Robust Watermarking of Point-Sampled Geometry

  • Authors:
  • Daniel Cotting;Tim Weyrich;Mark Pauly;Markus Gross

  • Affiliations:
  • Swiss Federal Institute of Technology;Swiss Federal Institute of Technology;Stanford University;Swiss Federal Institute of Technology

  • Venue:
  • SMI '04 Proceedings of the Shape Modeling International 2004
  • Year:
  • 2004

Quantified Score

Hi-index 0.00

Visualization

Abstract

We present a new scheme for digital watermarking ofpoint-sampled geometry based on spectral analysis. Byextending existing algorithms designed for polygonal datato unstructured point clouds, our method is particularlysuited for scanned models, where the watermark can bedirectly embedded in the raw data obtained from the 3Dacquisition device. To handle large data sets efficiently, weapply a fast hierarchical clustering algorithm that partitionsthe model into a set of patches. Each patch is mappedinto the space of eigenfunctions of an approximate Laplacianoperator to obtain a decomposition of the patch surfaceinto discrete frequency bands. The watermark is thenembedded into the low frequency components to minimizevisual artifacts in the model geometry. During extraction,the target model is resampled at optimal resolution usingan MLS projection. After extracting a watermark from thismodel, the corresponding bit stream is analyzed using statisticalmethods based on correlation. We have applied ourmethod to a number of point-sampled models of differentgeometric and topological complexity. These experimentsshow that our watermarking scheme is robust againstnumerous attacks, including low-pass filtering, resampling,affine transformations, cropping, additive randomnoise, and combinations of the above.