
H. Trumel, F. Rabette, F. Willot, R. Brenner, E. Ongari, M. Biessy, D. Picart (2019): Understanding the thermomechanical behavior of a TATBbased explosive via microstructurelevel simulations. Part I: Microcracking and viscoelasticity. Europyro 44th International Pyrotechnics Seminar, Tours (France).
In view of a better understanding of the thermomechanical behavior of pressed explosives, a Fourierbased computational tool is used to perform numerical homogenization and compare predictions to experimental macroscopic properties. This is first done in a purely thermoelastic context on simplified polycrystalline virtual microstructures, then extended to cracked polycrystalline ones. A further extension is proposed, aiming at predicting the nucleation and propagation of (micro)cracks. Besides, a meanfield (selfconsistent) approach is also followed, providing accurate thermoelastic...
F. Rabette, F. Willot, H. Trumel (2019): Homogénéisation en champs complets par FFT pour un matériau énergétique à forte anisotropie cristalline : prise en compte de la microfissuration par une méthode de champ de phase. Colloque National MECAMAT 2019: Rupture des Matériaux et des Structures, Aussois (France).
H. Thomas, C.R. Qi, J.E. Deschaud, B. Marcotegui, F. Goulette, L.J. Guibas (2019): KPConv: Flexible and Deformable Convolution for Point Clouds. The IEEE International Conference on Computer Vision (ICCV), Séoul (South Korea).
We present Kernel Point Convolution (KPConv), a new design of point convolution, i.e. that operates on point clouds without any intermediate representation. The convolution weights of KPConv are located in Euclidean space by kernel points, and applied to the input points close to them. Its capacity to use any number of kernel points gives KPConv more flexibility than fixed grid convolutions. Furthermore, these locations are continuous in space and can be learned by the network. Therefore, KPConv can be extended to deformable convolutions that learn to adapt kernel points to local geometry....
D. DuqueArias, S. VelascoForero, J.E. Deschaud, F. Goulette, B. Marcotegui (2019): A graphbased color lines model for image analysis. International Conference on Image Analysis and Processing, Trento (Italy).
This paper addresses the problem of obtaining a concise description of spectral representation for color images. The proposed method is a graphbased formulation of the wellknown Color Lines model. It generalizes the lines to piecewise lines, been able to fit more complex structures. We illustrate the goodness of proposed method by measuring the quality of the simplified representations in images and videos. The quality of video sequences reconstructed by means of proposed color lines extracted from the first frame demonstrates the robustness of our representation. Our formalism allows to...
E. Bazan, P. Dokládal, E. Dokladalova (2019): Quantitative Analysis of Similarity Measures of Distributions. British Machine Vision Conference (BMVC), Cardiff (United Kingdom).
There are many measures of dissimilarity that, depending on the application, do not always have optimal behavior. In this paper, we present a qualitative analysis of the similarity measures most used in the literature and the Earth Mover's Distance (EMD). The EMD is a metric based on the theory of optimal transport with interesting geometrical properties for the comparison of distributions. However, the use of this measure is limited in comparison with other similarity measures. The main reason was, until recently, the computational complexity. We show the superiority of the EMD through three...
S. VelascoForero, B. Ponchon, S. Blusseau, J. Angulo, I. Bloch (2019): On approximating mathematical morphology operators via deep learning techniques. 15th International Congress for Stereology and Image Analysis, Aarhus (Denmark) 51.
Mathematical Morphology (MM) is a wellestablished discipline whose aim is mainly to provide tools to characterise complex object via their shape/size features. This study addresses the problem of robust approximation of mathematical morphology (MM) operators by deep learning methods. We present two cases, (a) Asymmetric autoencoders for partbased approximations of classical MM in the sense of [1] and, (b) imagetoimage translation networks [2] to produce robust MM operators in presence of noise.
S. Blusseau, Y. Zhang, S. VelascoForero, I. Bloch, J. Angulo (2019): Pruning neural networks thanks to morphological layers. 15th International Congress for Stereology and Image Analysis, Aarhus (Denmark) 17.
Motivated by recent advances in morphological neural networks, we further study the properties of morphological units when incorporated in layers of conventional neural networks. We confirm and extend the observation that a Maxplus layer can be used to select relevant filters and reduce redundancy in its previous layer, without incurring performance loss. We present several experiments in image processing, showing that this filter selection property seems efficient and robust. We also point out the close connection between Maxout networks and our pruned Maxplus networks. The code related to...
