Search

International audience; Unmixing is a ubiquitous task in hyperspectral image analysis which consists in jointly extracting typical spectral signatures and estimating their respective proportions in the voxels, providing an explicit spatial mapping of these elementary signatures over the observed scene. Inspired by this approach, this paper aims at proposing a new framework for analyzing dynamic positron emission tomography (PET) images. More precisely, a PET-dedicated mixing model and an associated unmixing algorithm are derived to jointly estimate time-activity curves (TAC) characterizing each type of tissues, and the proportions of those tissues in the voxels of the imaged brain. In particular, the TAC corresponding to the specific binding class is expected to be voxel-wise dependent. The proposed approach allows this intrinsic spatial variability to be properly modeled, mitigated and quantified. Finally, the main contributions of the paper are twofold: first, we demonstrate that the unmixing concept is an appropriate analysis tool for dynamic PET images; and second, we propose a novel unmixing algorithm allowing for variability, which significantly improves the analysis and interpretation of dynamic PET images when compared with state-of-the-art unmixing algorithms.

Over the past decades, privacy-preservation has received considerable attention, not only as a consequence of regulations such as the General Data Protection Regulation in the EU, but also from the fact that people are more concerned about data abuse as the world is becoming increasingly digitized. In this paper we propose a convex optimization-based subspace perturbation approach to solve privacy-preserving distributed least squares problems. Based on the primal-dual method of multipliers, the introduced dual variables will only converge in a subspace determined by the graph topology and do not converge in its orthogonal complement. We, therefore, propose to exploit this property for privacy-preservation by using the nonconverging part of the dual variables to perturb the private data, thereby protecting it from being revealed. Moreover, we prove that the proposed approach is secure under both eavesdropping and passive adversaries. Computer simulations are conducted to demonstrate the benefits of the proposed approach through its convergence properties and accuracy.

The use of wireless acoustic sensor networks (WASNs) has received increased attention over the last decade. The advantages of WASNs over stand-alone multi-microphone devices are that the microphone array is not anymore limited by the dimensions of a single device, and that microphones can be placed at arbitrary locations. One of the disadvantages, however, is that for many applications, like beamforming, the clocks of all devices in the network need to be synchronised and that the microphone gains need to be equalised. In this paper we will prove that a specific class of beamformers is clock-offset and gain mismatch invariant. The parameters for these beamformers (acoustic transfer function and power spectral density matrices) can be estimated directly from the uncalibrated microphone signals, instead of first synchronising the clocks and equalising the gains and then estimating them. The resulting beamformers are applied to the non-calibrated microphone signals. We will substantiate, by means of computer simulations, that the proposed approach gives identical results compared to the setup where microphone signals are first calibrated, so that clock-offset compensation and microphone gain equalisation becomes unnecessary.

Atrial fibrillation (AF) is the most common arrhythmia in the heart. Two main types of AF are defined as paroxysmal and persistent. In this paper, we present a method to discriminate between the characteristics of paroxysmal and persistent using tensor decompositions of a multi-channel electrocardiogram (ECG) signal. For this purpose, ECG signals are segmented by applying a Hilbert transform on the thresholded signal. Dynamic time warping is used to align the separated segments of each channel and then a tensor is constructed with three dimensions as time, heartbeats and channels. A Canonical polyadic decomposition with rank 2 is computed from this tensor and the resulting loading vectors describe the characteristics of paroxysmal and persistent AF in these three dimensions. The time loading vector reveals the pattern of a single P wave or abnormal AF patterns. The heartbeat loading vector shows whether the pattern is present or absent in a specific beat. The results can be used to distinguish between the patterns in paroxysmal AF and persistent AF.

In various applications within signal processing, system identification, pattern recognition, and scientific computing, the canonical polyadic decomposition (CPD) of a higher-order tensor is only known via general linear measurements. In this paper, we show that the computation of such a CPD can be reformulated as a sum of CPDs with linearly constrained factor matrices by assuming that the measurement matrix can be approximated by a sum of a (small) number of Kronecker products. By properly exploiting the hypothesized structure, we can derive an efficient non-linear least squares algorithm, allowing us to tackle large-scale problems.

