A Multiscale Method for HOG-Based Face Recognition

X Wei, H. Wang, Gongde Guo, Huan Wan

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

7 Citations (Scopus)

Abstract

Image representation is an important process in image classification, and there are many different methods for representing images. HOG (Histograms of Oriented Gradients) is a popular one which has been used in many applications including face recognition, pedestrian detection and palmprint recognition. In this paper, a novel method is presented to improve HOG-based image classification by using the multiscale features of images. For each image, multiple HOG feature vectors are extracted under different spatial dimensions (or ’scales’). These ’multiscale’ feature vectors are then fused into a distance function to calculate the distance between two images. Experiments have been conducted on ORL face database, AR face database and FERET face database. Results show the use of multiscale HOG features has led to significant improvement in performance over the use of single scale HOG features. Results also show that the nearest neighbour classifier equipped with our distance function is comparable to the well-known and widely-used benchmark classifier.
LanguageEnglish
Title of host publicationInternational Conference on Intelligent Robotics and Applications
Subtitle of host publicationICIRA 2015
Pages535-545
Number of pages11
Volume9244
ISBN (Electronic)978-3-319-22879-2
Publication statusPublished - 20 Aug 2015
EventInternational Conference on Intelligent Robotics and Applications (ICIRA) -
Duration: 1 Jan 2015 → …

Publication series

NameLecture Notes in Computer Science
PublisherSpringer Nature
Volume9244

Conference

ConferenceInternational Conference on Intelligent Robotics and Applications (ICIRA)
Period1/01/15 → …

Fingerprint

Face recognition
Image classification
Classifiers
Palmprint recognition
Experiments

Keywords

  • Multiscale image descriptor
  • face recognition
  • HOG

Cite this

Wei, X., Wang, H., Guo, G., & Wan, H. (2015). A Multiscale Method for HOG-Based Face Recognition. In International Conference on Intelligent Robotics and Applications: ICIRA 2015 (Vol. 9244, pp. 535-545). (Lecture Notes in Computer Science; Vol. 9244).
Wei, X ; Wang, H. ; Guo, Gongde ; Wan, Huan. / A Multiscale Method for HOG-Based Face Recognition. International Conference on Intelligent Robotics and Applications: ICIRA 2015. Vol. 9244 2015. pp. 535-545 (Lecture Notes in Computer Science).
@inproceedings{8a9d2f64650847699866a82332461701,
title = "A Multiscale Method for HOG-Based Face Recognition",
abstract = "Image representation is an important process in image classification, and there are many different methods for representing images. HOG (Histograms of Oriented Gradients) is a popular one which has been used in many applications including face recognition, pedestrian detection and palmprint recognition. In this paper, a novel method is presented to improve HOG-based image classification by using the multiscale features of images. For each image, multiple HOG feature vectors are extracted under different spatial dimensions (or ’scales’). These ’multiscale’ feature vectors are then fused into a distance function to calculate the distance between two images. Experiments have been conducted on ORL face database, AR face database and FERET face database. Results show the use of multiscale HOG features has led to significant improvement in performance over the use of single scale HOG features. Results also show that the nearest neighbour classifier equipped with our distance function is comparable to the well-known and widely-used benchmark classifier.",
keywords = "Multiscale image descriptor, face recognition, HOG",
author = "X Wei and H. Wang and Gongde Guo and Huan Wan",
year = "2015",
month = "8",
day = "20",
language = "English",
isbn = "978-3-319-22878-5",
volume = "9244",
series = "Lecture Notes in Computer Science",
publisher = "Springer Nature",
pages = "535--545",
booktitle = "International Conference on Intelligent Robotics and Applications",

}

Wei, X, Wang, H, Guo, G & Wan, H 2015, A Multiscale Method for HOG-Based Face Recognition. in International Conference on Intelligent Robotics and Applications: ICIRA 2015. vol. 9244, Lecture Notes in Computer Science, vol. 9244, pp. 535-545, International Conference on Intelligent Robotics and Applications (ICIRA), 1/01/15.

A Multiscale Method for HOG-Based Face Recognition. / Wei, X; Wang, H.; Guo, Gongde; Wan, Huan.

International Conference on Intelligent Robotics and Applications: ICIRA 2015. Vol. 9244 2015. p. 535-545 (Lecture Notes in Computer Science; Vol. 9244).

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

TY - GEN

T1 - A Multiscale Method for HOG-Based Face Recognition

AU - Wei, X

AU - Wang, H.

AU - Guo, Gongde

AU - Wan, Huan

PY - 2015/8/20

Y1 - 2015/8/20

N2 - Image representation is an important process in image classification, and there are many different methods for representing images. HOG (Histograms of Oriented Gradients) is a popular one which has been used in many applications including face recognition, pedestrian detection and palmprint recognition. In this paper, a novel method is presented to improve HOG-based image classification by using the multiscale features of images. For each image, multiple HOG feature vectors are extracted under different spatial dimensions (or ’scales’). These ’multiscale’ feature vectors are then fused into a distance function to calculate the distance between two images. Experiments have been conducted on ORL face database, AR face database and FERET face database. Results show the use of multiscale HOG features has led to significant improvement in performance over the use of single scale HOG features. Results also show that the nearest neighbour classifier equipped with our distance function is comparable to the well-known and widely-used benchmark classifier.

AB - Image representation is an important process in image classification, and there are many different methods for representing images. HOG (Histograms of Oriented Gradients) is a popular one which has been used in many applications including face recognition, pedestrian detection and palmprint recognition. In this paper, a novel method is presented to improve HOG-based image classification by using the multiscale features of images. For each image, multiple HOG feature vectors are extracted under different spatial dimensions (or ’scales’). These ’multiscale’ feature vectors are then fused into a distance function to calculate the distance between two images. Experiments have been conducted on ORL face database, AR face database and FERET face database. Results show the use of multiscale HOG features has led to significant improvement in performance over the use of single scale HOG features. Results also show that the nearest neighbour classifier equipped with our distance function is comparable to the well-known and widely-used benchmark classifier.

KW - Multiscale image descriptor

KW - face recognition

KW - HOG

M3 - Conference contribution

SN - 978-3-319-22878-5

VL - 9244

T3 - Lecture Notes in Computer Science

SP - 535

EP - 545

BT - International Conference on Intelligent Robotics and Applications

ER -

Wei X, Wang H, Guo G, Wan H. A Multiscale Method for HOG-Based Face Recognition. In International Conference on Intelligent Robotics and Applications: ICIRA 2015. Vol. 9244. 2015. p. 535-545. (Lecture Notes in Computer Science).