Scale invariant texture analysis using multi-scale local autocorrelation features

Yousun Kang, Kenichi Morooka, Hiroshi Nagahashi

Research output: Contribution to journalArticle

10 Citations (Scopus)

Abstract

We have developed a new framework for scale invariant texture analysis using multi-scale local autocorrelation features. The multi-scale features are made of concatenated feature vectors of different scales, which are calculated from higher-order local autocorrelation functions. To classify different types of textures among the given test images, a linear discriminant classifier (LDA) is employed in the multi-scale feature space. The scale rate of test patterns in their reduced subspace can also be estimated by principal component analysis (PCA). This subspace represents the scale variation of each scale step by principal components of a training texture image. Experimental results show that the proposed method is effective in not only scale invariant texture classification including estimation of scale rate, but also scale invariant segmentation of 2D image for scene analysis.

Original languageEnglish
Pages (from-to)363-373
Number of pages11
JournalUnknown Journal
Volume3459
Publication statusPublished - 2005
Externally publishedYes

Fingerprint

Autocorrelation
autocorrelation
Textures
texture
Principal component analysis
segmentation
principal component analysis
Classifiers
analysis
rate
test

All Science Journal Classification (ASJC) codes

  • Computer Science (miscellaneous)

Cite this

Scale invariant texture analysis using multi-scale local autocorrelation features. / Kang, Yousun; Morooka, Kenichi; Nagahashi, Hiroshi.

In: Unknown Journal, Vol. 3459, 2005, p. 363-373.

Research output: Contribution to journalArticle

@article{747ba26797d248babb823d2bccd9e0f7,
title = "Scale invariant texture analysis using multi-scale local autocorrelation features",
abstract = "We have developed a new framework for scale invariant texture analysis using multi-scale local autocorrelation features. The multi-scale features are made of concatenated feature vectors of different scales, which are calculated from higher-order local autocorrelation functions. To classify different types of textures among the given test images, a linear discriminant classifier (LDA) is employed in the multi-scale feature space. The scale rate of test patterns in their reduced subspace can also be estimated by principal component analysis (PCA). This subspace represents the scale variation of each scale step by principal components of a training texture image. Experimental results show that the proposed method is effective in not only scale invariant texture classification including estimation of scale rate, but also scale invariant segmentation of 2D image for scene analysis.",
author = "Yousun Kang and Kenichi Morooka and Hiroshi Nagahashi",
year = "2005",
language = "English",
volume = "3459",
pages = "363--373",
journal = "Quaternary International",
issn = "1040-6182",
publisher = "Elsevier Limited",

}

TY - JOUR

T1 - Scale invariant texture analysis using multi-scale local autocorrelation features

AU - Kang, Yousun

AU - Morooka, Kenichi

AU - Nagahashi, Hiroshi

PY - 2005

Y1 - 2005

N2 - We have developed a new framework for scale invariant texture analysis using multi-scale local autocorrelation features. The multi-scale features are made of concatenated feature vectors of different scales, which are calculated from higher-order local autocorrelation functions. To classify different types of textures among the given test images, a linear discriminant classifier (LDA) is employed in the multi-scale feature space. The scale rate of test patterns in their reduced subspace can also be estimated by principal component analysis (PCA). This subspace represents the scale variation of each scale step by principal components of a training texture image. Experimental results show that the proposed method is effective in not only scale invariant texture classification including estimation of scale rate, but also scale invariant segmentation of 2D image for scene analysis.

AB - We have developed a new framework for scale invariant texture analysis using multi-scale local autocorrelation features. The multi-scale features are made of concatenated feature vectors of different scales, which are calculated from higher-order local autocorrelation functions. To classify different types of textures among the given test images, a linear discriminant classifier (LDA) is employed in the multi-scale feature space. The scale rate of test patterns in their reduced subspace can also be estimated by principal component analysis (PCA). This subspace represents the scale variation of each scale step by principal components of a training texture image. Experimental results show that the proposed method is effective in not only scale invariant texture classification including estimation of scale rate, but also scale invariant segmentation of 2D image for scene analysis.

UR - http://www.scopus.com/inward/record.url?scp=24644479763&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=24644479763&partnerID=8YFLogxK

M3 - Article

VL - 3459

SP - 363

EP - 373

JO - Quaternary International

JF - Quaternary International

SN - 1040-6182

ER -