Y. Yan, P.H. Conze, E. Decencière, M. Lamard, G. Quellec, B. Cochener, G. Coatrieux (2019): Cascaded multiscale convolutional encoderdecoders for breast mass segmentation in highresolution mammograms. IEEE International Engineering in Medicine and Biology Conference, Berlin (Germany).
Y. Xiao, E. Decencière, S. VelascoForero, H. Burdin, T. Bornschlögl, F. Bernerd, E. Warrick, T. Baldeweck (2019): A new color augmentation method for deep learning segmentation of histological images. 2019 IEEE 16th International Symposium on Biomedical Imaging (ISBI), Venise (France).
This paper addresses the problem of labeled data insufficiency in neural network training for semantic segmentation of colorstained histological images acquired via Whole Slide Imaging. It proposes an efficient image augmentation method to alleviate the demand for a large amount of labeled data and improve the network's generalization capacity. Typical image augmentation in bioimaging involves geometric transformation. Here, we propose a new image augmentation technique by combining the structure of one image with the color appearance of another image to construct augmented images onthefly...
F. Cadiou, A. Etiemble, T. Douillard, F. Willot, O. Valentin, J.C. Badot, B. Lestriez, E. Maire (2019): Numerical Prediction of Multiscale Electronic Conductivity of LithiumIon Battery Positive Electrodes. Journal of The Electrochemical Society 166(8) A1692—A1703.
The electronic conductivity, at the multiscale, of lithiumion positive composite electrodes based on LiNi_{1/3}Mn_{1/3}Co_{1/3}O_2 and/or carboncoated LiFePO_4, carbon black and poly(vinylidene fluoride) mixture is modeled. The electrode microstructures are acquired numerically in 3D by Xray tomography and FIB/SEM nanotomography and numerically segmented to perform electrostatic simulations using Fast Fourier Transform (FFT) method. Such simulations are easy and quick to perform because they are directly computed on the grid represented by the voxels in the 3D volumes. Numerical results...
P. CettourJanet, C. Cazorla, V. Machairas, Q. Delannoy, N. Bednarek, F. Rousseau, E. Decencière, N. Passat (2019): Watervoxels. Image Processing On Line 9 317—328.
In this article, we present the $n$dimensional version of the waterpixels, namely the watervoxels. Waterpixels constitute a simple, yet efficient alternative to standard superpixel paradigms, initially developed in the field of computer vision for reducing the space cost of input images without altering the accuracy of further image processing / analysis procedures. Waterpixels were initially proposed in a 2dimensional version. Their extension to 3dimensionsand more generally $n$dimensionsis however possible, in particular in the Cartesian grid. Indeed, waterpixels mainly rely on a...
Kaeshammer, P. Dokládal, F. Willot, B. Erzar, S. Belon, L. Borne (2019): A Morphological Study of Energetic Materials: Analysis of Microcomputed Tomography Images to Generate Representative Microstructures. Europyro 2019 44th International Pyrotechnics Seminar, Tours (France).
In previous work, impact experiments were performed at ISL on three energetic materials composed of 70% in weight of RDX particles embedded in a wax matrix. These materials differ by the microstructural properties of the explosive particles. The experimental results reveal that the detonation thresholds, and so the sensitivity to shock, are different for each sample. To better understand these results, we characterize the microstructural properties of these compositions and generate virtual microstructures representative of the real microstructures. First, the microstructures of the three...
Kaeshammer, P. Dokladal, F. Willot, S. Belon, L. Borne (2019): Generation of Virtual Microstructures of Energetic Materials Based on Microcomputed Tomography Images Analysis. 50th International Annual Conference of the Fraunhofer ICT, Karlsruhe (Germany).
Impact experiments were performed at the frenchgerman research Institute of SaintLouis on three energetic materials composed of 70 % in weight of RDX particles embedded in a wax matrix. These materials differ by the microstructural properties of the explosive particles. The experimental results reveal that the detonation thresholds, and so the sensitivity to shock, are different for each sample. To better understand these results, we characterize the microstructural properties of these compositions. The microstructures of the three materials are imaged with microcomputed tomography (µCT)...
