Proceedings of the 2nd Mediterranean Conference on Pattern Recognition and Artificial Intelligence
In this paper, we are interested in the processing of real underwater acoustics signals which hav... more In this paper, we are interested in the processing of real underwater acoustics signals which have many fields of application as detection and denoising by proposing an iterative and fast version of the CSMF, called CSMFST for Constrained Subspace Matched Filter Subspace Tracking, which is based on the Subspace Projection-type algorithms and we apply it to a nonstationary sinusoidal signal which is similar to a dolphin click and a nonwhite noise which is the sound of ocean waves.
The aim of this paper is the detection of a bioacoustic signal embedded in several noises such as... more The aim of this paper is the detection of a bioacoustic signal embedded in several noises such as sea noise and other bioacoustic signals (dolphins, sperm whales). All the signals are real world signals. Only second order statistics are use through the estimated correlation matrices of the signals. This paper proposes an extension of the Constrained Stochastic Matched Filter (CSMF) based on the optimization of the Signal to Noise Ratio after linear filtering. The approach proposed is a multicriteria one, merging three different versions of the CSMF, and is named Multicriteria CSMF (MCSMF). The objective is that the results obtained are better than the other methods, or at least equal to the best among the three. The results are provided on ROC curves and the method is compared to the classical method Stochastic Matched Filter (SMF).
Proceedings of the 10th International Symposium on Image and Signal Processing and Analysis, 2017
This paper introduces a new fast algorithm named CSMFST which estimates the p-dimensional optimal... more This paper introduces a new fast algorithm named CSMFST which estimates the p-dimensional optimal subspace, i.e. where the signal-to-noise ratio is maximized in the case of n-dimensional nonstationary signals. We assume that we treat both signal and noise which are characterized by their samples. This algorithm is an SP-type algorithm and uses the same principles as the Yet Another Subspace Tracking (YAST) algorithm when estimating the covariance matrices. At each step, it estimates a matrix which spans the optimal subspace.
Les problemes de detection et de classification de signaux aleatoires (transitoires, textures,. .... more Les problemes de detection et de classification de signaux aleatoires (transitoires, textures,. . . )sont d'une grande importance. Le filtre adapte stochastique sous contrainte se propose de maximiser le rapport signal-a-bruit dans un sous-espace de dimension donnee a priori pour atteindre ces objectifs. Il constitue une extension du filtre adapte et du filtre stochastique dont il partage la demarche. Cette approche est justifiee notamment quand les densites de probabilites des signaux sont inconnues. La methode proposee limite alors ses hypotheses a la connaissance des seules proprietes statisiques d'ordre deux des processus aleatoires en jeu, par le biais de leur matrice de covariance. La mise en equation conduit a resoudre une equation aux valeurs propres dont la matrice inconnue est parametree par le RSB, terme a maximiser qui s'ecrit sous la forme d'un rapport de sommes de formes quadratiques : un algorithme est propose dont on demontre qu'il converge vers l...
Proceedings of the 5th Wseas International Conference on Signal Processing Robotics and Automation, Feb 15, 2006
This paper deals with stochastic zero-mean signals detection. Second order statistical moments of... more This paper deals with stochastic zero-mean signals detection. Second order statistical moments of signals is assumed to be known. We propose an adaptated method based on a linear filtering under constraint optimizing a criterion based on different signal to noise ratios. This optimisation leads to the choice of a 1D subspace on which the different signals are projected. We also present simulations and results obtained with different signals.
L'objet de cet article est de proposer une nouvelle méthode de discrimination de deux textures bi... more L'objet de cet article est de proposer une nouvelle méthode de discrimination de deux textures bidimensionnelles en présence d'une texture dite 'blanche'. Elle permet le calcul de filtres optimaux adaptés à la texture jugée utile et désadaptés à (réjecteurs de) la texture indésirable. La méthode proposée se distingue des méthodes classiques en posant le problème sous l'angle de la maximisation d'un rapport signal à bruit qui conduit à l'obtention de filtres qui tendent à mettre en évidence une texture vis à vis d'une seconde. L'amélioration des performances est mise en évidence en présentant les probabilités de détection et de fausse alarme pour la méthode proposée et une méthode classique.
