Publication detail

Hair Segmentation for Color Estimation in Surveillance Systems

KŘUPKA, A. PŘINOSIL, J. ŘÍHA, K. MINÁŘ, J. DUTTA, M.

Original Title

Hair Segmentation for Color Estimation in Surveillance Systems

English Title

Hair Segmentation for Color Estimation in Surveillance Systems

Type

conference paper

Language

en

Original Abstract

The paper proposes a novel method for hair segmentation, that can be used in real-time video surveillance systems or multimedia services. The method utilizes an approach of video subtraction to obtain a person silhouette. Subsequently, the head part can be identified on the silhouette by exploiting information about face position. From the head, the skin is separated using floodfilling procedure and the hair area is determined as the difference between the head and the skin. The precision of the method is evaluated using manually extracted hair masks. The purpose of the segmentation method is to specify a hair area which can be then used for a hair color estimation. Therefore, the usability of the hair segmentation procedure is tested by a proposed scheme of hair color estimation.

English abstract

The paper proposes a novel method for hair segmentation, that can be used in real-time video surveillance systems or multimedia services. The method utilizes an approach of video subtraction to obtain a person silhouette. Subsequently, the head part can be identified on the silhouette by exploiting information about face position. From the head, the skin is separated using floodfilling procedure and the hair area is determined as the difference between the head and the skin. The precision of the method is evaluated using manually extracted hair masks. The purpose of the segmentation method is to specify a hair area which can be then used for a hair color estimation. Therefore, the usability of the hair segmentation procedure is tested by a proposed scheme of hair color estimation.

Keywords

segmentation, hair, color, face detection, video subtraction, floodfilling

RIV year

2014

Released

22.02.2014

ISBN

978-1-61208-320-9

Book

MMEDIA 2014, The Sixth International Conferences on Advances in Multimedia

Pages from

102

Pages to

107

Pages count

6

URL

BibTex


@inproceedings{BUT106756,
  author="Aleš {Křupka} and Jiří {Přinosil} and Kamil {Říha} and Jiří {Minář} and Malay Kishore {Dutta}",
  title="Hair Segmentation for Color Estimation in Surveillance Systems",
  annote="The paper proposes a novel method for hair segmentation, that can be used in real-time video surveillance systems or multimedia services. The method utilizes an approach of video subtraction to obtain a person silhouette. Subsequently, the head part can be identified on the silhouette by exploiting information about face position. From the head, the skin is separated using floodfilling procedure and the hair area is determined as the difference between the head and the skin. The precision of the method is evaluated using manually extracted hair masks. The purpose of the segmentation method is to specify a hair area which can be then used for a hair color estimation. Therefore, the usability of the hair segmentation procedure is tested by a proposed scheme of hair color estimation.",
  booktitle="MMEDIA 2014, The Sixth International Conferences on Advances in Multimedia",
  chapter="106756",
  howpublished="online",
  year="2014",
  month="february",
  pages="102--107",
  type="conference paper"
}