Color image segmentation based on the normal distribution and the dynamic thresholding

Seon D. Kang, Hun W. Yoo, Dong Sik Jang

Research output: Chapter in Book/Report/Conference proceedingConference contribution

4 Citations (Scopus)

Abstract

A new color image segmentation method is proposed in this paper. The proposed method is based on the human perception that in general human has attention on 3 or 4 major color objects in the image at first. Therefore, to determine the objects, three intensity distributions are constructed by sampling them randomly and sufficiently from three R, G, and B channel images. And three means are computed from three intensity distributions. Next, these steps are repeated many times to obtain three mean distribution sets. Each of these distributions comes to show normal shape based on the central limit theorem. To segment objects, each of the normal distribution is divided into 4 sections according to the standard deviation (section1 below - σ , section 2 between - σ and μ , section 3 between μ and σ , and section 4 over σ). Then sections with similar representative values are merged based on the threshold. This threshold is not chosen as constant but varies based on the difference of representative values of each section to reflect various characteristics for various images. Above merging process is iterated to reduce fine textures such as speckles remained even after the merging. Finally, segmented results of each channel images are combined to obtain a final segmentation result. The performance of the proposed method is evaluated through experiments over some images.

Original languageEnglish
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Pages372-384
Number of pages13
Volume4705 LNCS
EditionPART 1
Publication statusPublished - 2007 Dec 1
EventInternational Conference on Computational Science and its Applications, ICCSA 2007 - Kuala Lumpur, Malaysia
Duration: 2007 Aug 262007 Aug 29

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
NumberPART 1
Volume4705 LNCS
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

OtherInternational Conference on Computational Science and its Applications, ICCSA 2007
CountryMalaysia
CityKuala Lumpur
Period07/8/2607/8/29

Fingerprint

Color Image Segmentation
Normal Distribution
Normal distribution
Thresholding
Image segmentation
Merging
Gaussian distribution
Color
Speckle
Textures
Sampling
Human Perception
Experiments
Central limit theorem
Standard deviation
Texture
Segmentation
Vary
Experiment
Object

Keywords

  • Central limit theorem
  • Dividing
  • Merging
  • Normal distribution
  • Segmentation
  • Standard deviation
  • Threshold

ASJC Scopus subject areas

  • Biochemistry, Genetics and Molecular Biology(all)
  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Kang, S. D., Yoo, H. W., & Jang, D. S. (2007). Color image segmentation based on the normal distribution and the dynamic thresholding. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (PART 1 ed., Vol. 4705 LNCS, pp. 372-384). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 4705 LNCS, No. PART 1).

Color image segmentation based on the normal distribution and the dynamic thresholding. / Kang, Seon D.; Yoo, Hun W.; Jang, Dong Sik.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 4705 LNCS PART 1. ed. 2007. p. 372-384 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 4705 LNCS, No. PART 1).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Kang, SD, Yoo, HW & Jang, DS 2007, Color image segmentation based on the normal distribution and the dynamic thresholding. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). PART 1 edn, vol. 4705 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), no. PART 1, vol. 4705 LNCS, pp. 372-384, International Conference on Computational Science and its Applications, ICCSA 2007, Kuala Lumpur, Malaysia, 07/8/26.
Kang SD, Yoo HW, Jang DS. Color image segmentation based on the normal distribution and the dynamic thresholding. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). PART 1 ed. Vol. 4705 LNCS. 2007. p. 372-384. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 1).
Kang, Seon D. ; Yoo, Hun W. ; Jang, Dong Sik. / Color image segmentation based on the normal distribution and the dynamic thresholding. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 4705 LNCS PART 1. ed. 2007. pp. 372-384 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 1).
@inproceedings{d5bd17f217bd457dba043493d2aa2131,
title = "Color image segmentation based on the normal distribution and the dynamic thresholding",
abstract = "A new color image segmentation method is proposed in this paper. The proposed method is based on the human perception that in general human has attention on 3 or 4 major color objects in the image at first. Therefore, to determine the objects, three intensity distributions are constructed by sampling them randomly and sufficiently from three R, G, and B channel images. And three means are computed from three intensity distributions. Next, these steps are repeated many times to obtain three mean distribution sets. Each of these distributions comes to show normal shape based on the central limit theorem. To segment objects, each of the normal distribution is divided into 4 sections according to the standard deviation (section1 below - σ , section 2 between - σ and μ , section 3 between μ and σ , and section 4 over σ). Then sections with similar representative values are merged based on the threshold. This threshold is not chosen as constant but varies based on the difference of representative values of each section to reflect various characteristics for various images. Above merging process is iterated to reduce fine textures such as speckles remained even after the merging. Finally, segmented results of each channel images are combined to obtain a final segmentation result. The performance of the proposed method is evaluated through experiments over some images.",
keywords = "Central limit theorem, Dividing, Merging, Normal distribution, Segmentation, Standard deviation, Threshold",
author = "Kang, {Seon D.} and Yoo, {Hun W.} and Jang, {Dong Sik}",
year = "2007",
month = "12",
day = "1",
language = "English",
isbn = "9783540744689",
volume = "4705 LNCS",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
number = "PART 1",
pages = "372--384",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
edition = "PART 1",

