Accelerated multi-view stereo using parallel processing capababilities of the GPUS
THALES Security Solutions and Services, 1 Rue du General de Gaulle, 95523 Cergy-Pontoise, France
3DTV Conference: The True Vision – Capture, Transmission and Display of 3D Video, 2009
@conference{moslah2009accelerated,
title={Accelerated multi-view stereo using parallel processing capababilities of the GPUs},
author={Moslah, O. and Valles-Such, A. and Guitteny, V. and Couvet, S. and Philipp-Foliguet, S.},
booktitle={3DTV Conference: The True Vision-Capture, Transmission and Display of 3D Video, 2009},
pages={1–4},
year={2009},
organization={IEEE}
}
This paper presents an accelerated implementation of a multi-view stereo pipeline using parallel processing capabilities of the GPUs. Our system takes as input a set of calibrated photographs and produces a textured 3D mesh of the scene. The pipeline is divided into three parts: dense stereo matching, multi-view correspondence linking and 3D model generation. First, we use a combined vertical aggregation and dynamic programming (DP) scheme to produce disparity maps between pairs of photographs. Then, the depth maps are computed using a multi-view correspondence linking algorithm. Finally, we use a Delaunay triangulation algorithm and texture mapping to produce the 3D model of the scene.
April 18, 2011 by hgpu
