Sensor Noise Modeling by Stacking Pseudo-Periodic Grid Images Affected by Vibrations

Frédéric Sur 1 Michel Grediac 2
1 MAGRIT - Visual Augmentation of Complex Environments
Inria Nancy - Grand Est, LORIA - ALGO - Department of Algorithms, Computation, Image and Geometry
Abstract : This letter addresses the problem of noise estimation in raw images from digital sensors. Assuming that a series of images of a static scene are available, a possibility is to characterize the noise at a given pixel by considering the random fluctuations of the gray level across the images. However, mechanical vibrations, even tiny ones, affect the experimental setup, making this approach ineffective. The contribution of this letter is twofold. It is shown that noise estimation in the presence of vibrations is actually biased. Focusing on images of a pseudo-periodic grid, an algorithm to discard their effect is also given. An application to the generalized Anscombe transform is discussed.
Liste complète des métadonnées

Littérature citée [16 références]  Voir  Masquer  Télécharger
Contributeur : Frédéric Sur <>
Soumis le : mercredi 5 mars 2014 - 12:06:33
Dernière modification le : mardi 18 décembre 2018 - 16:18:26
Document(s) archivé(s) le : jeudi 5 juin 2014 - 11:15:48


Fichiers produits par l'(les) auteur(s)



Frédéric Sur, Michel Grediac. Sensor Noise Modeling by Stacking Pseudo-Periodic Grid Images Affected by Vibrations. IEEE Signal Processing Letters, Institute of Electrical and Electronics Engineers, 2014, 21 (4), pp.432-436. 〈10.1109/LSP.2014.2304570〉. 〈hal-00955709v2〉



Consultations de la notice


Téléchargements de fichiers