Robust frontal view search using extended manifold learning

Chao Wang, Xubo Song

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

Many 2D face processing algorithms can perform better using frontal or near frontal faces. In this paper, we present a robust frontal view search method based on manifold learning, with the assumption that with the pose being the only variable, face images should lie in a smooth and low-dimensional manifold. In 2D embedding, we find that manifold geometry of face images with varying poses has the shape of a parabola with the frontal view in the vertex. However, background clutter and illumination variations make frontal view deviate from the vertex. To address this problem, we propose a pairwise K-nearest neighbor protocol to extend manifold learning. In addition, we present an illumination-robust localized edge orientation histogram to represent face image in the extended manifold learning. The experimental results show that the extended algorithms have higher search accuracy, even under varying illuminations.

Original languageEnglish (US)
Pages (from-to)1147-1154
Number of pages8
JournalJournal of Visual Communication and Image Representation
Volume24
Issue number7
DOIs
StatePublished - 2013

Fingerprint

Lighting
Geometry
Processing

Keywords

  • Frontal face
  • Frontal view
  • Frontal view search
  • K-nearest neighbor protocol
  • Laplacian eigenmaps
  • Localized edge orientation histogram
  • Locally linear embedding
  • Manifold learning
  • Pairwise
  • Pose estimation

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Media Technology
  • Computer Vision and Pattern Recognition
  • Signal Processing

Cite this

Robust frontal view search using extended manifold learning. / Wang, Chao; Song, Xubo.

In: Journal of Visual Communication and Image Representation, Vol. 24, No. 7, 2013, p. 1147-1154.

Research output: Contribution to journalArticle

@article{973279eb336b4454bb18c179ee225b42,
title = "Robust frontal view search using extended manifold learning",
abstract = "Many 2D face processing algorithms can perform better using frontal or near frontal faces. In this paper, we present a robust frontal view search method based on manifold learning, with the assumption that with the pose being the only variable, face images should lie in a smooth and low-dimensional manifold. In 2D embedding, we find that manifold geometry of face images with varying poses has the shape of a parabola with the frontal view in the vertex. However, background clutter and illumination variations make frontal view deviate from the vertex. To address this problem, we propose a pairwise K-nearest neighbor protocol to extend manifold learning. In addition, we present an illumination-robust localized edge orientation histogram to represent face image in the extended manifold learning. The experimental results show that the extended algorithms have higher search accuracy, even under varying illuminations.",
keywords = "Frontal face, Frontal view, Frontal view search, K-nearest neighbor protocol, Laplacian eigenmaps, Localized edge orientation histogram, Locally linear embedding, Manifold learning, Pairwise, Pose estimation",
author = "Chao Wang and Xubo Song",
year = "2013",
doi = "10.1016/j.jvcir.2013.06.013",
language = "English (US)",
volume = "24",
pages = "1147--1154",
journal = "Journal of Visual Communication and Image Representation",
issn = "1047-3203",
publisher = "Academic Press Inc.",
number = "7",

}

TY - JOUR

T1 - Robust frontal view search using extended manifold learning

AU - Wang, Chao

AU - Song, Xubo

PY - 2013

Y1 - 2013

N2 - Many 2D face processing algorithms can perform better using frontal or near frontal faces. In this paper, we present a robust frontal view search method based on manifold learning, with the assumption that with the pose being the only variable, face images should lie in a smooth and low-dimensional manifold. In 2D embedding, we find that manifold geometry of face images with varying poses has the shape of a parabola with the frontal view in the vertex. However, background clutter and illumination variations make frontal view deviate from the vertex. To address this problem, we propose a pairwise K-nearest neighbor protocol to extend manifold learning. In addition, we present an illumination-robust localized edge orientation histogram to represent face image in the extended manifold learning. The experimental results show that the extended algorithms have higher search accuracy, even under varying illuminations.

AB - Many 2D face processing algorithms can perform better using frontal or near frontal faces. In this paper, we present a robust frontal view search method based on manifold learning, with the assumption that with the pose being the only variable, face images should lie in a smooth and low-dimensional manifold. In 2D embedding, we find that manifold geometry of face images with varying poses has the shape of a parabola with the frontal view in the vertex. However, background clutter and illumination variations make frontal view deviate from the vertex. To address this problem, we propose a pairwise K-nearest neighbor protocol to extend manifold learning. In addition, we present an illumination-robust localized edge orientation histogram to represent face image in the extended manifold learning. The experimental results show that the extended algorithms have higher search accuracy, even under varying illuminations.

KW - Frontal face

KW - Frontal view

KW - Frontal view search

KW - K-nearest neighbor protocol

KW - Laplacian eigenmaps

KW - Localized edge orientation histogram

KW - Locally linear embedding

KW - Manifold learning

KW - Pairwise

KW - Pose estimation

UR - http://www.scopus.com/inward/record.url?scp=84882786423&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84882786423&partnerID=8YFLogxK

U2 - 10.1016/j.jvcir.2013.06.013

DO - 10.1016/j.jvcir.2013.06.013

M3 - Article

VL - 24

SP - 1147

EP - 1154

JO - Journal of Visual Communication and Image Representation

JF - Journal of Visual Communication and Image Representation

SN - 1047-3203

IS - 7

ER -