Thermal hand image segmentation for biometric recogniton

Image segmentation is a first step for most of the biometric systems. In this paper we present a hand image segmentation algorithm for thermographic images. This work is based on a specially acquired database with a thermographic camera TESTO 882-3, which acquires thermal images of 320×240 pixels as well as visible images of 640×480 pixels. Experimental results reveal that thermal images are harder to segment from the background than visible images. Specially for some users. However, thermal images provide some interesting information, such as the vein pattern distribution.