Didier Auroux Laboratoire MIP Institut de Mathematiques de Toulouse Universite Paul Sabatier Toulouse France ups tlse fr

Didier Auroux

Description

Didier Auroux Laboratoire MIP - Institut de Mathematiques de Toulouse Universite Paul Sabatier Toulouse 3 France Image restoration and classification by topological asymptotic expansion. Mathematics and Image Analysis 2006, MIA'06, Paris, France September 18-21, 2006

• image restoration


• general formulation


• preserve them


• u? ??


• restoration-based preprocessing


• becoming blurred


• topological gradient


• numerical results

Sujets

• Analysis
• Sabatier
• Auroux
• Image processing

Informations

• Publié par: pefav
• Nombre de lectures: 14
• Langue: English

Extrait

MIA’06,Paris,France MathematicsandImageAnalysis2006, Imagerestorationandclassification Didier Auroux September18-21,2006 bytopologicalasymptoticexpansion. France [e-mail] LaboratoireMIP-InstitutdeMath´ematiquesdeToulouse Universit´ePaulSabatierToulouse3

Talkoverview 1.Introductiontotopologicalasymptoticexpansion 2.Imagerestoration (a)Linearapproach (b)Nonlinearvariationalapproach (c)Topologicalgradientapproach 3.Regularizedimageclassification (a)Topologicalgradientappliedtotheclassicalapproach (b)Restoration-basedpreprocessingmethod 4.Afewmorenumericalresults 5.Conclusion MIA’06,Paris,September20,2006/146

Introduction Imagerestorationproblem:howtode-noisetheimage?Inordertoprevent theimagebecomingblurred,wemustidentifyitscontoursinordertopreserve m.eht Tofindanoptimalshape(thesetofcontours)isequivalenttofindits characteristicfunction(0−1optimizationproblem). Thegoalofimageclassificationistofindapartitionoftheimageinsubsets withaconstantcolorlevelineachsubset. Mainprobleminbothcases:thenondifferentiability. MIA’06,Paris,September20,200664/2

MIA’06, A SYMPTOTIC A NALYSIS T OPOLOGICAL Paris,September20,2006 G RADI ADN ENT 463/