Higher Order Prediction for Geometry Compression

  • Authors:
  • Stefan Gumhold;Rachida Amjoun

  • Affiliations:
  • -;-

  • Venue:
  • SMI '03 Proceedings of the Shape Modeling International 2003
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

A lot of techniques have been developed for the encodingof triangular meshes as this is a widely used representationfor the description of surface models. Although methodsfor the encoding of the neighbor information, the connectivity,are near optimal, there is still room for better en-codingsof vertex locations, the geometry. Our geometryencoding strategy follows the predictive coding paradigm,which is based on a region growing encoding order. Onlythe delta vectors between original and predicted locationsare encoded in a local coordinate system, which splits intotwo tangential and one normal component.In this paper we introduce so-called higher order predictionfor an improved encoding of the normal component. Wefirst encode the tangential components with parallelogramprediction. Then we fit a higher order surface to the so farencoded geometry. As we encode the normal component asa bending angle, it is found by intersecting the higher ordersurface with the circle defined by the tangential components.Experimental results show that our strategy allowssaving one bit per vertex for the normal component independentof the tangential prediction rule used.