Fusion de données RVB-D par stéréophotométrie colorée

Yvain Quéau; Bastien Durix; Tom Lucas; Jade Boumaza; Jean-Denis Durou; Francois Bernard Lauze

Fusion de données RVB-D par stéréophotométrie colorée

Yvain Quéau, Bastien Durix, Tom Lucas, Jade Boumaza, Jean-Denis Durou, Francois Bernard Lauze

Department of Computer Science

Abstract

We propose a new approach of photometric stereo with ani-
sotropic pointwise sources such as LEDs. This approach is
both differential and variational, in order to recover the
depth directly and in a robust manner, without estimating
the normals and the albedo in a first step. We then show
how to use this approach in order to refine the depth map
obtained with a RGB-D sensor, by integrating on the sensor
three colored LEDs allowing us to apply colored photome-
tric stereo on the RGB image.

Translated title of the contribution	RGB-D data fusion by colored photometric stereo
Original language	French
Publication date	2016
Number of pages	8
Publication status	Published - 2016
Event	Rencontres Francophones d'Intelligence Artificielle 2016 - Université Blaise Pascal, Clermont-Ferrand, France Duration: 27 Jun 2016 → 1 Jul 2016 http://rfia2016.iut-auvergne.com/

Conference

Conference	Rencontres Francophones d'Intelligence Artificielle 2016
Location	Université Blaise Pascal
Country/Territory	France
City	Clermont-Ferrand
Period	27/06/2016 → 01/07/2016
Internet address	http://rfia2016.iut-auvergne.com/

Access to Document

http://rfia2016.iut-auvergne.com/media/articles/CPV03.pdf

Cite this

@conference{08c6dabb3fcc425dba0f84e1102d73bc,

title = "Fusion de donn{\'e}es RVB-D par st{\'e}r{\'e}ophotom{\'e}trie color{\'e}e",

abstract = "We propose a new approach of photometric stereo with ani-sotropic pointwise sources such as LEDs. This approach isboth differential and variational, in order to recover thedepth directly and in a robust manner, without estimatingthe normals and the albedo in a first step. We then showhow to use this approach in order to refine the depth mapobtained with a RGB-D sensor, by integrating on the sensorthree colored LEDs allowing us to apply colored photome-tric stereo on the RGB image.",

author = "Yvain Qu{\'e}au and Bastien Durix and Tom Lucas and Jade Boumaza and Jean-Denis Durou and Lauze, {Francois Bernard}",

year = "2016",

language = "Fransk",

note = "Rencontres Francophones d'Intelligence Artificielle 2016 ; Conference date: 27-06-2016 Through 01-07-2016",

url = "http://rfia2016.iut-auvergne.com/",

}

TY - CONF

T1 - Fusion de données RVB-D par stéréophotométrie colorée

AU - Quéau, Yvain

AU - Durix, Bastien

AU - Lucas, Tom

AU - Boumaza, Jade

AU - Durou, Jean-Denis

AU - Lauze, Francois Bernard

PY - 2016

Y1 - 2016

N2 - We propose a new approach of photometric stereo with ani-sotropic pointwise sources such as LEDs. This approach isboth differential and variational, in order to recover thedepth directly and in a robust manner, without estimatingthe normals and the albedo in a first step. We then showhow to use this approach in order to refine the depth mapobtained with a RGB-D sensor, by integrating on the sensorthree colored LEDs allowing us to apply colored photome-tric stereo on the RGB image.

AB - We propose a new approach of photometric stereo with ani-sotropic pointwise sources such as LEDs. This approach isboth differential and variational, in order to recover thedepth directly and in a robust manner, without estimatingthe normals and the albedo in a first step. We then showhow to use this approach in order to refine the depth mapobtained with a RGB-D sensor, by integrating on the sensorthree colored LEDs allowing us to apply colored photome-tric stereo on the RGB image.

M3 - Paper

T2 - Rencontres Francophones d'Intelligence Artificielle 2016

Y2 - 27 June 2016 through 1 July 2016

ER -