Quadratic projection based feature extraction with its application to biometric recognition

PATTERN RECOGNITION

卷:56

Article

Yan, Yan¹  Wang, Hanzi¹  Chen, Si²  Cao, Xiaochun³  Zhang, David⁴ 

[1] Xiamen Univ, Sch Informat Sci & Engn, Fujian Key Lab Sensing & Comp Smart City, Xiamen 361005, Fujian, Peoples R China

[2] Xiamen Univ Technol, Sch Comp & Informat Engn, Xiamen 361024, Fujian, Peoples R China

[3] Chinese Acad Sci, Inst Informat Engn, State Key Lab Informat Secur, Beijing 100093, Peoples R China

[4] Hong Kong Polytech Univ, Biometr Res Ctr, Hong Kong, Hong Kong, Peoples R China

关键词: Biometric recognition;    Feature extraction;    Quadratic projection;    Semidefinite programming;    Lagrange duality;   

DOI  :  10.1016/j.patcog.2016.02.010

来源: Elsevier

【 摘 要 】

This paper presents a novel quadratic projection based feature extraction framework, where a set of quadratic matrices is learned to distinguish each class from all other classes. We formulate quadratic matrix learning (QML) as a standard semidefinite programming (SDP) problem. However, the conventional interior-point SDP solvers do not scale well to the problem of QML for high-dimensional data. To solve the scalability of QML, we develop an efficient algorithm, termed DualQML, based on the Lagrange duality theory, to extract nonlinear features. To evaluate the feasibility and effectiveness of the proposed framework, we conduct extensive experiments on biometric recognition. Experimental results on three representative biometric recognition tasks, including face, palmprint, and ear recognition, demonstrate the superiority of the DualQML-based feature extraction algorithm compared to the current state-of-the-art algorithms. (C) 2016 Elsevier Ltd. All rights reserved.

【 授权许可】

Free