Information

Abstract

Abstract 3D surface reconstruction from point clouds is a key step in areas such as content creation, archaeology, digital cultural heritage and engineering. Current approaches either try to optimize a non-data-driven surface representation to fit the points, or learn a data-driven prior over the distribution of commonly occurring surfaces and how they correlate with potentially noisy point clouds. Data-driven methods enable robust handling of noise and typically either focus on a global or a local prior, which trade-off between robustness to noise on the global end and surface detail preservation on the local end. We propose PPSurf as a method that combines a global prior based on point convolutions and a local prior based on processing local point cloud patches. We show that this approach is robust to noise while recovering surface details more accurately than the current state-of-the-art. Our source code, pre-trained model and dataset are available at https://github.com/cg-tuwien/ppsurf.

Additional Files and Images

Additional images and videos

teaser: PPSurf teaser with comparison teaser: PPSurf teaser with comparison

Additional files

paper_repro: Paper Reproduction Code and Models
Note: use the repo instead of this messy code paper_repro: Paper Reproduction Code and Models Note: use the repo instead of this messy code
paper: PPSurf (ArXiv Version) paper: PPSurf (ArXiv Version)
ppsurf_50nn_model: PPSurf 50NN Model Checkpoint ppsurf_50nn_model: PPSurf 50NN Model Checkpoint
ppsurf_50nn_results: PPSurf 50NN Results (Meshes and Tables) ppsurf_50nn_results: PPSurf 50NN Results (Meshes and Tables)
slides_eg24_pdf: Eurographics 2024 Slides (PDF) slides_eg24_pdf: Eurographics 2024 Slides (PDF)
slides_eg24: Eurographics 2024 Slides slides_eg24: Eurographics 2024 Slides
testsets: Testsets (ABC, Famous, Thingi10k) testsets: Testsets (ABC, Famous, Thingi10k)
trainset: ABC Training Set trainset: ABC Training Set

Weblinks

BibTeX

@article{erler_2024_ppsurf,
  title =      "PPSurf: Combining Patches and Point Convolutions for
               Detailed Surface Reconstruction",
  author =     "Philipp Erler and Lizeth Fuentes-Perez and Pedro
               Hermosilla-Casajus and Paul Guerrero and Renato Pajarola and
               Michael Wimmer",
  year =       "2024",
  abstract =   "Abstract 3D surface reconstruction from point clouds is a
               key step in areas such as content creation, archaeology,
               digital cultural heritage and engineering. Current
               approaches either try to optimize a non-data-driven surface
               representation to fit the points, or learn a data-driven
               prior over the distribution of commonly occurring surfaces
               and how they correlate with potentially noisy point clouds.
               Data-driven methods enable robust handling of noise and
               typically either focus on a global or a local prior, which
               trade-off between robustness to noise on the global end and
               surface detail preservation on the local end. We propose
               PPSurf as a method that combines a global prior based on
               point convolutions and a local prior based on processing
               local point cloud patches. We show that this approach is
               robust to noise while recovering surface details more
               accurately than the current state-of-the-art. Our source
               code, pre-trained model and dataset are available at
               https://github.com/cg-tuwien/ppsurf.",
  month =      jan,
  journal =    "Computer Graphics Forum",
  volume =     "43",
  number =     "1",
  issn =       "1467-8659",
  doi =        "https://doi.org/10.1111/cgf.15000",
  pages =      "12",
  publisher =  "WILEY",
  pages =      "tbd--tbd",
  keywords =   "modeling, surface reconstruction",
  URL =        "https://www.cg.tuwien.ac.at/research/publications/2024/erler_2024_ppsurf/",
}