Fusion and quality assessment of multispectral, hyperspectral and panchromatic satellite images

Directeur de thèse :     Jocelyn CHANUSSOT     Jocelyn CHANUSSOT

École doctorale : Electronique, electrotechnique, automatique, traitement du signal (eeats)

Spécialité : Signal, image, parole, télécoms

Structure de rattachement : Grenoble-INP

Établissement d'origine : INPG

Financement(s) : bourse attribuée par un gouvernement étranger ; contrat à durée déterminée

Date d'entrée en thèse : 08/11/2006

Date de soutenance : 18/11/2009

Composition du jury :
M. Lucien WALD, Président
M. Roger KING, Rapporteur
M. Thierry RANCHIN, Rapporteur
Mme. Annick MONTANVERT,Examinateur
M. Andrea GARZELLI,Examinateur
M. Jocelyn CHANUSSOT, Directeur
M. Luciano ALPARONE, Co-directeur

Résumé : Fusion or Pansharpening improves the spatial quality of Multispectral satellite images by making use of details present in a high resolution Panchromatic image. The different nature of Panchromatic and Multispectral sensors requires that the details are added using a detail injection model. Most of the available detail injection model parameters are estimated at degraded resolution and then interpolated for use at the higher resolution. In this work we have presented a detail injection model which estimates the detail injection parameters directly at the desired higher resolution. The model optimizes the Quality Not Requiring a Reference index. The proposed method is then extended for pansharpening of Hyperspectral images. Finally, a quality assessment index, not requiring a high resolution reference, is proposed for determining the quality of pansharpened images at full resolution. The novelty of the method lies in the use of sensors Modulation Transfer Function filters for separating the low and high frequency information for determining the spectral and spatial distortions, respectively.