Rehashing for Bayesian Geometric Hashing

Michael Lifshits, Ilya Blayvas, Roman Goldenberg, Ehud Rivlin, and Michael Rudzsky.
Rehashing for Bayesian Geometric Hashing.
In ICPR (3), 99-102, 2004

Online Version

A pdf version is available for download.

Abstract

Geometric hashing is a model-based recognition technique based on matching of transformation-invariant object representations stored in a hash table. In the last decade a number of enhancements have been suggested to the basic method improving its performance and reliability. One of the important enhancements is rehashing, improving the computational performance by dealing with the problem of non-uniform occupancy of hash bins. However the proposed rehashing schemes aim to redistribute the hash entries uniformly, which is not appropriate for Bayesian approach, another enhancement optimizing the recognition rate in presence of noise. In this paper we derive the rehashing for Bayesian voting scheme, thus improving the computational performance by minimizing the hash table size and the number of bins accessed, while maintaining optimal recognition rate.

Co-authors

Bibtex Entry

@inproceedings{LifshitsBGRR04i,
  title = {Rehashing for Bayesian Geometric Hashing.},
  author = {Michael Lifshits and Ilya Blayvas and Roman Goldenberg and Ehud Rivlin and Michael Rudzsky},
  year = {2004},
  booktitle = {ICPR (3)},
  pages = {99-102},
  abstract = {Geometric hashing is a model-based recognition technique based on matching of transformation-invariant object representations stored in a hash table. In the last decade a number of enhancements have been suggested to the basic method improving its performance and reliability. One of the important enhancements is rehashing, improving the computational performance by dealing with the problem of non-uniform occupancy of hash bins. However the proposed rehashing schemes aim to redistribute the hash entries uniformly, which is not appropriate for Bayesian approach, another enhancement optimizing the recognition rate in presence of noise. In this paper we derive the rehashing for Bayesian voting scheme, thus improving the computational performance by minimizing the hash table size and the number of bins accessed, while maintaining optimal recognition rate.}
}