Titre :	Automatisation des techniques de reconnaissance d'objets au sol a partir d'images satellites
Type de document :	texte imprimé
Auteurs :	BENABADJI Noureddine, Auteur
Année de publication :	2006
Importance :	121 p.
Langues :	Français (fre)
Catégories :	Physique:Option Rayonnement et Matière
Mots-clés :	Satellite imagery , image processing , segmentation , automatic classification,spectral signatures , expert system , rules basis , inferce engine
Résumé :	We propose in this work a methodology of classification in satellite imagery based on the detection and the recognition of objects , using their radiometric properties (spectral signatures) this article describes an expert system model characterized by its modularity uniformity and simplicity , the declarative aspect , necessary to make rules in predetermined situtions , is naturally coded in a simple formalism but efficient in a first step the rules basis is established following an original approach with consists of transcribing a plot diagram of typical spectraml signatures for SPOT satellite ( in multiband channels ) in a second step an inference engine algorithm is used to evaluate the rules basis for automatic detection and recognition of objects
Directeur de thèse :	BELBACHIR Ahmad hafid

Automatisation des techniques de reconnaissance d'objets au sol a partir d'images satellites [texte imprimé] / BENABADJI Noureddine, Auteur . - 2006 . - 121 p.
Langues : Français (fre)

Catégories :	Physique:Option Rayonnement et Matière
Mots-clés :	Satellite imagery , image processing , segmentation , automatic classification,spectral signatures , expert system , rules basis , inferce engine
Résumé :	We propose in this work a methodology of classification in satellite imagery based on the detection and the recognition of objects , using their radiometric properties (spectral signatures) this article describes an expert system model characterized by its modularity uniformity and simplicity , the declarative aspect , necessary to make rules in predetermined situtions , is naturally coded in a simple formalism but efficient in a first step the rules basis is established following an original approach with consists of transcribing a plot diagram of typical spectraml signatures for SPOT satellite ( in multiband channels ) in a second step an inference engine algorithm is used to evaluate the rules basis for automatic detection and recognition of objects
Directeur de thèse :	BELBACHIR Ahmad hafid