Detail publikace

Thermal Face Recognition - Matching Algorithms Performance

Originální název

Thermal Face Recognition - Matching Algorithms Performance

Anglický název

Thermal Face Recognition - Matching Algorithms Performance

Jazyk

en

Originální abstrakt

Face recognition based on thermal images belongs to less known biometric methods. The method works with images of human face captured in near infrared light spectrum. Infrared face recognition can be either use in stand-alone applications, or as a part of multimodal biometric system in combination with other methods (e.g. face recognition based on 2D or 3D images in visible light spectrum). This article presents different methods of normalization and matching algorithms and shows how they affect the performance of the biometric system. Results of all described methods are summarized to see, which combination of normalization and matching algorithms have the best performance.

Anglický abstrakt

Face recognition based on thermal images belongs to less known biometric methods. The method works with images of human face captured in near infrared light spectrum. Infrared face recognition can be either use in stand-alone applications, or as a part of multimodal biometric system in combination with other methods (e.g. face recognition based on 2D or 3D images in visible light spectrum). This article presents different methods of normalization and matching algorithms and shows how they affect the performance of the biometric system. Results of all described methods are summarized to see, which combination of normalization and matching algorithms have the best performance.

BibTex


@inproceedings{BUT76327,
  author="Jan {Váňa} and Martin {Drahanský} and Radim {Dvořák}",
  title="Thermal Face Recognition - Matching Algorithms Performance",
  annote="Face recognition based on thermal images belongs to less known biometric methods.
The method works with images of human face captured in near infrared light
spectrum. Infrared face recognition can be either use in stand-alone
applications, or as a part of multimodal biometric system in combination with
other methods (e.g. face recognition based on 2D or 3D images in visible light
spectrum). This article presents different methods of normalization and matching
algorithms and shows how they affect the performance of the biometric system.
Results of all described methods are summarized to see, which combination of
normalization and matching algorithms have the best performance.",
  address="University of Defence in Brno",
  booktitle="Proceedings of the Conference Security and Protection of Information 2011",
  chapter="76327",
  edition="NEUVEDEN",
  howpublished="print",
  institution="University of Defence in Brno",
  year="2011",
  month="may",
  pages="140--149",
  publisher="University of Defence in Brno",
  type="conference paper"
}