels and the results are compared to EMD-shrinkage and wavelet transform (WT) coupled with TKEO. Speech enhancement performance is evaluated using output signal to noise ratio (SNR) and perceptual evaluation of speech quality (PESQ) measure. Based on the analyzed speech signals, the proposed enhancement scheme performs better than WT-TKEO and EMD-shrinkage approaches in terms of output SNR and PESQ. The noise is greatly reduced using time-adaptive thresholding than universal thresholding. The study is limited to signals corrupted by additive white Gaussian noise.
- 1 January 2014
- Published by Acoustical Society of America (ASA) in The Journal of the Acoustical Society of America
- Vol. 135 (1) , 451-459
Speech enhancement using empirical mode decomposition and the Teager–Kaiser energy operator
In this paper a speech denoising strategy based on time adaptive thresholding of intrinsic modes functions (IMFs) of the signal, extracted by empirical mode decomposition (EMD), is introduced. The denoised signal is reconstructed by the superposition of its adaptive thresholded IMFs. Adaptive thresholds are estimated using the Teager–Kaiser energy operator (TKEO) of signal IMFs. More precisely, TKEO identifies the type of frame by expanding differences between speech and non-speech frames in each IMF. Based on the EMD, the proposed speech denoising scheme is a fully data-driven approach. The method is tested on speech signals with different noise levels and the results are compared to EMD-shrinkage and wavelet transform (WT) coupled with TKEO. Speech enhancement performance is evaluated using output signal to noise ratio (SNR) and perceptual evaluation of speech quality (PESQ) measure. Based on the analyzed speech signals, the proposed enhancement scheme performs better than WT-TKEO and EMD-shrinkage approaches in terms of output SNR and PESQ. The noise is greatly reduced using time-adaptive thresholding than universal thresholding. The study is limited to signals corrupted by additive white Gaussian noise.
- 19 August 2013
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 63 (1) , 27-34
EMD-Based Filtering Using Similarity Measure Between Probability Density Functions of IMFs
This paper introduces a new signal-filtering, which combines the empirical mode decomposition (EMD) and a similarity measure. A noisy signal is adaptively broken down into oscillatory components called intrinsic mode functions by EMD followed by an estimation of the probability density function (pdf) of each extracted mode. The key idea of this paper is to make use of partial reconstruction, the relevant modes being selected on the basis of a striking similarity between the pdf of the input signal and that of each mode. Different similarity measures are investigated and compared. The obtained results, on simulated and real signals, show the effectiveness of the pdf-based filtering strategy for removing both white Gaussian and colored noises and demonstrate its superior performance over partial reconstruction approaches reported in the literature.
- 19 August 2013
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 63 (1) , 27-34
EMD-Based Filtering Using Similarity Measure Between Probability Density Functions of IMFs
This paper introduces a new signal-filtering, which combines the empirical mode decomposition (EMD) and a similarity measure. A noisy signal is adaptively broken down into oscillatory components called intrinsic mode functions by EMD followed by an estimation of the probability density function (pdf) of each extracted mode. The key idea of this paper is to make use of partial reconstruction, the relevant modes being selected on the basis of a striking similarity between the pdf of the input signal and that of each mode. Different similarity measures are investigated and compared. The obtained results, on simulated and real signals, show the effectiveness of the pdf-based filtering strategy for removing both white Gaussian and colored noises and demonstrate its superior performance over partial reconstruction approaches reported in the literature.
- 1 April 2012
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 3845, 1-6
Floating Costa Scheme with fractal structure for information embedding
Nowadays, multimedia data protection widely uses data hiding technology like Digital Watermarking or steganography. Among the large offer of watermarking techniques several ones are now based on the side information scheme proposed by Costa. These techniques mainly use scalar quantization for embedding a given information. Such solutions have good properties in terms of robustness and capacity but has lack of statistical transparency. In this paper we propose a new watermarking scheme based on a floating quantizer with a fractal structure. The aim of this study is to get rid of the unwanted fluctuations in the probability density function (PDF) of the watermarked signal, caused by the use of a scalar quantizer in the Scalar Costa Scheme approach (SCS). We present here the structure of a Fractal based Costa Scheme (FCS) and detail its performances. Afterwards, we compare the performances of our FCS proposed method with those obtained with the reference SCS. We show that we are able to reduce the statistical distortions obtained with this proposed scheme and keep good robustness properties.
- 1 April 2012
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 3845, 1-6
Floating Costa Scheme with fractal structure for information embedding
Nowadays, multimedia data protection widely uses data hiding technology like Digital Watermarking or steganography. Among the large offer of watermarking techniques several ones are now based on the side information scheme proposed by Costa. These techniques mainly use scalar quantization for embedding a given information. Such solutions have good properties in terms of robustness and capacity but has lack of statistical transparency. In this paper we propose a new watermarking scheme based on a floating quantizer with a fractal structure. The aim of this study is to get rid of the unwanted fluctuations in the probability density function (PDF) of the watermarked signal, caused by the use of a scalar quantizer in the Scalar Costa Scheme approach (SCS). We present here the structure of a Fractal based Costa Scheme (FCS) and detail its performances. Afterwards, we compare the performances of our FCS proposed method with those obtained with the reference SCS. We show that we are able to reduce the statistical distortions obtained with this proposed scheme and keep good robustness properties.
- 1 January 2012
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 21627843,p. 000292-000297
Emd-based filtering using the Hausdorff distance
This paper introduces a new signal-filtering method, which combines the empirical mode decomposition (EMD) and the Hausdorff distance (ℌD). A noisy signal is adaptively broken down into oscillatory components called intrinsic mode functions (IMFs) by EMD followed by an estimation of the probability density function (pdf) of each extracted mode. The filtered signal is obtained by performing a partial reconstruction using a selected set of empirical modes (termed relevant modes). The article focuses on an intuitive geometrical approach to identify the relevant modes based on the ℌD between the pdf of the noisy signal and that of each mode. Obtained results of the new filtering are compared to those of the EMD-CMSE (Consecutive Minimum Square Error) [1], the White Noise characteristics approach EMD-WN [2] and the Selection Criterion approach EMD-SC [3]. Dn different signals with varying signal-to-noise ratio values are used to validate the obtained results. The study is limited to signals corrupted by additive white Gaussian noise.
- 1 January 2012
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 21627843,p. 000292-000297
Emd-based filtering using the Hausdorff distance
This paper introduces a new signal-filtering method, which combines the empirical mode decomposition (EMD) and the Hausdorff distance (ℌD). A noisy signal is adaptively broken down into oscillatory components called intrinsic mode functions (IMFs) by EMD followed by an estimation of the probability density function (pdf) of each extracted mode. The filtered signal is obtained by performing a partial reconstruction using a selected set of empirical modes (termed relevant modes). The article focuses on an intuitive geometrical approach to identify the relevant modes based on the ℌD between the pdf of the noisy signal and that of each mode. Obtained results of the new filtering are compared to those of the EMD-CMSE (Consecutive Minimum Square Error) [1], the White Noise characteristics approach EMD-WN [2] and the Selection Criterion approach EMD-SC [3]. Dn different signals with varying signal-to-noise ratio values are used to validate the obtained results. The study is limited to signals corrupted by additive white Gaussian noise.