Continuous Projection for Fast L1 Reconstruction

Information

• Publication Type: Journal Paper with Conference Talk
• Workgroup(s)/Project(s):
  • Harvest4D - Harvesting Dynamic 3D Worlds from Commodity Sensor Clouds
• Date: August 2014
• Journal: ACM Transactions on Graphics (Proc. of ACM SIGGRAPH 2014)
• Volume: 33
• Pages (to): 47:13
• Pages (from): 47:1
• Number: 4
• Location: Vancouver, Canada
• Lecturer: Reinhold Preiner
• ISSN: 0730-0301
• Event: ACM SIGGRAPH 2014
• DOI: 10.1145/2601097.2601172
• Date (to): 14. August 2014
• Date (from): 10. August 2014
• Keywords: point set, Gaussian mixture, Hierarchical EM, upsampling, dynamic reconstruction, L1 reconstruction

Abstract

With better and faster acquisition devices comes a demand for fast robust reconstruction algorithms, but no L1-based technique has been fast enough for online use so far. In this paper, we present a novel continuous formulation of the weighted locally optimal projection (WLOP) operator based on a Gaussian mixture describing the input point density. Our method is up to 7 times faster than an optimized GPU implementation of WLOP, and achieves interactive frame rates for moderately sized point clouds. We give a comprehensive quality analysis showing that our continuous operator achieves a generally higher reconstruction quality than its discrete counterpart. Additionally, we show how to apply our continuous formulation to spherical mixtures of normal directions, to also achieve a fast robust normal reconstruction.

Project Page: https://www.cg.tuwien.ac.at/~preiner/projects/clop/

Additional Files and Images

Weblinks

• DOI: 10.1145/2601097.2601172

BibTeX

@article{preiner2014clop,
  title =      "Continuous Projection for Fast L1 Reconstruction",
  author =     "Reinhold Preiner and Oliver Mattausch and Murat Arikan and
               Renato Pajarola and Michael Wimmer",
  year =       "2014",
  abstract =   "With better and faster acquisition devices comes a demand
               for fast robust reconstruction algorithms, but no L1-based
               technique has been fast enough for online use so far. In
               this paper, we present a novel continuous formulation of the
               weighted locally optimal projection (WLOP) operator based on
               a Gaussian mixture describing the input point density. Our
               method is up to 7 times faster than an optimized GPU
               implementation of WLOP, and achieves interactive frame rates
               for moderately sized point clouds. We give a comprehensive
               quality analysis showing that our continuous operator
               achieves a generally higher reconstruction quality than its
               discrete counterpart. Additionally, we show how to apply our
               continuous formulation to spherical mixtures of normal
               directions, to also achieve a fast robust normal
               reconstruction.  Project Page:
               https://www.cg.tuwien.ac.at/~preiner/projects/clop/",
  month =      aug,
  journal =    "ACM Transactions on Graphics (Proc. of ACM SIGGRAPH 2014)",
  volume =     "33",
  number =     "4",
  issn =       "0730-0301",
  doi =        "10.1145/2601097.2601172",
  pages =      "47:1--47:13",
  keywords =   "point set, Gaussian mixture, Hierarchical EM, upsampling,
               dynamic reconstruction, L1 reconstruction",
  URL =        "https://www.cg.tuwien.ac.at/research/publications/2014/preiner2014clop/",
}