Pattern matching by scan-converting polygons

Research output: Contribution to journalConference article

3 Citations (Scopus)

Abstract

Pattern matching is one of the well-known pattern recognition techniques. When using points as matching features, a pattern matching problem becomes a point pattern matching problem. This paper proposes a novel point pattern matching algorithm that searches transformation space by transformation sampling. The algorithm defines a constraint set (a polygonal region in transformation space) for each possible pairing of a template point and a target point. Under constrained polynomial transformations that have no more than two parameters on each coordinate, the constraint sets and the transformation space can be represented as Cartesian products of 2D polygonal regions. The algorithm then rasterizes the transformation space into a discrete canvas and calculates the optimal matching at each sampled transformation efficiently by scan-converting polygons. Preliminary experiments on randomly generated point patterns show that the algorithm is effective and efficient. In addition, the running time of the algorithm is stable with respect to missing points.

Original languageEnglish (US)
Pages (from-to)101-110
Number of pages10
JournalProceedings of SPIE - The International Society for Optical Engineering
Volume5438
DOIs
StatePublished - Oct 26 2004
EventVisual Information Processing XIII - Orlando, FL, United States
Duration: Apr 15 2004Apr 16 2004

Fingerprint

polygons
Pattern matching
Pattern Matching
Polygon
Matching Problem
Polynomial Transformation
Pattern recognition
Feature Matching
Cartesian product
Matching Algorithm
Polynomials
Pairing
Sampling
pattern recognition
Pattern Recognition
Two Parameters
Template
polynomials
templates
sampling

Keywords

  • Geometric constraint
  • Point pattern matching
  • Scan-converting polygon
  • Transformation sampling

ASJC Scopus subject areas

  • Electronic, Optical and Magnetic Materials
  • Condensed Matter Physics
  • Computer Science Applications
  • Applied Mathematics
  • Electrical and Electronic Engineering

Cite this

Pattern matching by scan-converting polygons. / Ni, Mingtian; Reichenbach, Stephen E.

In: Proceedings of SPIE - The International Society for Optical Engineering, Vol. 5438, 26.10.2004, p. 101-110.

Research output: Contribution to journalConference article

@article{f35121b86d16492abf5761000cc31e6d,
title = "Pattern matching by scan-converting polygons",
abstract = "Pattern matching is one of the well-known pattern recognition techniques. When using points as matching features, a pattern matching problem becomes a point pattern matching problem. This paper proposes a novel point pattern matching algorithm that searches transformation space by transformation sampling. The algorithm defines a constraint set (a polygonal region in transformation space) for each possible pairing of a template point and a target point. Under constrained polynomial transformations that have no more than two parameters on each coordinate, the constraint sets and the transformation space can be represented as Cartesian products of 2D polygonal regions. The algorithm then rasterizes the transformation space into a discrete canvas and calculates the optimal matching at each sampled transformation efficiently by scan-converting polygons. Preliminary experiments on randomly generated point patterns show that the algorithm is effective and efficient. In addition, the running time of the algorithm is stable with respect to missing points.",
keywords = "Geometric constraint, Point pattern matching, Scan-converting polygon, Transformation sampling",
author = "Mingtian Ni and Reichenbach, {Stephen E}",
year = "2004",
month = "10",
day = "26",
doi = "10.1117/12.543057",
language = "English (US)",
volume = "5438",
pages = "101--110",
journal = "Proceedings of SPIE - The International Society for Optical Engineering",
issn = "0277-786X",
publisher = "SPIE",

}

TY - JOUR

T1 - Pattern matching by scan-converting polygons

AU - Ni, Mingtian

AU - Reichenbach, Stephen E

PY - 2004/10/26

Y1 - 2004/10/26

N2 - Pattern matching is one of the well-known pattern recognition techniques. When using points as matching features, a pattern matching problem becomes a point pattern matching problem. This paper proposes a novel point pattern matching algorithm that searches transformation space by transformation sampling. The algorithm defines a constraint set (a polygonal region in transformation space) for each possible pairing of a template point and a target point. Under constrained polynomial transformations that have no more than two parameters on each coordinate, the constraint sets and the transformation space can be represented as Cartesian products of 2D polygonal regions. The algorithm then rasterizes the transformation space into a discrete canvas and calculates the optimal matching at each sampled transformation efficiently by scan-converting polygons. Preliminary experiments on randomly generated point patterns show that the algorithm is effective and efficient. In addition, the running time of the algorithm is stable with respect to missing points.

AB - Pattern matching is one of the well-known pattern recognition techniques. When using points as matching features, a pattern matching problem becomes a point pattern matching problem. This paper proposes a novel point pattern matching algorithm that searches transformation space by transformation sampling. The algorithm defines a constraint set (a polygonal region in transformation space) for each possible pairing of a template point and a target point. Under constrained polynomial transformations that have no more than two parameters on each coordinate, the constraint sets and the transformation space can be represented as Cartesian products of 2D polygonal regions. The algorithm then rasterizes the transformation space into a discrete canvas and calculates the optimal matching at each sampled transformation efficiently by scan-converting polygons. Preliminary experiments on randomly generated point patterns show that the algorithm is effective and efficient. In addition, the running time of the algorithm is stable with respect to missing points.

KW - Geometric constraint

KW - Point pattern matching

KW - Scan-converting polygon

KW - Transformation sampling

UR - http://www.scopus.com/inward/record.url?scp=5544277228&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=5544277228&partnerID=8YFLogxK

U2 - 10.1117/12.543057

DO - 10.1117/12.543057

M3 - Conference article

VL - 5438

SP - 101

EP - 110

JO - Proceedings of SPIE - The International Society for Optical Engineering

JF - Proceedings of SPIE - The International Society for Optical Engineering

SN - 0277-786X

ER -