Title: Exploit more information of the sample for representation based face recognition

Authors: Kezong Tang; Xuan Xiao; Zuoyong Li

Addresses: Information Engineering Institute, Jingdezhen Ceramic Institute, Jingdezhen 333000, Jiangxi, China ' Information Engineering Institute, Jingdezhen Ceramic Institute, Jingdezhen 333000, Jiangxi, China ' Department of Computer Science, Minjiang University, Fuzhou 350108, Fujian, China

Abstract: Besides the representation based classification method defined in the original space for face recognition, the method has also been extended to the feature space. As defined in the original space the method uses a linear combination of all the training samples to approximate the test sample in the original space and then exploits the determined linear combination of all the training samples of each class to classify the test sample. As defined in the feature space the method first implicitly transforms the sample into a high-dimensional space, then classify the test sample in the feature space. In this paper, we propose to integrate the original data of the sample and its representation in the feature space for face recognition based on their respective rationales. The proposed method outperforms the representation based classification method defined in the original space.

Keywords: face recognition; classification methods; feature space; biometrics.

DOI: 10.1504/IJWMC.2015.070938

International Journal of Wireless and Mobile Computing, 2015 Vol.8 No.4, pp.401 - 405

Received: 12 Jul 2014
Accepted: 22 Sep 2014
Published online: 03 Aug 2015 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article