In this paper, we address the singing voice separation problem and propose a novel unsupervised approach based on robust principal component analysis (RPCA) exploiting rank-1 constraint (CRPCA). RPCA is a recently proposed singing voice separation algorithm that can separate singing voice from the monaural recordings. Although RPCA has been successfully applied to singing voice separation task, it ignores the different characteristic values of singular value decomposition and computational complexity to minimize the nuclear norm for separating singing voice. Since rank-l constraint in the background music, as the background music has a large variation in richness than singing voice among different songs. Furthermore, rank-1 constraint can utilize a prior target rank to separate singing voice and background music from the mixture music signal. Accordingly, the proposed CRPCA method utilizes rank-1 constraint minimization of singular values in RPCA instead of minimizing the whole nuclear norm, which can not only describe the different values of singular value decomposition, but also the computation complexity is reduced. The experiment evaluation results reveal that CRPCA can achieve better separation performance than the previous methods, especially with regard to use time frequency masking on ccMixter and DSD100 datasets. In addition, the running time on CRPCA is shorter than others under the same conditions.

© EURASIP 2018. When signal coordination techniques can not eliminate all crosstalk in a digital subscriber line (DSL) system, competition for data rate among different users is strong. In such scenarios, employing a static resource allocation fails to capitalize on the time dependent nature of the traffic carried by the DSL network. An alternative approach is adaptive resource allocation, consisting of dividing time into slots of short duration and using a different resource allocation in each slot. A cross-layer scheduler then decides on the resource allocation for each time slot by solving a network utility maximization (NUM) problem. For many DSL systems however, this NUM problem is non-convex and solving it is NP-hard. This paper presents a fast algorithm for finding a local solution to the NUM problem, which is referred to as NUM-DSB. The algorithm is able to handle many DSL deployment scenarios, and is applicable regardless of the utility function's properties. ispartof: pages:792-796 ispartof: Proc. of the 26th European Signal Processing Conference vol:2018-September pages:792-796 ispartof: 26th European Signal Processing Conference location:Rome, Italy date:3 Sep - 7 Sep 2018 status: published

A recent trend in color image processing combines the quaternion algebra with dictionary learning methods. This paper aims to present a generalization of the quaternion dictionary learning method by using the octonion algebra. The octonion algebra combined with dictionary learning methods is well suited for representation of multispectral images with up to 7 color channels. Opposed to the classical dictionary learning techniques that treat multispectral images by concatenating spectral bands into a large monochrome image, we treat all the spectral bands simultaneously. Our approach leads to better preservation of color fidelity in true and false color images of the reconstructed multispectral image. To show the potential of the octonion based model, experiments are conducted for image reconstruction and denoising of color images as well as of extensively used Landsat 7 images.

The emergence of efficient algorithms in variational and Bayesian frameworks braught significant advances to the field of inverse problems. However, such problems remain challenging when the observation operator is not perfectly known. In this paper we propose a Bayesian Plug-and-Play (PP) algorithm for solving a wide range of inverse problems where the signal/image is sparse in the original domain and the observation operator has to be estimated. The principle consists of plugging the prior considered for the target observation operator and keep using the same algorithm. The proposed method relies on a generic proximal non-smooth sampling scheme. This genericity makes the proposed algorithm novel in the sense that it can be used to solve a wide range or inverse problems. Our method is illustrated on a deblurring problem with unknown blur operator where promising results are obtained.

International audience; Bayesian sparse signal recovery has been widely investigated during the last decade due to its ability to automatically estimate regularization parameters. Prior based on mixtures of Bernoulli and continuous distributions have recently been used in a number of recent works to model the target signals , often leading to complicated posteriors. Inference is therefore usually performed using Markov chain Monte Carlo algorithms. In this paper, a Bernoulli-generalized Gaussian distribution is used in a sparse Bayesian regularization framework to promote a two-level flexible sparsity. Since the resulting conditional posterior has a non-differentiable energy function , the inference is conducted using the recently proposed non-smooth Hamiltonian Monte Carlo algorithm. Promising results obtained with synthetic data show the efficiency of the proposed regularization scheme.

International audience; In this paper, we present a chaos-based decoupled multiuser ranging (DEMR) estimator for multiuser DS-UWB ranging system. In the DEMR estimator, users are decoupled by the knowledge of all the users' limited number of data bits. Then, the ranging performance of each user mainly depends on the non-cyclic autocorrelation property of the spreading code. Based on this property, we improve DEMR estimator by using the selected binary chaotic sequences instead of the Gold sequences in order to increase the system capacity and to improve the ranging accuracy. Simulations in CM1 channel show that the chaos-based DEMR estimator is quite near-far resistant and achieves a noticeable ranging accuracy even in a heavily loaded system. Compared with using Gold sequences, chaos-based DEMR not only works with more users than full load of Gold sequences but also improves the ranging accuracy especially under low SNR condition.