}

TY - GEN

T1 - Color image segmentation based on the normal distribution and the dynamic thresholding

AU - Kang, Seon D.

AU - Yoo, Hun W.

AU - Jang, Dong Sik

PY - 2007/12/1

Y1 - 2007/12/1

N2 - A new color image segmentation method is proposed in this paper. The proposed method is based on the human perception that in general human has attention on 3 or 4 major color objects in the image at first. Therefore, to determine the objects, three intensity distributions are constructed by sampling them randomly and sufficiently from three R, G, and B channel images. And three means are computed from three intensity distributions. Next, these steps are repeated many times to obtain three mean distribution sets. Each of these distributions comes to show normal shape based on the central limit theorem. To segment objects, each of the normal distribution is divided into 4 sections according to the standard deviation (section1 below - σ , section 2 between - σ and μ , section 3 between μ and σ , and section 4 over σ). Then sections with similar representative values are merged based on the threshold. This threshold is not chosen as constant but varies based on the difference of representative values of each section to reflect various characteristics for various images. Above merging process is iterated to reduce fine textures such as speckles remained even after the merging. Finally, segmented results of each channel images are combined to obtain a final segmentation result. The performance of the proposed method is evaluated through experiments over some images.

AB - A new color image segmentation method is proposed in this paper. The proposed method is based on the human perception that in general human has attention on 3 or 4 major color objects in the image at first. Therefore, to determine the objects, three intensity distributions are constructed by sampling them randomly and sufficiently from three R, G, and B channel images. And three means are computed from three intensity distributions. Next, these steps are repeated many times to obtain three mean distribution sets. Each of these distributions comes to show normal shape based on the central limit theorem. To segment objects, each of the normal distribution is divided into 4 sections according to the standard deviation (section1 below - σ , section 2 between - σ and μ , section 3 between μ and σ , and section 4 over σ). Then sections with similar representative values are merged based on the threshold. This threshold is not chosen as constant but varies based on the difference of representative values of each section to reflect various characteristics for various images. Above merging process is iterated to reduce fine textures such as speckles remained even after the merging. Finally, segmented results of each channel images are combined to obtain a final segmentation result. The performance of the proposed method is evaluated through experiments over some images.

KW - Central limit theorem

KW - Dividing

KW - Merging

KW - Normal distribution

KW - Segmentation

KW - Standard deviation

KW - Threshold

UR - http://www.scopus.com/inward/record.url?scp=38049027308&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=38049027308&partnerID=8YFLogxK

M3 - Conference contribution

SN - 9783540744689

VL - 4705 LNCS

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 372

EP - 384

BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

ER -