Parallel Heat Kernel Volume Based Local Binary Pattern on Multi-Orientation Planes for Face Representation

Appropriate representation is one of the keys to successful face recognition technologies. Actual facial appearance sometimes differs dramatically because of variations in pose, illumination, expression, and occlusion. However, existing face representation methods remain insufficiently powerful and...

Full description

Saved in:
Bibliographic Details
Published in:International journal of parallel programming Vol. 46; no. 5; pp. 943 - 962
Main Authors: Lu, Wei, Yang, Xiaomin, Gou, Xu, Jian, Lihua, Wu, Wei, Jeon, Gwanggil
Format: Journal Article
Language:English
Published: New York Springer US 01.10.2018
Springer Nature B.V
Subjects:
Algorithms
Computer programming
Computer Science
Face
Face recognition
Feature extraction
Kernels
Macrostructure
Occlusion
Orientation
Parallel processing
Pattern analysis
Pattern recognition
Planes
Processor Architectures
Representations
Software Engineering/Programming and Operating Systems
Special Issue on Parallel Approaches for Data Mining in the Internet of Things Realm
Theory of Computation
Face presentation
Face recognition
Parallel multi-scale heat kernel face
Local binary patterns
ISSN:0885-7458, 1573-7640
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Appropriate representation is one of the keys to successful face recognition technologies. Actual facial appearance sometimes differs dramatically because of variations in pose, illumination, expression, and occlusion. However, existing face representation methods remain insufficiently powerful and robust. Hence, we propose a new feature extraction approach for face representation based on heat kernel volume and local binary patterns. Multi-scale heat kernel faces are created in our proposed framework. We then reformulate these multi-scale heat kernel faces as three-dimensional volume. Furthermore, we generate multi-orientation planes from the heat kernel volume, which reflects orientation co-occurrence statistics among different heat kernel faces. Finally, we apply local binary pattern (LBP) analysis on the multi-orientation planes of the heat kernel volume to capture the microstructure and macrostructure of face appearance. Hence, we obtain the heat kernel volume based local binary pattern on multi-orientation planes (HKV–LBP–MOP) descriptor. The proposed method is successfully be paralleled. We applied the method to face recognition and obtain the performance of 99.28 and 87.82% on ORL and Yale databases respectively. Experimental results on the show that the proposed algorithm significantly outperforms other well-known approaches in terms of recognition rate.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0885-7458
1573-7640
DOI:10.1007/s10766-017-0552-8