Proc. SPIE. 9138, Optics, Photonics, and Digital Technologies for Multimedia Applications III
KEYWORDS: Facial recognition systems, Detection and tracking algorithms, Visualization, Imaging systems, Cameras, Databases, High dynamic range imaging, Associative arrays, Information visualization, Time multiplexed optical shutter
The gaining popularity of the new High Dynamic Range (HDR) imaging systems is raising new privacy issues caused by the methods used for visualization. HDR images require tone mapping methods for an appropriate visualization on conventional and non-expensive LDR displays. These visualization methods might result in completely different visualization raising several issues on privacy intrusion. In fact, some visualization methods result in a perceptual recognition of the individuals, while others do not even show any identity. Although perceptual recognition might be possible, a natural question that can rise is how computer based recognition will perform using tone mapping generated images? In this paper, a study where automatic face recognition using sparse representation is tested with images that result from common tone mapping operators applied to HDR images. Its ability for the face identity recognition is described. Furthermore, typical LDR images are used for the face recognition training.