Une brève histoire de l'IRCCyN
Cellule parachèvement des composites
Image and video quality assessment
Network simulation and emulation
NOLIACPA et SAC
Nouvelle ligne d'assemblage
QoE 3D, image and video
Simulateur de conduite
UGV et NTC
Visual attention and eye tracking
Watermarking and Security
Présentation de l'équipe ADTSI
Thématiques de recherche
Collaborations et Thèses
Publications de l'équipe
Analyse et Décision en Traitement du Signal et de l'Image
Résultats : de 11 à 20 (295 au total)
Estimating fat, paste and gas in a proving Danish paste by MRI – Description of the method and evaluation of its performances (bias and accuracy, sensitivity threshold)
Guylaine Collewet, Vincent Perrouin, Cécile Deligny, Jérôme Idier, Tiphaine Lucas,
Regularisation, Proving, Food, Quantification, Image restoration, Noise, MRI, Conjugate gradient, Puff pastry, Spin echo
This paper presents a method to characterize the development of the structures of puff pastries during proving using MRI. Since the resolution is large, each pixel contains an unknown proportion of three components: fat, paste and gas. The signal to noise ratio is low since gas which reaches 80% at the end of proving gives no signal. The signal is the sum of reference signals, corresponding to pixels filled with one component, weighted by the proportions. The reference signals were supposed to be known. We adopted an edge-preserving approach based on the minimization of a penalized least-square criterion. This criterion is the weighted sum of a term accounting for the fidelity to data and a regularization term. The minimization of the criterion is based on a non-linear conjugate gradient algorithm. The settings of the weights of the two terms is based on simulations. Then simulation results are presented. The mean error was similar with or without regularization and depended on the components and their proportion (less that 1% up to 6%). Fat and gas proportions were overestimated, paste proportion was underestimated. The dispersion of the results was lower with regularization (from 0.3% up to 1.5 %). Monte-Carlo simulations showed that these results were not influenced very much by the uncertainty on the reference signals at the end of the proving. Larger uncertainties were found at the beginning of proving. We showed that the regularization of the solutions did improve the visualization of the structures confirming the interest of this approach.We also found that layers down to 40 microns thick and bubbles of which size exceeded 2.5mm could be easily distinguished in fat and gas proportion maps respectively. Experiments on genuine MRI images of Danish paste confirmed the results obtained with the simulation study. We were able to validate the method at the scale of the Danish paste by observing the evolution of the sum of the fat proportion over time, as well as the evolution of the gas content compared with the evolution of the size of the pastry. This confirmed the possibility to use the method to study the proving of a Danish paste.
Estimation de l’état du conducteur pour la détection de distraction
Ablamvi Ameyoe, Philippe Chevrel, Eric Le Carpentier, Hervé Illy,
Présentation au Groupe de Travail Automatique et Automobile (GTAA). 2016. 〈hal-01317989〉
Modélisation conducteur, Estimation, Observateur, Détection distraction, ADAS
Sliding mode observer design for the road curvature estimation in Traffic Jam Pilot system
Joan Davins-Valldaura, Plestan Franck, Saïd Moussaoui, Guillermo Pita Gil,
14th International Workshop on Variable Structure Systems , Jun 2016, Nanjing, China. 2016. 〈hal-01309577〉
The road curvature is an important characteristic, specially for systems tracking a target trajectory. However, in some cases, this value is not available and its dynamics is unknown and has to be estimated to be used by control. By considering vehicle lateral dynamics, the steering system and the Radar/Camera information, the standard Traffic Jam Pilot (TJP) system is presented. Current TJP solutions add the target road curvature in the state, which is typically estimated by a linear solution based on Kalman observer. A drawback of this method is that this class of observers is designed and tuned for several ranges of velocity, each range having its own observers gains. This feature strongly increases the amount of observer parameters to be tuned. The aim of this paper is to propose a new scheme of observation for TJP system, based on a nonlinear Sliding Mode observer, requiring only one set of gains. Such an observer solution is tested by simulation in different time-varying velocity conditions, curvature variation and types of roads. Another contribution, is the proposition of a recent automatically tuning procedure to optimize observer gains for any system with a large scale base. Finally, the road curvature estimation is used to discuss the performances obtained.
Improving the axial and lateral resolution of three-dimensional fluorescence microscopy using random speckle illuminations
A. Negash, S. Labouesse, Nicolas Sandeau, Marc Allain, Hugues Giovannini, Jérome Idier, Rainer Heintzmann, Patrick C. Chaumet, Kamal Belkebir, Anne Sentenac,
Journal of the Optical Society of America. A, Optics and image science, Optical Society of America, 2016, 33, pp.1089-1094. 〈10.1364/JOSAA.33.001089〉. 〈hal-01314573〉
We consider a fluorescence microscope in which several three-dimensional images of a sample are recorded for different speckle illuminations. We show, on synthetic data, that by summing the positive deconvolution of each speckle image, one obtains a sample reconstruction with axial and transverse resolutions that compare favorably to that of an ideal confocal microscope.
