Publication detail

NEURAL NETWORK BASED SEGMENTATION OF COLOR IMAGES FOR SKIN DETECTION

HYNČICA, T. JIRSÍK, V.

Original Title

NEURAL NETWORK BASED SEGMENTATION OF COLOR IMAGES FOR SKIN DETECTION

Czech Title

Segmentace barevných snímků pro detekci kůže za použití neuronových sítí

English Title

NEURAL NETWORK BASED SEGMENTATION OF COLOR IMAGES FOR SKIN DETECTION

Type

conference paper

Language

en

Original Abstract

Skin segmentation is an important task in image processing and computer vision. Since there are many problems (for example different lighting conditions, different skin colors across the population and so on) it is necessary to use a robust and effective detection method. At first the proper color representation (color space) must be chosen. The description of four color spaces frequently used in image processing (RGB, HSL, YCbCr and L*a*b*) is provided in this paper. Also their suitability for representing skin colors is evaluated. For skin pixel detection a RCE (reduced Coulomb energy) neural network is used.

Czech abstract

Segmentace kůže je důležitým úkolem při zpracování obrazu. Vzhledem k množství problémů (například různé světelné podmínky, různé barvy pokožky v celé populaci, a tak dále), je nutné použít robustní a efektivní metody detekce. Pro detekci pixelů kůže sje použita RCE neuronová síť.

English abstract

Skin segmentation is an important task in image processing and computer vision. Since there are many problems (for example different lighting conditions, different skin colors across the population and so on) it is necessary to use a robust and effective detection method. At first the proper color representation (color space) must be chosen. The description of four color spaces frequently used in image processing (RGB, HSL, YCbCr and L*a*b*) is provided in this paper. Also their suitability for representing skin colors is evaluated. For skin pixel detection a RCE (reduced Coulomb energy) neural network is used.

Keywords

RCE neural network, color spaces, skin segmentation

RIV year

2012

Released

27.06.2012

Publisher

Nosová Hana, Koperníkova 4362/21, Brno,

Location

Brno

ISBN

978-80-214-4540-6

Book

MENDEL 2012 18th Intermational Conference on Soft Computing

Edition

1

Edition number

1

Pages from

251

Pages to

255

Pages count

5

BibTex


@inproceedings{BUT93438,
  author="Tomáš {Hynčica} and Václav {Jirsík}",
  title="NEURAL NETWORK BASED SEGMENTATION OF COLOR IMAGES FOR SKIN DETECTION",
  annote="Skin segmentation is an important task in image processing and computer vision. Since there are many problems (for example different lighting conditions, different skin colors across the population and so on) it is necessary to use a robust and effective detection method. At first the proper color representation (color space) must be chosen. The description of four color spaces frequently used in image processing (RGB, HSL, YCbCr and L*a*b*) is provided in this paper. Also their suitability for representing skin colors is evaluated.  For skin pixel detection a RCE (reduced Coulomb energy) neural network is used.",
  address="Nosová Hana, Koperníkova 4362/21, Brno,",
  booktitle="MENDEL 2012 18th Intermational Conference on Soft Computing",
  chapter="93438",
  edition="1",
  howpublished="print",
  institution="Nosová Hana, Koperníkova 4362/21, Brno,",
  year="2012",
  month="june",
  pages="251--255",
  publisher="Nosová Hana, Koperníkova 4362/21, Brno,",
  type="conference paper"
}