M. Neumann, B. Abdallah, L. Holzer, F. Willot, V. Schmidt (2019): Stochastic 3D Modeling of ThreePhase Microstructures for Predicting Transport Properties: A Case Study. Transport in Porous Media 128(1) 179—200.
We compare two conceptually different stochastic microstructure models , i.e., a graphbased model and a pluriGaussian model, that have been introduced to model the transport properties of threephase microstructures occurring, e.g., in solid oxide fuel cell electrodes. Besides comparing both models, we present new results regarding the relationship between model parameters and certain microstructure characteristics. In particular, an analytical expression is obtained for the expected length of triple phase boundary per unit volume in the pluriGaussian model. As a case study, we consider...
J. Serra, F. Willot (2019): Special topic on multiscale modeling of granular media: a tribute to Prof. Dominique Jeulin. Image Analysis and Stereology 38(1) 1.
A few words on the present special topic, devoted to the multiscale modeling of granular media, and published in honor of Prof. Dominique Jeulin's enduring contribution to the wide field of image analysis, random structures and material science.
B. Ponchon, S. VelascoForero, S. Blusseau, J. Angulo, I. Bloch (2019): Partbased approximations for morphological operators using asymmetric autoencoders. International Symposium on Mathematical Morphology, Saarbrücken (Germany).
This paper addresses the issue of building a partbased representation of a dataset of images. More precisely, we look for a nonnegative, sparse decomposition of the images on a reduced set of atoms, in order to unveil a morphological and interpretable structure of the data. Additionally, we want this decomposition to be computed online for any new sample that is not part of the initial dataset. Therefore, our solution relies on a sparse, nonnegative autoencoder where the encoder is deep (for accuracy) and the decoder shallow (for interpretability). This method compares favorably to the...
Y. Zhang, S. Blusseau, S. VelascoForero, I. Bloch, J. Angulo (2019): Maxplus Operators Applied to Filter Selection and Model Pruning in Neural Networks. International Symposium on Mathematical Morphology, Saarbrücken (Germany).
Following recent advances in morphological neural networks, we propose to study in more depth how Maxplus operators can be exploited to define morphological units and how they behave when incorporated in layers of conventional neural networks. Besides showing that they can be easily implemented with modern machine learning frameworks , we confirm and extend the observation that a Maxplus layer can be used to select important filters and reduce redundancy in its previous layer, without incurring performance loss. Experimental results demonstrate that the filter selection strategy enabled by...
F. Willot, H. Trumel, D. Jeulin (2019): The thermoelastic response of cracked polycrystals with hexagonal symmetry. Philosophical Magazine 99(5) 606—630.
The influence of a population of randomlyoriented cracks on the macroscopic thermal and linearelastic response of a hexagonal polycrystal is addressed using a selfconsistent method. Coupling between microcracks and crystal anisotropy is taken into account through the effective medium where all inhomogeneities are embedded. In the absence of cracks, the proposed approach reduces to the selfconsistent estimate of Berryman (2005). The accuracy of the present method is first assessed using numerical, Fourierbased computations. In the absence of crystal anisotropy, the estimates for the...
E.H. Diop, J. Angulo (2019): Levelings based on SpatiallyAdaptive ScaleSpaces using Local Image Features. IET Image Processing 13(10) 1597—1607.
A. Borocco, B. Marcotegui (2019): Nonrigid shape registration using curvature information. 14th International Conference on Computer Vision Theory and Applications, Prague (Czech Republic).
This paper addresses a registration problem for an industrial control application: it meets the need to registrate a model on an image of a flexible object. We propose a nonrigid shape registration approach that deals with a great disparity of the number of points in the model and in the manufactured object. We have developed a method based on a classical minimization process combining a distance term and a regularization term. We observed that, even if the control points fall on the object boundary, the registration failed on high curvature points. In this paper we add a curvaturebased...