Algorithmes de type homotopiques pour la minimisation des moindres carrés régularisés par la "norme" $\ell_0$
Charles Soussen, Jérôme Idier, Junbo Duan, David Brie,
Cette communication concerne la conception d'algorithmes d'approximation parcimonieuse pour les problèmes inverses mal conditionnés. Les algorithmes heuristiques proposés sont conçus pour minimiser des critères mixtes 2-0 du type min_x J (x; λ) = || y − Ax ||^2 + λ || x ||_0. Ce sont des algorithmes gloutons " bidirectionnels " définis en tant qu'extensions d'Orthogonal Least Squares (OLS). Leur développement est motivé par le très bon comportement empirique d'OLS et de ses versions dérivées lorsque le dictionnaire A est une matrice mal conditionnée. Nous présentons dans un premier temps l'algorithme Single Best Replacement (SBR) pour minimiser J (x; λ) à λ fixé , en mettant en avant ses propriétés de convergence. Nous proposons ensuite deux algorithmes permettant de minimiser J pour un continuum de valeurs de λ, ce qui conduit à estimer le chemin de régularisation L0 . Ces algorithmes sont inspirés de l'algorithme d'homotopie pour la régularisation L1 [3, 4] et exploitent le caractère constant par morceaux du chemin de régularisation L0. Continuation Single Best Replacement (CSBR) est basé sur des appels à SBR pour des valeurs décroissantes de λ, calculées de manièere adaptative. L'algorithme plus sophistiqué L0-regularization Path Descent (L0-PD) effectue une reconstruction (sous-optimale) du chemin de régularisation en maintenant (i) une liste de supports candidats pour des valeurs décroissantes de λ; et (ii) une liste de valeurs critiques de λ autour desquelles la solution change. Les simulations numériques montrent l'efficacité des deux algorithmes pour des problèmes inverses difficiles comme la déconvolution impulsionnelle par un filtre passe-bas. Nous montrons finalement que les algorithmes proposés peuvent être avantageusement couplés avec des méthodes de sélection d'ordre comme le MDL (Minimum Description Length) afin de sélectionner automatiquement l'une des solutions parcimonieuses obtenues.
Detection of Overlapping Acoustic Events using a Temporally-Constrained Probabilistic Model
Emmanouil Benetos, Grégoire Lafay, Mathieu Lagrange, Mark Plumbley,
ICASSP, Mar 2016, Shanghai, China. 2016. 〈hal-01255074v2〉
Index Terms Acoustic event detection, Probabilistic latent component analysis, Hidden Markov models
In this paper, a system for overlapping acoustic event detection is proposed, which models the temporal evolution of sound events. The system is based on probabilistic latent component analysis, supporting the use of a sound event dictionary where each exemplar consists of a succession of spectral templates. The temporal succession of the templates is controlled through event class-wise Hidden Markov Models (HMMs). As input time/frequency representation, the Equivalent Rectangular Bandwidth (ERB) spectrogram is used. Experiments are carried out on polyphonic datasets of office sounds generated using an acoustic scene synthesizer-simulator, as well as real and synthesized monophonic datasets for comparative purposes. Results show that the proposed system outperforms several state-of-the-art methods for overlapping acoustic event detection on the same task, using both frame-based and event-based metrics, and is robust to varying event density and noise levels.
Exact Sparse Approximation Problems via Mixed-Integer Programming: Formulations and Computational Performance
Sébastien Bourguignon, Jordan Ninin, Hervé Carfantan, Marcel Mongeau,
IEEE Transactions on Signal Processing, Institute of Electrical and Electronics Engineers, 2016, 64 (6), pp.1405-1419. 〈10.1109/TSP.2015.2496367〉. 〈hal-01254856〉
Deconvolution, Mixed-integer programming, L0-norm-based problems, Optimization, Sparse approximation
Sparse approximation addresses the problem of approximately fitting a linear model with a solution having as few non-zero components as possible. While most sparse estimation algorithms rely on suboptimal formulations, this work studies the performance of exact optimization of problems through Mixed-Integer Programs (MIPs). Nine different sparse optimization problems are formulated based on or data misfit measures, and involving whether constrained or penalized formulations. For each problem, MIP reformulations allow exact optimization, with optimality proof, for moderate-size yet difficult sparse estimation problems. Algorithmic efficiency of all formulations is evaluated on sparse deconvolution problems. This study promotes error-constrained minimization of the norm as the most efficient choice when associated with and misfits, while the misfit is more efficiently optimized with sparsity-constrained and sparsity-penalized problems. Then, exact optimization is shown to outperform classical methods in terms of solution quality, both for over-and under-determined problems. Finally, numerical simulations emphasize the relevance of the different fitting possibilities as a function of the noise statistical distribution. Such exact approaches are shown to be an efficient alternative, in moderate dimension, to classical (suboptimal) sparse approximation algorithms with data misfit. They also provide an algorithmic solution to less common sparse optimization problems based on and misfits. For each formulation, simulated test problems are proposed where optima have been successfully computed. Data and optimal solutions are made available as potential benchmarks for evaluating other sparse approximation methods.
