Near-Field Photometric Stereo in Ambient Light - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2016

Near-Field Photometric Stereo in Ambient Light

Résumé

Shape recovery from shading information has recently regained importance due to the improvement towards making the Photometric Stereo technique more reliable in terms of appearance of reflective objects. However, although more advanced models have been lately proposed, 3D scanners based on this technology do not provide reliable reconstructions as long as the considered irradiance equation neglects any additive bias. Depending on the context, such bias assumes different physical meanings. For example, in murky water it is known as saturated backscattered effect or for acquisition in pure air medium it is known as ambient light. Although the theoretical part covers both cases, this work mostly focuses on the pure air acquisition case. Indeed, we present a new approach based on ratios of differences of images where an exhaustive set of physical features are tackled while dealing with Photometric Stereo acquisition with considerable importance for the ambient light. To the best of our knowledge, this is the first attempt to recover the shape from Photometric Stereo considering simultaneously perspective viewing geometry, non-linear light propagation, both specular and diffuse reflectance plus the additive bias of the ambient light. Proof of concept is provided by showing experimental results on synthetic and real data.
Fichier principal
Vignette du fichier
logothetis_17235.pdf (1 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-01436196 , version 1 (16-01-2017)

Identifiants

  • HAL Id : hal-01436196 , version 1
  • OATAO : 17235

Citer

Fotios Logothetis, Roberto Mecca, Yvain Quéau, Roberto Cipolla. Near-Field Photometric Stereo in Ambient Light. 27th British Machine Vision Conference (BMVC 2016), Sep 2016, York, United Kingdom. pp. 1-12. ⟨hal-01436196⟩
144 Consultations
123 Téléchargements

Partager

Gmail Facebook X LinkedIn More