The influence of ambient temperature is a big challenge to robust infrared face recognition. This paper proposes a new ambient temperature normalization algorithm to improve the performance of infrared face recognition under variable ambient temperatures. Based on statistical regression theory, a second order polynomial model is learned to describe the ambient temperature’s impact on infrared face image. Then, infrared image was normalized to reference ambient temperature by the second order polynomial model. Finally, this normalization method is applied to infrared face recognition to verify its efficiency. The experiments demonstrate that the proposed temperature normalization method is feasible and can significantly improve the robustness of infrared face recognition.

Date Published: 22 August 2015
PDF: 6 pages
Proc. SPIE 9656, International Symposium on Photonics and Optoelectronics 2015, 96560O (22 August 2015); doi: 10.1117/12.2197248

Published in SPIE Proceedings Vol. 9656:
International Symposium on Photonics and Optoelectronics 2015
Zhiping Zhou, Editor(s)