|Institution:||University of Notre Dame|
|Department:||Computer Science and Engineering|
|Keywords:||computer vision; iris biometrics|
|Full text PDF:||http://etd.nd.edu/ETD-db/theses/available/etd-10222012-113048/|
Iris biometric sensors and systems have advanced in complexity and performance as more detailed models of iris appearance are explored by the research community. Current research in iris biometrics works to enhance iris sensors, improve matching algorithms, and accommodate more characteristics of the human iris. However, many intrinsic human and sensor based factors and extrinsic environmental factors affect the advancement of iris biometrics. In this thesis, we examine several types of intrinsic (human and sensor based) and extrinsic (environmental) factors and their effects on iris recognitions. In particular, we study the illumination scheme of the LG IrisAccess 4000 and propose a diffuse illumination system in order to reduce the effects of specular highlights, an intrinsic sensor based factor. We also look at a human based intrinsic factor, eye dominance, as well as an extrinsic factor in the usages of the LG TD 100. Our experiments suggests that eye dominance does affect iris recognition performance, whereas variations in the use of the LG TD 100 does not.