A super-resolution framework for high-accuracy multiview reconstruction

• Verfasst von: Goldlücke, Bastian [VerfasserIn]
• Verfasst von: Aubry, Mathieu [VerfasserIn]
• Verfasst von: Kolev, Kalin [VerfasserIn]
• Verfasst von: Cremers, Daniel [VerfasserIn]
• Titel: A super-resolution framework for high-accuracy multiview reconstruction
• Verf.angabe: Bastian Goldlücke, Mathieu Aubry, Kalin Kolev, Daniel Cremers
• Jahr: 2014
• Jahr des Originals: 2013
• Umfang: 20 S.
• Fussnoten: Published online: 25 August 2013 ; Gesehen am 06.10.2020
• Titel Quelle: Enthalten in: International journal of computer vision
• Ort Quelle: Dordrecht [u.a.] : Springer Science + Business Media B.V, 1987
• Jahr Quelle: 2014
• Band/Heft Quelle: 106(2014), 2, Seite 172-191
• ISSN Quelle: 1573-1405
• DOI: doi:10.1007/s11263-013-0654-8
• Datenträger: Online-Ressource
• Sprache: eng
• K10plus-PPN: 1734807512

Abstract

We present a variational framework to estimate super-resolved texture maps on a 3D geometry model of a surface from multiple images. Given the calibrated images and the reconstructed geometry, the proposed functional is convex in the super-resolution texture. Using a conformal atlas of the surface, we transform the model from the curved geometry to the flat charts and solve it using state-of-the-art and provably convergent primal-dual algorithms. In order to improve image alignment and quality of the texture, we extend the functional to also optimize for a normal displacement map on the surface as well as the camera calibration parameters. Since the sub-problems for displacement and camera parameters are non-convex, we revert to relaxation schemes in order to robustly estimate a minimizer via sequential convex programming. Experimental results confirm that the proposed super-resolution framework allows to recover textured models with significantly higher level-of-detail than the individual input images.