Abstract

While near infrared and visible fusion recognition has been actively researched in recent years, most theoretical results and algorithms concentrate on the sufficient training samples setting. This paper focuses on the general fusion method when there are insufficient training samples with one pair of near-infrared and visible face images. Compared with existing methods, the proposed method requires neither sufficient samples nor the training step. To get a robust and time-efficient fusion model for unconstrained face recognition in the single sample situation, two models are proposed to fuse the local binary pattern based descriptors and the sparse representation based classification: the first fusion model directly fuses the representation error, while the second fusion model is an accelerated version that learns from a cross-spectral dictionary. Experiments are performed on the HITSZ LAB2 database, and the experiment results showed that the proposed fusion model extracted the complementary features of near-infrared and visible-light images. The fusion face recognition method had superior performance to state of the art fusion methods.

© 2019 Optical Society of America

PDF Article

References

You do not have subscription access to this journal. Citation lists with outbound citation links are available to subscribers only. You may subscribe either as an OSA member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access OSA Member Subscription

Cited By

You do not have subscription access to this journal. Cited by links are available to subscribers only. You may subscribe either as an OSA member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access OSA Member Subscription