Exploiting repeated patterns for efficient compression of massive models

  • Authors:
  • Kangying Cai;Wencheng Wang;Zhibo Chen;QuQing Chen;Jun Teng

  • Affiliations:
  • Thomson C.R. Beijing;Chinese Academy of Sciences;Thomson C.R. Beijing;Thomson C.R. Beijing;Thomson C.R. Beijing

  • Venue:
  • Proceedings of the 8th International Conference on Virtual Reality Continuum and its Applications in Industry
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

We propose an efficient compression algorithm for massive models, which consist of a large number of small to medium sized connected components. It is based on efficiently exploiting repetitive patterns in the input model. Compared with [Shikhare et al. 2001], the state-of-the-art work for utilizing repetitive patterns for compressing massive models, our new algorithm is more efficient on detecting repeated components and compressing transformations, so that it achieves a considerably higher compression ratio. By recognizing instances repeating in various scalings, which is missed in [Shikhare et al. 2001], we can reduce 40% of the repetitive patterns on average. By aligning components based on their quadric error metrics, we overcome the limitation of [Shikhare et al. 2001] that may regard two components as instances of the same pattern when they have same vertex position but different connectivity. For transformations of all instances, we give an efficient compression scheme, which can save 40% storage on average in comparison with gzip, the popular compression software used in [Shikhare et al. 2001]. By experiments, our new algorithm can achieve compression ratio at about 4% of the raw data, and gain around 40% over [Shikhare et al. 2001] on average.