Multi subspaces active appearance models

Yang Junyeong, Byun Hyeran

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

Abstract

The original Active Appearance Model(AAM) uses the mean matrix of gradient matrixes instead of a gradient matrix which should be recomputed with respect to a varying parameter at a fitting phase. By this property, the original AAM can guarantee a fast fitting speed because it avoids computation of a gradient matrix of which a computation complexity is high. However, the fixed gradient matrix is not good choice when the distribution of a training database is nonlinear because the mean can not represent the variation of a training database. To overcome this problem, this paper proposes multi subspaces AAM. First, we divide a training database into multi subspaces along the illumination direction, and build the independent AAM for each subspace. At a fitting phase, we adaptively choose a subspace well fit to a target image. However, the parameter update problem is occurred because a subspace can be changed during a fitting phase. To solve this problem, we propose a linear transform matrix on an eigenspace. In experiments, we apply the proposed method to Yale Face Database B and demonstrate that the method is robust for facial images under various illuminations.

Original languageEnglish
Title of host publication2008 8th IEEE International Conference on Automatic Face and Gesture Recognition, FG 2008
DOIs
Publication statusPublished - 2008 Dec 1
Event2008 8th IEEE International Conference on Automatic Face and Gesture Recognition, FG 2008 - Amsterdam, Netherlands
Duration: 2008 Sep 172008 Sep 19

Other

Other2008 8th IEEE International Conference on Automatic Face and Gesture Recognition, FG 2008
CountryNetherlands
CityAmsterdam
Period08/9/1708/9/19

Fingerprint

Lighting
Experiments

All Science Journal Classification (ASJC) codes

  • Computer Vision and Pattern Recognition
  • Human-Computer Interaction
  • Electrical and Electronic Engineering

Cite this

Junyeong, Y., & Hyeran, B. (2008). Multi subspaces active appearance models. In 2008 8th IEEE International Conference on Automatic Face and Gesture Recognition, FG 2008 [4813334] https://doi.org/10.1109/AFGR.2008.4813334
Junyeong, Yang ; Hyeran, Byun. / Multi subspaces active appearance models. 2008 8th IEEE International Conference on Automatic Face and Gesture Recognition, FG 2008. 2008.
@inproceedings{4c12ec49ca1b48f48a761f3920b52ca9,
title = "Multi subspaces active appearance models",
abstract = "The original Active Appearance Model(AAM) uses the mean matrix of gradient matrixes instead of a gradient matrix which should be recomputed with respect to a varying parameter at a fitting phase. By this property, the original AAM can guarantee a fast fitting speed because it avoids computation of a gradient matrix of which a computation complexity is high. However, the fixed gradient matrix is not good choice when the distribution of a training database is nonlinear because the mean can not represent the variation of a training database. To overcome this problem, this paper proposes multi subspaces AAM. First, we divide a training database into multi subspaces along the illumination direction, and build the independent AAM for each subspace. At a fitting phase, we adaptively choose a subspace well fit to a target image. However, the parameter update problem is occurred because a subspace can be changed during a fitting phase. To solve this problem, we propose a linear transform matrix on an eigenspace. In experiments, we apply the proposed method to Yale Face Database B and demonstrate that the method is robust for facial images under various illuminations.",
author = "Yang Junyeong and Byun Hyeran",
year = "2008",
month = "12",
day = "1",
doi = "10.1109/AFGR.2008.4813334",
language = "English",
isbn = "9781424421541",
booktitle = "2008 8th IEEE International Conference on Automatic Face and Gesture Recognition, FG 2008",

}

Junyeong, Y & Hyeran, B 2008, Multi subspaces active appearance models. in 2008 8th IEEE International Conference on Automatic Face and Gesture Recognition, FG 2008., 4813334, 2008 8th IEEE International Conference on Automatic Face and Gesture Recognition, FG 2008, Amsterdam, Netherlands, 08/9/17. https://doi.org/10.1109/AFGR.2008.4813334

Multi subspaces active appearance models. / Junyeong, Yang; Hyeran, Byun.

2008 8th IEEE International Conference on Automatic Face and Gesture Recognition, FG 2008. 2008. 4813334.

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

TY - GEN

T1 - Multi subspaces active appearance models

AU - Junyeong, Yang

AU - Hyeran, Byun

PY - 2008/12/1

Y1 - 2008/12/1

N2 - The original Active Appearance Model(AAM) uses the mean matrix of gradient matrixes instead of a gradient matrix which should be recomputed with respect to a varying parameter at a fitting phase. By this property, the original AAM can guarantee a fast fitting speed because it avoids computation of a gradient matrix of which a computation complexity is high. However, the fixed gradient matrix is not good choice when the distribution of a training database is nonlinear because the mean can not represent the variation of a training database. To overcome this problem, this paper proposes multi subspaces AAM. First, we divide a training database into multi subspaces along the illumination direction, and build the independent AAM for each subspace. At a fitting phase, we adaptively choose a subspace well fit to a target image. However, the parameter update problem is occurred because a subspace can be changed during a fitting phase. To solve this problem, we propose a linear transform matrix on an eigenspace. In experiments, we apply the proposed method to Yale Face Database B and demonstrate that the method is robust for facial images under various illuminations.

AB - The original Active Appearance Model(AAM) uses the mean matrix of gradient matrixes instead of a gradient matrix which should be recomputed with respect to a varying parameter at a fitting phase. By this property, the original AAM can guarantee a fast fitting speed because it avoids computation of a gradient matrix of which a computation complexity is high. However, the fixed gradient matrix is not good choice when the distribution of a training database is nonlinear because the mean can not represent the variation of a training database. To overcome this problem, this paper proposes multi subspaces AAM. First, we divide a training database into multi subspaces along the illumination direction, and build the independent AAM for each subspace. At a fitting phase, we adaptively choose a subspace well fit to a target image. However, the parameter update problem is occurred because a subspace can be changed during a fitting phase. To solve this problem, we propose a linear transform matrix on an eigenspace. In experiments, we apply the proposed method to Yale Face Database B and demonstrate that the method is robust for facial images under various illuminations.

UR - http://www.scopus.com/inward/record.url?scp=67650652518&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=67650652518&partnerID=8YFLogxK

U2 - 10.1109/AFGR.2008.4813334

DO - 10.1109/AFGR.2008.4813334

M3 - Conference contribution

AN - SCOPUS:67650652518

SN - 9781424421541

BT - 2008 8th IEEE International Conference on Automatic Face and Gesture Recognition, FG 2008

ER -

Junyeong Y, Hyeran B. Multi subspaces active appearance models. In 2008 8th IEEE International Conference on Automatic Face and Gesture Recognition, FG 2008. 2008. 4813334 https://doi.org/10.1109/AFGR.2008.4813334