Search results for " processing"
showing 10 items of 7549 documents
Multispectral imaging for computer vision
2018
Refitting solutions promoted by $\ell_{12}$ sparse analysis regularization with block penalties
2019
International audience; In inverse problems, the use of an $\ell_{12}$ analysis regularizer induces a bias in the estimated solution. We propose a general refitting framework for removing this artifact while keeping information of interest contained in the biased solution. This is done through the use of refitting block penalties that only act on the co-support of the estimation. Based on an analysis of related works in the literature, we propose a new penalty that is well suited for refitting purposes. We also present an efficient algorithmic method to obtain the refitted solution along with the original (biased) solution for any convex refitting block penalty. Experiments illustrate the g…
Approche multicritère pour la caractérisation des adventices par imagerie
2019
La réduction des produits phytosanitaires représente un des enjeux majeurs du secteur agricole. Les plans gouvernementaux Ecophyto, Ecophyto II et Ecophyto II+ visent à réduire fortement leurs usages et les solutions actuelles ne permettent pas d’obtenir les résultats escomptés. La détection des adventices par imagerie est un des axes de travail devant permettre cette réduction. La qualité de la discrimination cultures/adventices est fortement liée au type de méthodes utilisées, à la résolution spatiale des images et au stade de développement des plantes présentes. L'objectif de ce travail, est donc d'évaluer l’impact des différents critères pouvant être extraits depuis des images acquises …
A nonlinear derivative
2014
International audience; A nonlinear derivative is directly defined in the discrete domain. This derivative is motivated by the asymmetry of pattern in discrete signal as step or edge in 2D signal. Thanks to the special definition (in the discrete domain) of this derivative, pattern can be detected in a univocal way. This derivative is the only one able to perfectly detect and localize ideal edges in image. Beside this fundamental benefit, the derivative has the nice property to reduce noise. Ap- plications to edge detection, noise reduction and noise estimation are described and their performances are studied.
Novel weighting and resampling schemes in Population Monte Carlo
2017
International audience; In this paper we focus on novel population Monte Carlo schemes that present added flexibility and superior performance compared to the standard approach. The new schemes combine different weighting and resampling strategies to achieve more efficient performance at the cost of a reasonable computational complexity increment. Computer simulations compare the different alternatives when applied to the problem of frequency estimation in superimposed sinusoids.; Dans cet article, nous nous concentrons sur de nouveaux schémas de population Monte Carlo présentant une flexibilité et des performance supérieures par rapportàrapport`rapportà l'approche standard. Ces nouveaux sc…
Italian People at Work. Jobs in Lexical Syntax
2004
A novel approach for image sweeping functions using approximating scheme
2009
International audience; Using Kolmogorov's superposition theorem, complex N-dimensional signal can be expressed by simpler 1D functions. Precisely, Kolmogorov has demonstrated that any multivariate function can be decomposed into sums and compositions of monovariate functions, that are called inner and external functions. We present one of the most recent method of monovariate functions construction. The algorithm proposed by Igelnik approximates the monovariate functions. Different layers are constructed and superposed. A layer is constituted by a couple of internal and external functions, that realizes an approximation of the multivariate function with a given precision, which corresponds…
Semantic Measures: A State of the Art
2014
Significant advances in terms of syntactic, structural and schematic heterogeneity have been achieved by adopting conventions and standards. The IT community is now trying to solve the problem of semantic heterogeneity (particularly in the Semantic Web field). To reach this objective, it is necessary to enable machines to understand the semantics of terms. Semantics, as opposed to syntax, defines the mental representation of concepts corresponding to the symbols used in texts or images. When a person reads a text, he uses a semantization process which enables him to associate an interpretation to each sign identified. This operation uses a number of underlying processes such as measuring se…
Formalizing Semantic of Natural Language through Conceptualization from Existence
2011
International audience; We propose an outline of an approach to formalize semantic from conceptualization for both natural language (NL) and logic expression mechanisms. This goes beyond the level of discussions at conceptual level which has to either end in conscious/unconscious relativity of understanding or subjective enforcement in the form of definitions instead of expected objective semantic. This approach supports to view from a pure mathematical perspective, and explore and locate the fundamental problems. The semantic formalization mechanism realizes the integration of problem description and the solution expression at absolute semantic level. So a problem describing process is equ…
Temporal Semantic Centrality for the Analysis of Communication Networks
2012
National audience; De nos jours, la compréhension des communautés en ligne devient un enjeu majeur du Web. Dans cet article nous proposons une nouvelle mesure, la Probabilité de Propagation Sémantique (SPP), qui caractérise la capacité de l'utilisateur à propager un concept sémantique à d'autres utilisateurs, d'une manière rapide et ciblée. La sémantique des messages est analysée selon une ontologie donnée. Nous utilisons cette mesure pour obtenir la Centralité Sémantique Temporelle (TSC) d'un utilisateur dans une communauté. Nous proposons et évaluons une expérimentation de cette mesure, en utilisant une ontologie et des données réelles issues du Web.