Home

About

Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network
FAQHELPSITE MAPCONTACT US


  Advanced Search  

 
Registration of Infra-Red and Color Images for Multimodal Face Recognition
 

Summary: Registration of Infra-Red and Color Images for Multimodal
Face Recognition
Faysal Boughorbel, Besma Abidi, and Mongi Abidi
Imaging Robotics and Intelligent Systems Laboratory
The University of Tennessee, Knoxville, Tennessee, 37996
fboughor@utk.edu
1. Introduction
The fusion of IR and color imagery for face recognition was shown to increase significantly the recognition rates of
currently deployed systems [1]. This improvement is due to the important amount of independent information
available in these modalities. In this context, shortcomings of color imagery such as sensitivity to illumination
changes can be compensated for by fusion with IR data. Furthermore, while color imagery provides information
about the surface of the face, IR images show the blood vessel and heat emission patterns unique to every person.
Another important application of IR-color fused images is the location of the eyes for recognition purposes. The
main practical restriction on using hardware registered IR-color images is the elevated cost of cameras with special
optics (more than $50k), which limits their large scale deployment in most real environments where face recognition
is needed, such as in airports. In this research we investigate the task of registering images acquired separately by
uncalibrated IR and color cameras. One of the few general purpose IR-visual registration approaches, proposed by
Irani et al [2], defined a criterion based on local statistical correlation of feature maps along with global motion
models and was mostly applied to outdoor man-made scenes. In our work we developed an efficient Gaussian
criterion for point-sets registration under various motion models and applied it to edge maps extracted from the

  

Source: Abidi, Mongi A. - Department of Electrical and Computer Engineering, University of Tennessee

 

Collections: Computer Technologies and Information Sciences