R.B. Metcalf, M. Meneghetti, C. Avestruz, F. Bellagamba, C.R. Bom, E. Bertin, R. Cabanac, F. Courbin, A. Davies, E. Decencière, R. Flamary, R. Gavazzi, M. Geiger, P. Hartley, M. HuertasCompany, N. Jackson, C. Jacobs, E. Jullo, J.P. Kneib, L.V.E. Koopmans, F. Lanusse, C.L. Li, Q. Ma, M. Makler, N. Li, M. Lightman, C.E. Petrillo, S. Serjeant, C. Schäfer, A. Sonnenfeld, A. Tagore, C. Tortora, D. Tuccillo, M.B. Valentín, S. VelascoForero, G.A. Verdoes Kleijn, G. Vernardos (2019): The Strong Gravitational Lens Finding Challenge. Astron.Astrophys. 625 A119.
Largescale imaging surveys will increase the number of galaxyscale strong lensing candidates by maybe three orders of magnitudes beyond the number known today. Finding these rare objects will require picking them out of at least tens of millions of images, and deriving scientific results from them will require quantifying the efficiency and bias of any search method. To achieve these objectives automated methods must be developed. Because gravitational lenses are rare objects, reducing false positives will be particularly important. We present a description and results of an open...
R. Rodriguez Salas, P. Dokládal, E. Dokladalova (2019): REDNN: RotationEquivariant Deep Neural Network for Classification and Prediction of Rotation.
In this work, we propose a new Convolutional Neural Network (CNN) for classification of rotated objects. This architecture is built around an ordered ensemble of oriented edge detectors to create a rototranslational space that transforms the input rotation into translation. This space allows the subsequent predictor to learn the internal spatial and angular relations of the objects regardless of their orientation. No data augmentation is needed and the model remains significantly smaller. It presents a selforganization capability and learns to predict the class and the rotation angle...
R. Rodriguez Salas, E. Dokladalova, P. Dokládal (2019): Rotation invariant CNN using scattering transform for image classification.
Deep convolutional neural networks accuracy is heavily impacted by rotations of the input data. In this paper, we propose a convolutional predictor that is invariant to rotations in the input. This architecture is capable of predicting the angular orientation without angleannotated data. Furthermore, the predictor maps continuously the random rotation of the input to a circular space of the prediction. For this purpose, we use the rototranslation properties existing in the Scattering Transform Networks with a series of 3D Convolutions. We validate the results by training with upright and...
J. DIRRENBERGER, S. FOREST, D. Jeulin (2019): Computational Homogenization of Architectured Materials. Architectured Materials in Nature and Engineering 89—139.
S. Bancelin, B. Lynch, C. BonodBidaud, P. Dokládal, F. Ruggiero, J.M. Allain, M.C. SchanneKlein (2019): Combination of Traction Assays and Multiphoton Imaging to Quantify Skin Biomechanics. Collagen : Methods and Protocols 1944 145—155.
An important issue in tissue biomechanics is to decipher the relationship between the mechanical behavior at macroscopic scale and the organization of the collagen fiber network at microscopic scale. Here, we present a protocol to combine traction assays with multiphoton microscopy in ex vivo murine skin. This multiscale approach provides simultaneously the stress/stretch response of a skin biopsy and the collagen reorganization in the dermis by use of second harmonic generation (SHG) signals and appropriate image processing.
R. Rodriguez Salas, P. Dokládal, E. Dokladalova (2019): Rotationinvariant NN for learning naturally unoriented data.
Deep convolutional neural networks accuracy is heavily impacted by the rotations of the input data. In this paper, we propose a convolutional predictor that is invariant to rotations in the input. This architecture is capable of predicting the angular orientation without angleannotated data. Furthermore, the predictor maps continuously the random rotation of the input to a circular space of the prediction. For this purpose, we use the rototranslation properties existing in the Scattering Transform Networks with a series of 3D Convolutions. We validate the results by training with upright...
E. Bazan, P. Dokládal, E. Dokladalova (2019): Quantitative Analysis of Similarity Measures of Distributions.
The Earth Mover's Distance (EMD) is a metric based on the theory of optimal transport that has interesting geometrical properties for distributions comparison. However, the use of this measure is limited in comparison with other similarity measures as the KullbackLeibler divergence. The main reason was, until recently, the computation complexity. In this paper, we present a comparative study of the dissimilarity measures most used in the literature for the comparison of distributions through a colorbased image classification system and other simple examples with synthetic data. We show that...
List of all publications from the CMM, recorded on the HAL depository under the tag
ENSMP_CMM
.
See also: