Optimal Gabor Filters for Texture Segmentation

Research output: Contribution to journalArticle

317 Citations (Scopus)

Abstract

Texture segmentation involves subdividing an image into differently textured regions. Many texture-segmentation schemes are based on a filter-bank model, where the filters, called Gabor filters, are derived from Gabor elementary functions. The goal is to transform texture differences into detectable filter-output discontinuities at texture boundaries. By locating these discontinuities, one can segment the image into differently textured regions. Distinct discontinuities occur, however, only if the Gabor filter parameters are suitably chosen. Some previous analysis has shown how to design filters for discriminating simple textures. Designing filters for more general natural textures, though, has largely been done ad hoc. We have devised a more rigorously based method for designing Gabor filters. It assumes that an image contains two different textures and that prototype samples of the textures are given a priori. We argue that Gabor filter outputs can be modeled as Rician random variables (often approximated well as Gaussian rv’s) and develop a decision-theoretic algorithm for selecting optimal filter parameters. To improve segmentations for difficult texture pairs, we also propose a multiple-filter segmentation scheme, motivated by the Rician model. Experimental results indicate that our method is superior to previous methods in providing useful Gabor filters for a wide range of texture pairs.

Original languageEnglish (US)
Pages (from-to)947-964
Number of pages18
JournalIEEE Transactions on Image Processing
Volume4
Issue number7
DOIs
StatePublished - Jan 1 1995

Fingerprint

Gabor filters
Textures
Filter banks
Random variables

All Science Journal Classification (ASJC) codes

  • Software
  • Computer Graphics and Computer-Aided Design

Cite this

@article{2cb1c82173cb49ab895286abe563afde,
title = "Optimal Gabor Filters for Texture Segmentation",
abstract = "Texture segmentation involves subdividing an image into differently textured regions. Many texture-segmentation schemes are based on a filter-bank model, where the filters, called Gabor filters, are derived from Gabor elementary functions. The goal is to transform texture differences into detectable filter-output discontinuities at texture boundaries. By locating these discontinuities, one can segment the image into differently textured regions. Distinct discontinuities occur, however, only if the Gabor filter parameters are suitably chosen. Some previous analysis has shown how to design filters for discriminating simple textures. Designing filters for more general natural textures, though, has largely been done ad hoc. We have devised a more rigorously based method for designing Gabor filters. It assumes that an image contains two different textures and that prototype samples of the textures are given a priori. We argue that Gabor filter outputs can be modeled as Rician random variables (often approximated well as Gaussian rv’s) and develop a decision-theoretic algorithm for selecting optimal filter parameters. To improve segmentations for difficult texture pairs, we also propose a multiple-filter segmentation scheme, motivated by the Rician model. Experimental results indicate that our method is superior to previous methods in providing useful Gabor filters for a wide range of texture pairs.",
author = "Dennis Dunn and Higgins, {William Evan}",
year = "1995",
month = "1",
day = "1",
doi = "10.1109/83.392336",
language = "English (US)",
volume = "4",
pages = "947--964",
journal = "IEEE Transactions on Image Processing",
issn = "1057-7149",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "7",

}

Optimal Gabor Filters for Texture Segmentation. / Dunn, Dennis; Higgins, William Evan.

In: IEEE Transactions on Image Processing, Vol. 4, No. 7, 01.01.1995, p. 947-964.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Optimal Gabor Filters for Texture Segmentation

AU - Dunn, Dennis

AU - Higgins, William Evan

PY - 1995/1/1

Y1 - 1995/1/1

N2 - Texture segmentation involves subdividing an image into differently textured regions. Many texture-segmentation schemes are based on a filter-bank model, where the filters, called Gabor filters, are derived from Gabor elementary functions. The goal is to transform texture differences into detectable filter-output discontinuities at texture boundaries. By locating these discontinuities, one can segment the image into differently textured regions. Distinct discontinuities occur, however, only if the Gabor filter parameters are suitably chosen. Some previous analysis has shown how to design filters for discriminating simple textures. Designing filters for more general natural textures, though, has largely been done ad hoc. We have devised a more rigorously based method for designing Gabor filters. It assumes that an image contains two different textures and that prototype samples of the textures are given a priori. We argue that Gabor filter outputs can be modeled as Rician random variables (often approximated well as Gaussian rv’s) and develop a decision-theoretic algorithm for selecting optimal filter parameters. To improve segmentations for difficult texture pairs, we also propose a multiple-filter segmentation scheme, motivated by the Rician model. Experimental results indicate that our method is superior to previous methods in providing useful Gabor filters for a wide range of texture pairs.

AB - Texture segmentation involves subdividing an image into differently textured regions. Many texture-segmentation schemes are based on a filter-bank model, where the filters, called Gabor filters, are derived from Gabor elementary functions. The goal is to transform texture differences into detectable filter-output discontinuities at texture boundaries. By locating these discontinuities, one can segment the image into differently textured regions. Distinct discontinuities occur, however, only if the Gabor filter parameters are suitably chosen. Some previous analysis has shown how to design filters for discriminating simple textures. Designing filters for more general natural textures, though, has largely been done ad hoc. We have devised a more rigorously based method for designing Gabor filters. It assumes that an image contains two different textures and that prototype samples of the textures are given a priori. We argue that Gabor filter outputs can be modeled as Rician random variables (often approximated well as Gaussian rv’s) and develop a decision-theoretic algorithm for selecting optimal filter parameters. To improve segmentations for difficult texture pairs, we also propose a multiple-filter segmentation scheme, motivated by the Rician model. Experimental results indicate that our method is superior to previous methods in providing useful Gabor filters for a wide range of texture pairs.

UR - http://www.scopus.com/inward/record.url?scp=0029343445&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0029343445&partnerID=8YFLogxK

U2 - 10.1109/83.392336

DO - 10.1109/83.392336

M3 - Article

C2 - 18290045

AN - SCOPUS:0029343445

VL - 4

SP - 947

EP - 964

JO - IEEE Transactions on Image Processing

JF - IEEE Transactions on Image Processing

SN - 1057-7149

IS - 7

ER -