Rotation- and scale-independent pattern recognation through optimization

Thomas L. Hemminger, Carlos A. Pomalaz-Raez

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

This paper presents a two-dimensional pattern recognition paradigm, which is independent of rotation and scale. It uses a Hopfield neural network with a configuration similar to that employed for the traveling salesman problem (TSP). Enhancements to the original Hopfield design are included which are based on the eigenvalues of the connection matrix and on modifications suggested by simulations. The goal of the method is to determine the underlying linear transformation between a test image and an observed input pattern by minimizing the energy of a Hopfield network. This procedure is easy to implement and is well suited for sparse binary images, e.g. skeletalized representations or the output from vertex detection algorithms. In this work the effects of noise are studied both analytically and through Monte Carlo techniques yielding some useful performance guidelines and greatly increasing the utility of the method. Experiments show that this technique performs successfully on a variety of rotated and scaled images and is robust against additive noise.

Original languageEnglish (US)
Pages (from-to)487-495
Number of pages9
JournalPattern Recognition
Volume29
Issue number3
DOIs
StatePublished - Mar 1996

Fingerprint

Hopfield neural networks
Linear transformations
Traveling salesman problem
Binary images
Additive noise
Pattern recognition
Experiments

All Science Journal Classification (ASJC) codes

  • Software
  • Signal Processing
  • Computer Vision and Pattern Recognition
  • Artificial Intelligence

Cite this

@article{bc5ef49ab8b343a29007ad6e23f19220,
title = "Rotation- and scale-independent pattern recognation through optimization",
abstract = "This paper presents a two-dimensional pattern recognition paradigm, which is independent of rotation and scale. It uses a Hopfield neural network with a configuration similar to that employed for the traveling salesman problem (TSP). Enhancements to the original Hopfield design are included which are based on the eigenvalues of the connection matrix and on modifications suggested by simulations. The goal of the method is to determine the underlying linear transformation between a test image and an observed input pattern by minimizing the energy of a Hopfield network. This procedure is easy to implement and is well suited for sparse binary images, e.g. skeletalized representations or the output from vertex detection algorithms. In this work the effects of noise are studied both analytically and through Monte Carlo techniques yielding some useful performance guidelines and greatly increasing the utility of the method. Experiments show that this technique performs successfully on a variety of rotated and scaled images and is robust against additive noise.",
author = "Hemminger, {Thomas L.} and Pomalaz-Raez, {Carlos A.}",
year = "1996",
month = "3",
doi = "10.1016/0031-3203(95)00100-X",
language = "English (US)",
volume = "29",
pages = "487--495",
journal = "Pattern Recognition",
issn = "0031-3203",
publisher = "Elsevier Limited",
number = "3",

}

Rotation- and scale-independent pattern recognation through optimization. / Hemminger, Thomas L.; Pomalaz-Raez, Carlos A.

In: Pattern Recognition, Vol. 29, No. 3, 03.1996, p. 487-495.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Rotation- and scale-independent pattern recognation through optimization

AU - Hemminger, Thomas L.

AU - Pomalaz-Raez, Carlos A.

PY - 1996/3

Y1 - 1996/3

N2 - This paper presents a two-dimensional pattern recognition paradigm, which is independent of rotation and scale. It uses a Hopfield neural network with a configuration similar to that employed for the traveling salesman problem (TSP). Enhancements to the original Hopfield design are included which are based on the eigenvalues of the connection matrix and on modifications suggested by simulations. The goal of the method is to determine the underlying linear transformation between a test image and an observed input pattern by minimizing the energy of a Hopfield network. This procedure is easy to implement and is well suited for sparse binary images, e.g. skeletalized representations or the output from vertex detection algorithms. In this work the effects of noise are studied both analytically and through Monte Carlo techniques yielding some useful performance guidelines and greatly increasing the utility of the method. Experiments show that this technique performs successfully on a variety of rotated and scaled images and is robust against additive noise.

AB - This paper presents a two-dimensional pattern recognition paradigm, which is independent of rotation and scale. It uses a Hopfield neural network with a configuration similar to that employed for the traveling salesman problem (TSP). Enhancements to the original Hopfield design are included which are based on the eigenvalues of the connection matrix and on modifications suggested by simulations. The goal of the method is to determine the underlying linear transformation between a test image and an observed input pattern by minimizing the energy of a Hopfield network. This procedure is easy to implement and is well suited for sparse binary images, e.g. skeletalized representations or the output from vertex detection algorithms. In this work the effects of noise are studied both analytically and through Monte Carlo techniques yielding some useful performance guidelines and greatly increasing the utility of the method. Experiments show that this technique performs successfully on a variety of rotated and scaled images and is robust against additive noise.

UR - http://www.scopus.com/inward/record.url?scp=0030107387&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0030107387&partnerID=8YFLogxK

U2 - 10.1016/0031-3203(95)00100-X

DO - 10.1016/0031-3203(95)00100-X

M3 - Article

AN - SCOPUS:0030107387

VL - 29

SP - 487

EP - 495

JO - Pattern Recognition

JF - Pattern Recognition

SN - 0031-3203

IS - 3

ER -