Online rapid prototyping of 3D objects using GPU-based 3D cloud computing: Application to 3D face modelling
Department of Computer Science, The University of Auckland, New Zealan
IAPR Conference on Machine Vision Applications (MVA2011), 2011
@article{nguyen2011online,
title={Online rapid prototyping of 3D objects using GPU-based 3D cloud computing: Application to 3D face modelling},
author={Nguyen, M. and Delmas, P. and Gimel’farb, G. and Chan, YH and Strozzi, A.G. and Woodward, A.},
year={2011}
}
An on-line web application that interacts with an Internet user’s 3D webcam (e.g. Minoru stereo webcamera) is described. The application instantly captures and processes stereo images to retrieve 3D object coordinates for further 3D modelling tasks. It offers on-demand semi-automatic camera calibration and automatic image pair rectification for further stereo matching operations. The reconstructed 3D objects are displayed on a Java3D panel with free observational navigation. The extracted 3D coordinates are saved for further use in a professional 3D graphical tool such as Blender. Real objects geometry was extracted from stereo images processed by a custom matching algorithm using either server-side (CUDA on high-end server) or client-side (Java Applet on user computer) processing. Specific applications in 3D face animation are described. Depth map and associated face texture were parsed into our animated 3D face system to rapidly generate specific facial expressions. The described application is portable, easy to set up and operate. Its current version is available at: http://www.ivs.auckland.ac.nz.
December 23, 2011 by hgpu
