Diffusion and inpainting of reflectance and height LiDAR orthoimages

Abstract : This paper presents a fully automatic framework for the production of both reflectance and height orthoimages at very high resolution (1cm 2 per pixel) from a mobile laser scan. Traditional orthoimages are obtained using aerial techniques which are limited by their accuracy and the occlusions due to hovering objects (trees, tunnels, bridges ...). Mobile laser scan is an acquisition technique made at ground level which overcomes those limitations. In this paper, we propose a framework to reconstruct high quality ground orthoimages from a point cloud acquired via mobile laser scan in a urban environment. The point cloud provided by the laser scan is first projected on a 2D-pixel grid to generate sparse and noisy re-flectance and height images. Dense images are computed using an anisotropic diffusion algorithm taking both channels into account. Next, we extract a mask of the large regions where diffusion does not provide a visually satisfying result. Finally, we propose an examplar-based inpainting method to fill-in these regions. It relies on both channels and an assumption about the alignment between structures to inpaint. A number of examples on real acquisition data demonstrates the effectiveness of the proposed pipeline in generating a very high resolution orthoimage of reflectance and height while filling holes of various sizes in a visually satisfying way.
Keywords : Orthophoto
Type de document :
Pré-publication, Document de travail
2016
Liste complète des métadonnées


https://hal.archives-ouvertes.fr/hal-01322822
Contributeur : Pierre Biasutti <>
Soumis le : vendredi 27 mai 2016 - 17:29:20
Dernière modification le : lundi 23 janvier 2017 - 10:13:04
Document(s) archivé(s) le : dimanche 28 août 2016 - 10:43:33

Fichier

paper.pdf
Fichiers produits par l'(les) auteur(s)

Licence


Distributed under a Creative Commons Paternité 4.0 International License

Identifiants

  • HAL Id : hal-01322822, version 1

Collections

Citation

Pierre Biasutti, Jean-François Aujol, Mathieu Brédif, Aurélie Bugeau. Diffusion and inpainting of reflectance and height LiDAR orthoimages. 2016. <hal-01322822>

Partager

Métriques

Consultations de
la notice

200

Téléchargements du document

115