Sensory-motor Anticipation and Local Information Fusion for Reliable Humanoid Approach
Hendry Ferreira Chame, Christine Chevallereau,
P. Wenger et al. (eds.). New Trends in Medical and Service Robots, Mechanisms and Machine Science, 39, © Springer International Publishing Switzerland, 2016, 〈10.1007/978-3-319-30674-2_10〉. 〈hal-01265030〉
Humanoid robotics, Robot Vision, Ego-localization, Top-down visual attention, Cognitive robotics, Embodied cognition
The possibility of developing increasingly sophisticated robots, and the availability of cloud-connected resources, have boosted the interest in the study of real world applications of service robotics. However, in order to operate under natural or less structured conditions, and given the information processing bottleneck and the reactivity required for a secure execution of the task, it is desirable that the agent can exploit more efficiently the local information available, so that being more autonomous, and relying less on remote computation. This study explores a strategy for obtaining reliable approach tasks. It considers the anticipation of perception, by taking into account the statistical regularities and the information redundancies induced in the sensory-motor coupling. From an initial perception of the object assisted by remote computation, contextual features are defined for capturing bodily sensations emerging in the task. The observations based on proprioceptive and visual data are fused in a Bayesian Network, which is in charge of assessing the saliency during the object approach, thus constituing a local discriminative processing of the object. The strategy proposed reduces dependency on context-free models of behavior, while providing an estimate on the degree of confidence in the progress of the task.
A morphological model for simulating acoustic scenes and its application to sound event detection
Grégoire Lafay, Mathieu Lagrange, Mathias Rossignol, Emmanouil Benetos, Axel Roebel,
IEEE/ACM Transactions on Audio, Speech and Language Processing, Institute of Electrical and Electronics Engineers, 2016, 24 (10), pp.1854-1864. 〈10.1109/TASLP.2016.2587218〉. 〈hal-01111381v2〉
Auditory scene analysis, Acoustic event detection, Experimental validation
This paper introduces a model of environmental acoustic scenes which adopts a morphological approach by ab-stracting temporal structures of acoustic scenes. To demonstrate its potential, this model is employed to evaluate the performance of a large set of acoustic events detection systems. This model allows us to explicitly control key morphological aspects of the acoustic scene and isolate their impact on the performance of the system under evaluation. Thus, more information can be gained on the behavior of evaluated systems, providing guidance for further improvements. The proposed model is validated using submitted systems from the IEEE DCASE Challenge; results indicate that the proposed scheme is able to successfully build datasets useful for evaluating some aspects the performance of event detection systems, more particularly their robustness to new listening conditions and the increasing level of background sounds.
Influence of image features on face portraits social context interpretation: experimental methods, crowdsourcing based studies and models
Signal and Image Processing. Ecole Centrale de Nantes (ECN), 2015. English. 〈tel-01291459〉
Machine learning, Face portraits, Social context, Crowdsourcing, Image features, Apprentissage automatique, Contexte sociale, Caractéristiques d’images, Photos de portrait
Online communities and social networks are more and more present in everyday life. On these networks, people build a virtual profile and interact between them for many different purposes: to be in touch with friends, for business, to make new connections, to find a love partner... Online profiles should be coherent with these tasks, starting with the omnipresent profile picture, as different pictures of the same subject can convey very different messages. This thesis focuses on which elements inside a profile picture modify the perception of the context that best suits the picture itself. We define this concept borrowing the "social context" concept in psychology. Image features considered are both low and high level; while the first are more technical quantities related to the sole pixels values (i.e. brightness, contrast), the latter are related to the understanding of the scene depicted in the picture. These elements are underlined by results of research on psychology (i.e. influence of clothing or gaze direction in social interactions). These features have been evaluated through crowdsourcing, a relatively new technique that exploits the power of the web to outsource simple tasks. We adopted the same technique to gather social context evaluations, being this a subjective perception. Then, through different mathematical approaches, we modelled the social context with image features, to understand and quantify the influence of each feature. Results are in line with empirical experience.