Cette étude se positionne dans l'hypothèse d'une superposition additive de deux processus stochas... more Cette étude se positionne dans l'hypothèse d'une superposition additive de deux processus stochastiques dont l'un d'intérêt noté S(t) est appelé signal et le second noté b(t) est le terme perturbateur ou bruit. On associe classiquement à ce mélange le rapport signal à bruit noté ici RSB= E[S 2 (t)]/E[b 2 (t)] déni comme le rapport des puissances. Dans les problèmes de détection on est amené à trouver un ltre linéaire qui maximise le nouveau RSB après ltrage; néanmoins la ltrée du signal ainsi obtenue peut s'écarter fortement de S(t). Dans les problèmes de ltrage, de compression, de reconstruction etc... cet inconvénient peut devenir majeur. On est alors amené à maximiser le RSB sous contrainte.
2014 21st IEEE International Conference on Electronics, Circuits and Systems (ICECS), 2014
The method proposed in this paper aims both to detect the events occurred on the overall current ... more The method proposed in this paper aims both to detect the events occurred on the overall current measurement and to estimate the features of the electric load underlying these events. This method is within the scope of the non-intrusive approach consisting in extracting information from the electric consumption only measured at the electrical switchboard.
Proceedings of the Eighth International Symposium on Signal Processing and Its Applications, 2005.
This paper deals with signal-to-noise ratio (SNR) maximization in a subspace. This criterion can ... more This paper deals with signal-to-noise ratio (SNR) maximization in a subspace. This criterion can be applied to estimation or detection of a stochastic signals embedded in additive noise. Both signal and noise are considered to be realizations of random functions whose statistics are assumed known up to second-order. The proposed method leads to a filter we call "constrained stochastic matched filter" (CSMF), optimal in the sense that it maximizes the SNR in a subspace whose dimension is chosen a priori. It is an extension of the stochastic matched filter (SMF), itself an extension of the matched filter.
Proceedings of the Tenth IEEE Workshop on Statistical Signal and Array Processing (Cat. No.00TH8496), 2000
We are interested in the separation of N independent sources recorded simultaneously by N receive... more We are interested in the separation of N independent sources recorded simultaneously by N receivers. The mixture is realized instantaneously through an unknown constant matrix M. When the spectral densities of the sources are different, several methods using second order moments have been proposed whose results are convincing. Nevertheless, these methods are no more efficient when their spectral densities are the same. We are interested in this special case where sources may even be white. The method we propose is based on the evaluation of second order moments estimated from extracted series of the observations. We talk of conditional second order moments. An iterative algorithm is proposed which calculates, at each step, a matrix K/sub i/ so that K/sub n/,K/sub n-1/,...,K/sub 1/M tends, when n increases, to DII, product of a diagonal matrix and a permutation matrix. We show that restrictive conditions on the probability distributions of the sources must be verified to assure the separation. In the two-dimensional case, we prove that the algorithm separates uniformly distributed sources, and that it does not separate Gaussian sources. The algorithm proposed is robust towards the number of sources; simulations with more than 20 uniformly distributed sources were successful.
