15 October 2012 Intensity-based registration and fusion of thermal and visual facial-images
Author Affiliations +
Abstract
Fusion of images from different modalities provides information that cannot be obtained by viewing the images separately and consecutively. Automatic fusion of thermal and visual images is of great interest in defense and medical applications. In this study, we implemented automatic intensity-based illumination, translation and scale invariant registration of deformable objects in thermal and visual images by maximization of a similarity measure such as generalized correlation ratio. This method was originally used to register ultrasound (US) and magnetic resonance images (MRI) successfully. In our current work, we propose a major modification to the original algorithm by investigating appropriate information content in the input data. The registration of facial thermal and visual images in this algorithm is achieved by maximization of the similarity measure between the input images in the appropriate image channel. The algorithm is tested using real facial images with illumination, scale, and translation variations and the results show acceptable accuracy.
© (2012) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Musa Serdar Arslan, Musa Serdar Arslan, Mohamed I. Elbakaray, Mohamed I. Elbakaray, Shamim Reza, Shamim Reza, Khan M. Iftekharuddin, Khan M. Iftekharuddin, } "Intensity-based registration and fusion of thermal and visual facial-images", Proc. SPIE 8498, Optics and Photonics for Information Processing VI, 84980C (15 October 2012); doi: 10.1117/12.930972; https://doi.org/10.1117/12.930972
PROCEEDINGS
9 PAGES


SHARE
Back to Top