Increasing Computational Redundancy of Digital Images via Multiresolutional Matching

  • Authors:

  • Farzad Khalvati;Hamid R. Tizhoosh;Arsen R. Hajian

  • Affiliations:

  • Department of Systems Design Engineering, University of Waterloo, Waterloo, Canada;Department of Systems Design Engineering, University of Waterloo, Waterloo, Canada;Department of Systems Design Engineering, University of Waterloo, Waterloo, Canada

  • Venue:
  • ICIAR '09 Proceedings of the 6th International Conference on Image Analysis and Recognition
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

Computational redundancy of an image represents the amount of computations that can be skipped to improve performance. In order to calculate and exploit the computational redundancy of an image, a similarity measure is required to identify similar neighborhoods of pixels in the image. In this paper, we present two similarity measures: a position-invariant histogram-based measure and a rotation-invariant multiresolutional histogrambased measure. We demonstrate that by using the position-invariant and rotation-invariant similarity measures, on average, the computational redundancy of natural images increases by 34% and 28%, respectively, in comparison to the basic similarity measure. The increase in computational redundancy can lead to further performance improvement. For a case study, the average increase in actual speedup is 211% and 35% for position-invariant and rotation-invariant similarity measures, respectively.