Accurate foreground extraction using graph cut with trimap estimation

Jung Ho Ahn, Hyeran Byun

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

4 Citations (Scopus)

Abstract

This paper describes an accurate human silhouette extraction method as applied to video sequences. In computer vision applications that use a static camera, the background subtraction method is one of the most effective ways of extracting human silhouettes. However it is prone to errors so performance of silhouette-based gait and gesture recognition often decreases significantly. In this paper we propose two-step segmentation method: trimap estimation and fine segmentation using a graph cut. We first estimated foreground, background and unknown regions with an acceptable level of confidence. Then, the energy function was identified by focussing on the unknown region, and it was minimized via the graph cut method to achieve optimal segmentation. The proposed algorithm was evaluated with respect to ground truth data and it was shown to produce high quality human silhouettes.

Original languageEnglish
Title of host publicationAdvances in Image and Video Technology - First Pacific Rim Symposium, PSIVT 2006, Proceedings
Pages1185-1194
Number of pages10
DOIs
Publication statusPublished - 2006 Dec 1
Event1st Pacific Rim Symposium on Image and Video Technology, PSIVT 2006 - Hsinchu, Taiwan, Province of China
Duration: 2006 Dec 102006 Dec 13

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume4319 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other1st Pacific Rim Symposium on Image and Video Technology, PSIVT 2006
CountryTaiwan, Province of China
CityHsinchu
Period06/12/1006/12/13

Fingerprint

Gesture recognition
Graph Cuts
Silhouette
Computer vision
Cameras
Segmentation
Gait Recognition
Unknown
Gesture Recognition
Background Subtraction
Energy Function
Computer Vision
Confidence
Camera
Decrease
Human

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Ahn, J. H., & Byun, H. (2006). Accurate foreground extraction using graph cut with trimap estimation. In Advances in Image and Video Technology - First Pacific Rim Symposium, PSIVT 2006, Proceedings (pp. 1185-1194). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 4319 LNCS). https://doi.org/10.1007/11949534-120
Ahn, Jung Ho ; Byun, Hyeran. / Accurate foreground extraction using graph cut with trimap estimation. Advances in Image and Video Technology - First Pacific Rim Symposium, PSIVT 2006, Proceedings. 2006. pp. 1185-1194 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{d9d50b493fb04103a2df99a92c09bdaa,
title = "Accurate foreground extraction using graph cut with trimap estimation",
abstract = "This paper describes an accurate human silhouette extraction method as applied to video sequences. In computer vision applications that use a static camera, the background subtraction method is one of the most effective ways of extracting human silhouettes. However it is prone to errors so performance of silhouette-based gait and gesture recognition often decreases significantly. In this paper we propose two-step segmentation method: trimap estimation and fine segmentation using a graph cut. We first estimated foreground, background and unknown regions with an acceptable level of confidence. Then, the energy function was identified by focussing on the unknown region, and it was minimized via the graph cut method to achieve optimal segmentation. The proposed algorithm was evaluated with respect to ground truth data and it was shown to produce high quality human silhouettes.",
author = "Ahn, {Jung Ho} and Hyeran Byun",
year = "2006",
month = "12",
day = "1",
doi = "10.1007/11949534-120",
language = "English",
isbn = "354068297X",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "1185--1194",
booktitle = "Advances in Image and Video Technology - First Pacific Rim Symposium, PSIVT 2006, Proceedings",

}

Ahn, JH & Byun, H 2006, Accurate foreground extraction using graph cut with trimap estimation. in Advances in Image and Video Technology - First Pacific Rim Symposium, PSIVT 2006, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 4319 LNCS, pp. 1185-1194, 1st Pacific Rim Symposium on Image and Video Technology, PSIVT 2006, Hsinchu, Taiwan, Province of China, 06/12/10. https://doi.org/10.1007/11949534-120

Accurate foreground extraction using graph cut with trimap estimation. / Ahn, Jung Ho; Byun, Hyeran.

Advances in Image and Video Technology - First Pacific Rim Symposium, PSIVT 2006, Proceedings. 2006. p. 1185-1194 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 4319 LNCS).

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

TY - GEN

T1 - Accurate foreground extraction using graph cut with trimap estimation

AU - Ahn, Jung Ho

AU - Byun, Hyeran

PY - 2006/12/1

Y1 - 2006/12/1

N2 - This paper describes an accurate human silhouette extraction method as applied to video sequences. In computer vision applications that use a static camera, the background subtraction method is one of the most effective ways of extracting human silhouettes. However it is prone to errors so performance of silhouette-based gait and gesture recognition often decreases significantly. In this paper we propose two-step segmentation method: trimap estimation and fine segmentation using a graph cut. We first estimated foreground, background and unknown regions with an acceptable level of confidence. Then, the energy function was identified by focussing on the unknown region, and it was minimized via the graph cut method to achieve optimal segmentation. The proposed algorithm was evaluated with respect to ground truth data and it was shown to produce high quality human silhouettes.

AB - This paper describes an accurate human silhouette extraction method as applied to video sequences. In computer vision applications that use a static camera, the background subtraction method is one of the most effective ways of extracting human silhouettes. However it is prone to errors so performance of silhouette-based gait and gesture recognition often decreases significantly. In this paper we propose two-step segmentation method: trimap estimation and fine segmentation using a graph cut. We first estimated foreground, background and unknown regions with an acceptable level of confidence. Then, the energy function was identified by focussing on the unknown region, and it was minimized via the graph cut method to achieve optimal segmentation. The proposed algorithm was evaluated with respect to ground truth data and it was shown to produce high quality human silhouettes.

UR - http://www.scopus.com/inward/record.url?scp=60949094075&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=60949094075&partnerID=8YFLogxK

U2 - 10.1007/11949534-120

DO - 10.1007/11949534-120

M3 - Conference contribution

SN - 354068297X

SN - 9783540682974

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

SP - 1185

EP - 1194

BT - Advances in Image and Video Technology - First Pacific Rim Symposium, PSIVT 2006, Proceedings

ER -

Ahn JH, Byun H. Accurate foreground extraction using graph cut with trimap estimation. In Advances in Image and Video Technology - First Pacific Rim Symposium, PSIVT 2006, Proceedings. 2006. p. 1185-1194. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/11949534-120