Termal faceprint characterization using the image foresting transform

Abstract

Although the digital fingerprint is one of the most used biometric recognition approach, this one is very sensible to fingers' imperfection, which can be congenital or acquired with time. The iris recognition is an alternative, however such kind of systems still have prohibitive costs and are uncomfortable enough to the user. Another approaches have been used for face recognition instead of the aforementioned ones because of its invasiveness. The main problem of these approaches is the excessive brightness and face occlusions. In such a way, termal face images have been employed for face recognition purposes due to the different temperatures in certain regions of the face that allow to distinguish one person from other. However, some diseases can affect these temperature and, consequently, the infrared images. Recent advances have been proposed to overcome this situation by obtaining a faceprint and determining the isothermical regions. The present work has an objective to improve such techniques, that are related with face segmentation and isothermical regions identification for further classification, using a graph-based image processing technique named as Image Foresting Transform, which has been extensively used in all steps that will be necessary for the implementation of our work, but was not already used for these purpose.