Sparse deconvolution using adaptive mixed-Gaussian models

نویسندگان

  • Ignacio Santamaría
  • Carlos Pantaleón
  • Antonio Artés-Rodríguez
چکیده

In this paper we present a new algorithm to recover a sparse signal from a noisy register. The algorithm assumes a new prior distribution for the sparse signal that consists of a mixture of a narrow and a broad Gaussian both with zero mean. A penalty term which favors solutions driven from this model is added to the usual error cost function and the resultant global cost function is minimized by means of a gradient-type algorithm. A condition is derived for the step-size parameter in order to ensure convergence. In the paper we also propose a method (based on the Expectation-Maximization algorithm) to update the mixture parameters. The estimation of the sparse signal and the optimization of the Gaussian mixture are combined in the proposed algorithm: in each iteration a new signal estimate and a new model (which approximates the distribution of the new estimate) are obtained. In this way, the proposed method can be used without any statistical knowledge about the signal. Simulation experiments show that the accuracy of the proposed method is competitive with classical statistical detectors with a lower computational load. In diesem Beitrag ptisentieren wir einen neuen Algorithmus, urn ein diinnverteiltes Signal aus dem Rauschen zuriickzugewinnen. Der Algorithmus nimmt eine neue urspriingliche Verteilung tiir das diinnverteilte Signal an, welches aus einer Mischung aus einem schmalen und einem breiten GauBverteilten Signal besteht, beide ohne Gleichanteil. Ein Strafierm, welcher Lijsungen begiinstigt, die von diesem Model1 abgeleitet werden, wird zu der iiblichen Fehler-Kostenfunktion hinzuaddiert und die resultierende globale Kostenfunktion wird mittels eines Gradienten-Algorithmus minimiert. Eine Bedingung fir den Schrittgr6l3en-Parameter wird hergeleitet, urn Konvergenz sicherzustellen. Im Beitrag schlagen wir such eine Methode vor (gegriindet auf dem Erwartungs-Maximienmgs-Algorithmus), urn die Mischungsparameter zu aktualisieren. Die Sch&ung des diinnverteilten Signals und die Optimierung der GauDschen Mischung sind in dem vorgeschlagenen Algorithmus verbunden: in jeder Iteration wird ein neuer SignalschLzwert und ein neues Model1 (welches die Verteilung des neuen Schlzwertes annghert) erhalten. Auf diesem Weg kann die vorgeschlagene Methode ohne statistisches Wissen iiber das Signal angewendet werden. Simulationsexperimente zeigen, da13 die Genauigkeit der vorgeschlagenen Methode vergleichbar ist mit klassischen statistischen Detektoren bei einer niedrigeren Rechenlast. Nous presentons dans cet article un algorithme nouveau pour le recouvrement d’un signal Ctalt a partir d’un registre bruit& L’algorithme suppose une nouvelle distribution a priori pour le signal Ctal(t sous la forme d’un mClange d’une gaussienne 6.troite et d’une gaussienne large, toutes deux de moyenne nulle. Un terme de ptnalitk qui favorise les solutions correspondant *Tel.: 34-42-201552; fax: 34-42-201488; e-mail: [email protected]. 0165-1684/96/$15.00 @ 1996 Elsevier Science B.V. All rights reserved PIISO165-1684(96)00105-3 162 I. Suntamariu-Caballero et al. /Signal Processing 54 (1996) 161-172 ti ce modele est addition& a la fonction de codt d’erreur habituelle et la fonction de codt globale resultante est minimiste a l’aide d’un algorithme de type gradient. Une condition sur le parametre de pas d’iteration est donnee pour garantir la convergence. Dans cet article nous proposons Cgalement une methode (basee sur l’algorithme d’EspCrance-Maximisation) pour la mise b jour des parametres du melange. L’estimation du signal &ale et l’optimisation du melange gaussien sont combinees dans l’algorithme propose: a chaque iteration une nouvelle estimee du signal et un nouveau modele (qui approxime la distribution de la nouvelle estimee) sont obtenus. De cette man&e, la methode proposee peut etre utilisee sans aucune connaissance statistique sur le signal. Des simulations montrent que la precision de la methode proposee est competitive vis-a-vis des detecteurs statistiques classiques tout en demandant une charge de calcul plus faible.

برای دانلود رایگان متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

An adaptive sparse deconvolution method for distinguishing the overlapping echoes of ultrasonic guided waves for pipeline crack inspection

In guided wave pipeline inspection, echoes reflected from closely spaced reflectors generally overlap, meaning useful information is lost. To solve the overlapping problem, sparse deconvolution methods have been developed in the past decade. However, conventional sparse deconvolution methods have limitations in handling guided wave signals, because the input signal is directly used as the proto...

متن کامل

Parameter Estimation in Spatial Generalized Linear Mixed Models with Skew Gaussian Random Effects using Laplace Approximation

 Spatial generalized linear mixed models are used commonly for modelling non-Gaussian discrete spatial responses. We present an algorithm for parameter estimation of the models using Laplace approximation of likelihood function. In these models, the spatial correlation structure of data is carried out by random effects or latent variables. In most spatial analysis, it is assumed that rando...

متن کامل

Sparse Deconvolution Using Support Vector Machines

Sparse deconvolution is a classical subject in digital signal processing, having many practical applications. Support vector machine (SVM) algorithms show a series of characteristics, such as sparse solutions and implicit regularization, which make them attractive for solving sparse deconvolution problems. Here, a sparse deconvolution algorithm based on the SVM framework for signal processing i...

متن کامل

Non-Uniform Camera Shake Removal Using a Spatially-Adaptive Sparse Penalty

Typical blur from camera shake often deviates from the standard uniform convolutional assumption, in part because of problematic rotations which create greater blurring away from some unknown center point. Consequently, successful blind deconvolution for removing shake artifacts requires the estimation of a spatiallyvarying or non-uniform blur operator. Using ideas from Bayesian inference and c...

متن کامل

Bayesian Blind Deconvolution with General Sparse Image Priors

We present a general method for blind image deconvolution using Bayesian inference with super-Gaussian sparse image priors. We consider a large family of priors suitable for modeling natural images, and develop the general procedure for estimating the unknown image and the blur. Our formulation includes a number of existing modeling and inference methods as special cases while providing additio...

متن کامل

A recursive soft-decision approach to blind image deconvolution

This paper presents a new approach to blind image deconvolution based on soft-decision blur identification and hierarchical neural networks. Traditional blind algorithms require a hard-decision on whether the blur satisfies a parametric form before their formulations. As the blurring function is usually unknown a priori, this precondition inhibits the incorporation of parametric blur knowledge ...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

عنوان ژورنال:
  • Signal Processing

دوره 54  شماره 

صفحات  -

تاریخ انتشار 1996