An efficient fuzzy algorithm for aligning shapes under affine transformations

Zhong Xue, Dinggang Shen, Khwang Eam Teoh

Research output: Contribution to journalArticle

16 Citations (Scopus)

Abstract

A fuzzy algorithm for aligning object shapes under affine transformations is proposed in this paper. The algorithm, with the name of fuzzy alignment algorithm (FAA), extends Marques' algorithm to affine transformations. It can efficiently estimate the point correspondence and the relevant affine transformational parameters between the feature points of the object shape and the reference shape. In this algorithm, the fuzzy point-correspondence degrees are used to describe an uncertainty point assignment, then both the parameters of the affine transformation and the fuzzy correspondence degrees are iteratively calculated by minimizing a constrained fuzzy objective function. To prevent FAA from sinking into local minimum when the shapes are greatly deformed, an initialization method based on affine invariants is designed. Comparing to the eigenvector method, the effectiveness and robustness of the proposed algorithm is investigated with a sensitivity study based on randomly generated points. At last, good performance of FAA is illustrated with several experiments on aligning digits and object shapes.

Original languageEnglish
Pages (from-to)1171-1180
Number of pages10
JournalPattern Recognition
Volume34
Issue number6
DOIs
Publication statusPublished - 2001 Jun 1
Externally publishedYes

Fingerprint

Eigenvalues and eigenfunctions
Experiments
Uncertainty

Keywords

  • Affine invariant
  • Affine transformation
  • Correspondence
  • Fuzzy clustering
  • Pose estimation
  • Shape alignment

ASJC Scopus subject areas

  • Computer Vision and Pattern Recognition
  • Signal Processing
  • Electrical and Electronic Engineering

Cite this

An efficient fuzzy algorithm for aligning shapes under affine transformations. / Xue, Zhong; Shen, Dinggang; Teoh, Khwang Eam.

In: Pattern Recognition, Vol. 34, No. 6, 01.06.2001, p. 1171-1180.

Research output: Contribution to journalArticle

Xue, Zhong ; Shen, Dinggang ; Teoh, Khwang Eam. / An efficient fuzzy algorithm for aligning shapes under affine transformations. In: Pattern Recognition. 2001 ; Vol. 34, No. 6. pp. 1171-1180.
@article{84bc2e4aaeeb4a20a7cbd4cde9c5e259,
title = "An efficient fuzzy algorithm for aligning shapes under affine transformations",
abstract = "A fuzzy algorithm for aligning object shapes under affine transformations is proposed in this paper. The algorithm, with the name of fuzzy alignment algorithm (FAA), extends Marques' algorithm to affine transformations. It can efficiently estimate the point correspondence and the relevant affine transformational parameters between the feature points of the object shape and the reference shape. In this algorithm, the fuzzy point-correspondence degrees are used to describe an uncertainty point assignment, then both the parameters of the affine transformation and the fuzzy correspondence degrees are iteratively calculated by minimizing a constrained fuzzy objective function. To prevent FAA from sinking into local minimum when the shapes are greatly deformed, an initialization method based on affine invariants is designed. Comparing to the eigenvector method, the effectiveness and robustness of the proposed algorithm is investigated with a sensitivity study based on randomly generated points. At last, good performance of FAA is illustrated with several experiments on aligning digits and object shapes.",
keywords = "Affine invariant, Affine transformation, Correspondence, Fuzzy clustering, Pose estimation, Shape alignment",
author = "Zhong Xue and Dinggang Shen and Teoh, {Khwang Eam}",
year = "2001",
month = "6",
day = "1",
doi = "10.1016/S0031-3203(00)00067-4",
language = "English",
volume = "34",
pages = "1171--1180",
journal = "Pattern Recognition",
issn = "0031-3203",
publisher = "Elsevier Limited",
number = "6",

}

TY - JOUR

T1 - An efficient fuzzy algorithm for aligning shapes under affine transformations

AU - Xue, Zhong

AU - Shen, Dinggang

AU - Teoh, Khwang Eam

PY - 2001/6/1

Y1 - 2001/6/1

N2 - A fuzzy algorithm for aligning object shapes under affine transformations is proposed in this paper. The algorithm, with the name of fuzzy alignment algorithm (FAA), extends Marques' algorithm to affine transformations. It can efficiently estimate the point correspondence and the relevant affine transformational parameters between the feature points of the object shape and the reference shape. In this algorithm, the fuzzy point-correspondence degrees are used to describe an uncertainty point assignment, then both the parameters of the affine transformation and the fuzzy correspondence degrees are iteratively calculated by minimizing a constrained fuzzy objective function. To prevent FAA from sinking into local minimum when the shapes are greatly deformed, an initialization method based on affine invariants is designed. Comparing to the eigenvector method, the effectiveness and robustness of the proposed algorithm is investigated with a sensitivity study based on randomly generated points. At last, good performance of FAA is illustrated with several experiments on aligning digits and object shapes.

AB - A fuzzy algorithm for aligning object shapes under affine transformations is proposed in this paper. The algorithm, with the name of fuzzy alignment algorithm (FAA), extends Marques' algorithm to affine transformations. It can efficiently estimate the point correspondence and the relevant affine transformational parameters between the feature points of the object shape and the reference shape. In this algorithm, the fuzzy point-correspondence degrees are used to describe an uncertainty point assignment, then both the parameters of the affine transformation and the fuzzy correspondence degrees are iteratively calculated by minimizing a constrained fuzzy objective function. To prevent FAA from sinking into local minimum when the shapes are greatly deformed, an initialization method based on affine invariants is designed. Comparing to the eigenvector method, the effectiveness and robustness of the proposed algorithm is investigated with a sensitivity study based on randomly generated points. At last, good performance of FAA is illustrated with several experiments on aligning digits and object shapes.

KW - Affine invariant

KW - Affine transformation

KW - Correspondence

KW - Fuzzy clustering

KW - Pose estimation

KW - Shape alignment

UR - http://www.scopus.com/inward/record.url?scp=0035372531&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0035372531&partnerID=8YFLogxK

U2 - 10.1016/S0031-3203(00)00067-4

DO - 10.1016/S0031-3203(00)00067-4

M3 - Article

AN - SCOPUS:0035372531

VL - 34

SP - 1171

EP - 1180

JO - Pattern Recognition

JF - Pattern Recognition

SN - 0031-3203

IS - 6

ER -