Morphologie mathématique

Résumé

La morphologie mathématique est une théorie et technique mathématique et informatique d'analyse de structures qui est liée avec l'algèbre, la théorie des treillis, la topologie et les probabilités. Le développement de la morphologie mathématique est inspiré des problèmes de , domaine qui constitue son principal champ d'application. Elle fournit en particulier des outils de filtrage, , quantification et modélisation d'images. Elle est également utilisable en traitement du signal, par exemple pour filtrer les variations d'une mesure (physique, biologique) au cours du temps. Aperçu général Une des idées de base de la morphologie mathématique est d'étudier ou de traiter un ensemble à l'aide d'un autre ensemble, appelé élément structurant, qui sert de sonde. À chaque position de l'élément structurant, on regarde s'il touche ou s'il est inclus dans l'ensemble initial. En fonction de la réponse, on construit un ensemble de sortie. On obtient ainsi des opérateurs de base qui so

Source officielle

https://fr.wikipedia.org/wiki/Morphologie_math%C3%A9matique

À propos de ce résultat

Cette page est générée automatiquement et peut contenir des informations qui ne sont pas correctes, complètes, à jour ou pertinentes par rapport à votre recherche. Il en va de même pour toutes les autres pages de ce site. Veillez à vérifier les informations auprès des sources officielles de l'EPFL.

Publications associées (20)

Publications associées

Chargement

Personnes associées (2)

Personnes associées

Chargement

Unités associées (2)

Unités associées

Chargement

Concepts associés (2)

Concepts associés

Chargement

Cours associés (2)

Cours associés

Chargement

Séances de cours associées (2)

Séances de cours associées

Chargement

In this paper, we study an emerging class of neural networks, the Morphological Neural networks, from some modern perspectives. Our approach utilizes ideas from tropical geometry and mathematical morphology. First, we state the training of a binary morphological classifier as a Difference-of-Convex optimization problem and extend this method to multiclass tasks. We then focus on general morphological networks trained with gradient descent variants and show, quantitatively via pruning schemes as well as qualitatively, the sparsity of the resulted representations compared to FeedForward networks with ReLU activations as well as the effect the training optimizer has on such compression techniques. Finally, we show how morphological networks can be employed to guarantee monotonicity and present a softened version of a known architecture, based on Maslov Dequantization, which alleviates issues of gradient propagation associated with its "hard" counterparts and moderately improves performance.

IEEE2021

Chargement

Texturing the glass surface is a promising method for improving the light trapping properties of superstrate thin-film silicon solar cells, as it enables thinner absorber layers and, possibly, higher cell efficiencies. In this paper we present the optical and morphological properties of borosilicate glass superstrates textured with the aluminium induced texture (AIT) method. High haze values are achieved without any reduction in the total optical transmission of the glass sheets after the AIT process. Scanning electron microscope and atomic force microscope (AFM) measurements reveal a laterally uniform surface morphology of the AIT texture. We demonstrate that the surface roughness and thus the transmission haze can be controlled by adjusting the AIT process parameters. From the AFM images, we extract histograms of the local height and angle distributions of the texture. Samples with a wide angle distribution are shown to produce the highest optical haze. The results of this analysis provide a better understanding of the correlation between the AIT process parameters and the resulting surface morphology. This analysis is further extended to an amorphous silicon pin solar cell deposited onto the textured glass substrate. (C) 2012 The Japan Society of Applied Physics

Japan Society of Applied Physics2012

Chargement

A multi-scale methodology for the radiative transfer analysis of heterogeneous media composed of morphologically-complex components on two distinct scales is presented. The methodology incorporates the exact morphology at the various scales and utilizes volume-averaging approaches with the corresponding effective properties to couple the scales. At the continuum level, the volume-averaged coupled radiative transfer equations are solved utilizing i) effective radiative transport properties obtained by direct Monte Carlo simulations at the pore level, and ii) averaged bulk material properties obtained at particle level by Lorenz-Mie theory or discrete dipole approximation calculations. This model is applied to a soot-contaminated snow layer, and is experimentally validated with reflectance measurements of such layers. A quantitative and decoupled understanding of the morphological effect on the radiative transport is achieved, and a significant influence of the dual-scale morphology on the macroscopic optical behavior is observed. Our results show that with a small amount of soot particles, of the order of 1ppb in volume fraction, the reduction in reflectance of a snow layer with large ice grains can reach up to 77% (at a wavelength of 0.3 μm). Soot impurities modeled as compact agglomerates yield 2-3% lower reduction of the reflectance in a thick show layer compared to snow with soot impurities modeled as chain-like agglomerates. Soot impurities modeled as equivalent spherical particles underestimate the reflectance reduction by 2-8%. This study implies that the morphology of the heterogeneities in a media significantly affects the macroscopic optical behavior and, specifically for the soot-contaminated snow, indicates the non-negligible role of soot on the absorption behavior of snow layers. It can be equally used in technical applications for the assessment and optimization of optical performance in multi-scale media.

Elsevier2018

Chargement