This paper provides a novel passive underwater acoustic method to track a moving object called so... more This paper provides a novel passive underwater acoustic method to track a moving object called source or target, with the following constraints: the sensors location is fixed and imposed, and classical array processing techniques cannot be applied. The method proposed has been successfully used to track a surface vessel (2D problem) or an underwater target (3D problem). All the results presented have been obtained with real signals. The localization of the target requires the estimation of propagation delays, that means the duration between the instant of the signal emission and its reception on each receiver. The cross-correlation function is a suitable tool when the target is motionless, but needs to be extended to the ambiguity function when it is moving. The signal to noise ratio, the uniformity of power spectral density and the integration time are determining factors for the accuracy of the localization. We show that a whitening method and a Doppler compensation are necessary, and we propose a way to eliminate the significant problem of the reflected signals. Furthermore, a new configuration of the receivers is proposed, based on the idea of coupling receivers whose distance is chosen from experiment derived results. Furthermore, the algorithm proposed is susceptible to parallel implementation, thereby facilitating real-time uses. Experimental results with real time domain data are presented and compared to trajectories obtained by an active method. Résumé Cet article présente une nouvelle méthode d'écoute passive destinée à trajectographier un objet en mouvement appelé la "source" ou la "cible", avec les contraintes suivantes : le réseau de capteurs est fixe et imposé, et les méthodes classiques de traitement d'antenne ne peuvent pas s'appliquer. La méthode proposée ici a été utilisée avec succès pour trajectographier un bâtiment de surface (problème à deux dimensions) ou une cible sous-marine (problème à trois dimensions). Tous les résultats présentés ont été obtenus sur des signaux réels. Le calcul de la position de la cible nécessite l'estimation des temps de propagation, c'est-à-dire du temps écoulé entre l'instant d'émission du signal et sa réception sur chaque capteur. La fonction d'intercorrélation est un outil opportun lorsque la cible est immobile mais elle doit être étendue à la notion d'ambiguïté quand la source est en mouvement. Le rapport signal à bruit, l'uniformité de la densité spectrale de puissance et la durée d'intégration sont des facteurs déterminants pour la précision de la localisation. Nous montrons qu'une méthode de blanchiment et une compensation de l'effet Doppler sont nécessaires et nous proposons un moyen d'éliminer le problème des trajets multiples. Par ailleurs, une nouvelle configuration des capteurs est proposée, fondée sur l'idée d'associer des hydrophones séparés d'une distance déterminée de manière expérimentale. Des résultats de trajectographies sont présentés et comparés aux trajectoires obtenues par une méthode active.
This paper is concerned with the problem of blind separation of instantaneous mixture of sources ... more This paper is concerned with the problem of blind separation of instantaneous mixture of sources (BSS) which has been addressed in many ways. When power spectral densities of the sources are different, methods using second-order statistics are sufficient to solve this problem. Otherwise these methods fail and others (higher-order statistics,...) must be used. In this paper, we propose an iterative method to process the case of sources with the same power spectral density. This method is based on an evaluation of conditional first and second-order statistics only. Restrictions on characteristics of sources are given to reach a solution, and proofs of convergence of the algorithm are provided for particular cases of probability density functions. Robustness of this algorithm with respect to the number of sources is shown through computer simulations. A particular case of sources which have a probability density function with unbounded domain of definition is described; here the algorithm does not lead directly to a separation state, but to an a priori known mixture state. Finally, prospects of links with contrast functions are mentioned, with a possible generalization of them based on results obtained with particular sources.
In this paper, we propose a novel approach to perform detection of stochastic signals embedded in... more In this paper, we propose a novel approach to perform detection of stochastic signals embedded in an additive random noise. Both signal and noise are considered to be realizations of zero mean random processes whose only secondorder statistics are known (their covariance matrices). The method proposed, called Constrained Stochastic Matched Filter (CSMF), is an extension of the Stochastic Matched Filter itself derived from the Matched Filter. The CSMF is optimal in the sense that it maximizes the Signalto-Noise Ratio in a subspace whose dimension is fixed a priori. In this paper, after giving the reasons of our approach, we show that there is neither obvious nor analytic solution to the problem expressed. Then an algorithm, which is proved to converge, is proposed to obtain the optimal solution. The evaluation of the performance is completed through estimation of Receiver Operating Characteristic curves. Experiments on real signals show the improvement brought by this method and thus its significance.
Proceedings of the 2nd Mediterranean Conference on Pattern Recognition and Artificial Intelligence
In this paper, we are interested in the processing of real underwater acoustics signals which hav... more In this paper, we are interested in the processing of real underwater acoustics signals which have many fields of application as detection and denoising by proposing an iterative and fast version of the CSMF, called CSMFST for Constrained Subspace Matched Filter Subspace Tracking, which is based on the Subspace Projection-type algorithms and we apply it to a nonstationary sinusoidal signal which is similar to a dolphin click and a nonwhite noise which is the sound of ocean waves.
The aim of this paper is the detection of a bioacoustic signal embedded in several noises such as... more The aim of this paper is the detection of a bioacoustic signal embedded in several noises such as sea noise and other bioacoustic signals (dolphins, sperm whales). All the signals are real world signals. Only second order statistics are use through the estimated correlation matrices of the signals. This paper proposes an extension of the Constrained Stochastic Matched Filter (CSMF) based on the optimization of the Signal to Noise Ratio after linear filtering. The approach proposed is a multicriteria one, merging three different versions of the CSMF, and is named Multicriteria CSMF (MCSMF). The objective is that the results obtained are better than the other methods, or at least equal to the best among the three. The results are provided on ROC curves and the method is compared to the classical method Stochastic Matched Filter (SMF).
Proceedings of the 10th International Symposium on Image and Signal Processing and Analysis, 2017
This paper introduces a new fast algorithm named CSMFST which estimates the p-dimensional optimal... more This paper introduces a new fast algorithm named CSMFST which estimates the p-dimensional optimal subspace, i.e. where the signal-to-noise ratio is maximized in the case of n-dimensional nonstationary signals. We assume that we treat both signal and noise which are characterized by their samples. This algorithm is an SP-type algorithm and uses the same principles as the Yet Another Subspace Tracking (YAST) algorithm when estimating the covariance matrices. At each step, it estimates a matrix which spans the optimal subspace.
Les problemes de detection et de classification de signaux aleatoires (transitoires, textures,. .... more Les problemes de detection et de classification de signaux aleatoires (transitoires, textures,. . . )sont d'une grande importance. Le filtre adapte stochastique sous contrainte se propose de maximiser le rapport signal-a-bruit dans un sous-espace de dimension donnee a priori pour atteindre ces objectifs. Il constitue une extension du filtre adapte et du filtre stochastique dont il partage la demarche. Cette approche est justifiee notamment quand les densites de probabilites des signaux sont inconnues. La methode proposee limite alors ses hypotheses a la connaissance des seules proprietes statisiques d'ordre deux des processus aleatoires en jeu, par le biais de leur matrice de covariance. La mise en equation conduit a resoudre une equation aux valeurs propres dont la matrice inconnue est parametree par le RSB, terme a maximiser qui s'ecrit sous la forme d'un rapport de sommes de formes quadratiques : un algorithme est propose dont on demontre qu'il converge vers l...
Proceedings of the 5th Wseas International Conference on Signal Processing Robotics and Automation, Feb 15, 2006
This paper deals with stochastic zero-mean signals detection. Second order statistical moments of... more This paper deals with stochastic zero-mean signals detection. Second order statistical moments of signals is assumed to be known. We propose an adaptated method based on a linear filtering under constraint optimizing a criterion based on different signal to noise ratios. This optimisation leads to the choice of a 1D subspace on which the different signals are projected. We also present simulations and results obtained with different signals.
L'objet de cet article est de proposer une nouvelle méthode de discrimination de deux textures bi... more L'objet de cet article est de proposer une nouvelle méthode de discrimination de deux textures bidimensionnelles en présence d'une texture dite 'blanche'. Elle permet le calcul de filtres optimaux adaptés à la texture jugée utile et désadaptés à (réjecteurs de) la texture indésirable. La méthode proposée se distingue des méthodes classiques en posant le problème sous l'angle de la maximisation d'un rapport signal à bruit qui conduit à l'obtention de filtres qui tendent à mettre en évidence une texture vis à vis d'une seconde. L'amélioration des performances est mise en évidence en présentant les probabilités de détection et de fausse alarme pour la méthode proposée et une méthode classique.
Cette étude se positionne dans l'hypothèse d'une superposition additive de deux processus stochas... more Cette étude se positionne dans l'hypothèse d'une superposition additive de deux processus stochastiques dont l'un d'intérêt noté S(t) est appelé signal et le second noté b(t) est le terme perturbateur ou bruit. On associe classiquement à ce mélange le rapport signal à bruit noté ici RSB= E[S 2 (t)]/E[b 2 (t)] déni comme le rapport des puissances. Dans les problèmes de détection on est amené à trouver un ltre linéaire qui maximise le nouveau RSB après ltrage; néanmoins la ltrée du signal ainsi obtenue peut s'écarter fortement de S(t). Dans les problèmes de ltrage, de compression, de reconstruction etc... cet inconvénient peut devenir majeur. On est alors amené à maximiser le RSB sous contrainte.
2014 21st IEEE International Conference on Electronics, Circuits and Systems (ICECS), 2014
The method proposed in this paper aims both to detect the events occurred on the overall current ... more The method proposed in this paper aims both to detect the events occurred on the overall current measurement and to estimate the features of the electric load underlying these events. This method is within the scope of the non-intrusive approach consisting in extracting information from the electric consumption only measured at the electrical switchboard.
Proceedings of the Eighth International Symposium on Signal Processing and Its Applications, 2005.
This paper deals with signal-to-noise ratio (SNR) maximization in a subspace. This criterion can ... more This paper deals with signal-to-noise ratio (SNR) maximization in a subspace. This criterion can be applied to estimation or detection of a stochastic signals embedded in additive noise. Both signal and noise are considered to be realizations of random functions whose statistics are assumed known up to second-order. The proposed method leads to a filter we call "constrained stochastic matched filter" (CSMF), optimal in the sense that it maximizes the SNR in a subspace whose dimension is chosen a priori. It is an extension of the stochastic matched filter (SMF), itself an extension of the matched filter.
Proceedings of the Tenth IEEE Workshop on Statistical Signal and Array Processing (Cat. No.00TH8496), 2000
We are interested in the separation of N independent sources recorded simultaneously by N receive... more We are interested in the separation of N independent sources recorded simultaneously by N receivers. The mixture is realized instantaneously through an unknown constant matrix M. When the spectral densities of the sources are different, several methods using second order moments have been proposed whose results are convincing. Nevertheless, these methods are no more efficient when their spectral densities are the same. We are interested in this special case where sources may even be white. The method we propose is based on the evaluation of second order moments estimated from extracted series of the observations. We talk of conditional second order moments. An iterative algorithm is proposed which calculates, at each step, a matrix K/sub i/ so that K/sub n/,K/sub n-1/,...,K/sub 1/M tends, when n increases, to DII, product of a diagonal matrix and a permutation matrix. We show that restrictive conditions on the probability distributions of the sources must be verified to assure the separation. In the two-dimensional case, we prove that the algorithm separates uniformly distributed sources, and that it does not separate Gaussian sources. The algorithm proposed is robust towards the number of sources; simulations with more than 20 uniformly distributed sources were successful.
This paper provides a novel passive underwater acoustic method to track a moving object called so... more This paper provides a novel passive underwater acoustic method to track a moving object called source or target, with the following constraints: the sensors location is fixed and imposed, and classical array processing techniques cannot be applied. The method proposed has been successfully used to track a surface vessel (2D problem) or an underwater target (3D problem). All the results presented have been obtained with real signals. The localization of the target requires the estimation of propagation delays, that means the duration between the instant of the signal emission and its reception on each receiver. The cross-correlation function is a suitable tool when the target is motionless, but needs to be extended to the ambiguity function when it is moving. The signal to noise ratio, the uniformity of power spectral density and the integration time are determining factors for the accuracy of the localization. We show that a whitening method and a Doppler compensation are necessary, and we propose a way to eliminate the significant problem of the reflected signals. Furthermore, a new configuration of the receivers is proposed, based on the idea of coupling receivers whose distance is chosen from experiment derived results. Furthermore, the algorithm proposed is susceptible to parallel implementation, thereby facilitating real-time uses. Experimental results with real time domain data are presented and compared to trajectories obtained by an active method. Résumé Cet article présente une nouvelle méthode d'écoute passive destinée à trajectographier un objet en mouvement appelé la "source" ou la "cible", avec les contraintes suivantes : le réseau de capteurs est fixe et imposé, et les méthodes classiques de traitement d'antenne ne peuvent pas s'appliquer. La méthode proposée ici a été utilisée avec succès pour trajectographier un bâtiment de surface (problème à deux dimensions) ou une cible sous-marine (problème à trois dimensions). Tous les résultats présentés ont été obtenus sur des signaux réels. Le calcul de la position de la cible nécessite l'estimation des temps de propagation, c'est-à-dire du temps écoulé entre l'instant d'émission du signal et sa réception sur chaque capteur. La fonction d'intercorrélation est un outil opportun lorsque la cible est immobile mais elle doit être étendue à la notion d'ambiguïté quand la source est en mouvement. Le rapport signal à bruit, l'uniformité de la densité spectrale de puissance et la durée d'intégration sont des facteurs déterminants pour la précision de la localisation. Nous montrons qu'une méthode de blanchiment et une compensation de l'effet Doppler sont nécessaires et nous proposons un moyen d'éliminer le problème des trajets multiples. Par ailleurs, une nouvelle configuration des capteurs est proposée, fondée sur l'idée d'associer des hydrophones séparés d'une distance déterminée de manière expérimentale. Des résultats de trajectographies sont présentés et comparés aux trajectoires obtenues par une méthode active.
This paper is concerned with the problem of blind separation of instantaneous mixture of sources ... more This paper is concerned with the problem of blind separation of instantaneous mixture of sources (BSS) which has been addressed in many ways. When power spectral densities of the sources are different, methods using second-order statistics are sufficient to solve this problem. Otherwise these methods fail and others (higher-order statistics,...) must be used. In this paper, we propose an iterative method to process the case of sources with the same power spectral density. This method is based on an evaluation of conditional first and second-order statistics only. Restrictions on characteristics of sources are given to reach a solution, and proofs of convergence of the algorithm are provided for particular cases of probability density functions. Robustness of this algorithm with respect to the number of sources is shown through computer simulations. A particular case of sources which have a probability density function with unbounded domain of definition is described; here the algorithm does not lead directly to a separation state, but to an a priori known mixture state. Finally, prospects of links with contrast functions are mentioned, with a possible generalization of them based on results obtained with particular sources.
In this paper, we propose a novel approach to perform detection of stochastic signals embedded in... more In this paper, we propose a novel approach to perform detection of stochastic signals embedded in an additive random noise. Both signal and noise are considered to be realizations of zero mean random processes whose only secondorder statistics are known (their covariance matrices). The method proposed, called Constrained Stochastic Matched Filter (CSMF), is an extension of the Stochastic Matched Filter itself derived from the Matched Filter. The CSMF is optimal in the sense that it maximizes the Signalto-Noise Ratio in a subspace whose dimension is fixed a priori. In this paper, after giving the reasons of our approach, we show that there is neither obvious nor analytic solution to the problem expressed. Then an algorithm, which is proved to converge, is proposed to obtain the optimal solution. The evaluation of the performance is completed through estimation of Receiver Operating Characteristic curves. Experiments on real signals show the improvement brought by this method and thus its significance.
Uploads
Papers by bruno borloz