On Shape-Mediated Enrolment in Ear Biometrics
Ears are a new biometric with major advantage in that they appear to maintain their shape with increased age. Any automatic biometric system needs enrolment to extract the target area from the background. In ear biometrics the inputs are often human head
- PDF / 6,090,806 Bytes
- 10 Pages / 430 x 660 pts Page_size
- 4 Downloads / 174 Views
Abstract. Ears are a new biometric with major advantage in that they appear to maintain their shape with increased age. Any automatic biometric system needs enrolment to extract the target area from the background. In ear biometrics the inputs are often human head profile images. Furthermore ear biometrics is concerned with the effects of partial occlusion mostly caused by hair and earrings. We propose an ear enrolment algorithm based on finding the elliptical shape of the ear using a Hough Transform (HT) accruing tolerance to noise and occlusion. Robustness is improved further by enforcing some prior knowledge. We assess our enrolment on two face profile datasets; as well as synthetic occlusion.
1
Introduction
Ears have long been considered as a potential means of personal identification, yet it is only in the last 10 years that machine vision experts started to tackle the idea of using ears as a biometric. Empirical evidence supporting the ear’s uniqueness was provided in studies by Iannarelli [1], conducted over 10,000 ears. Ears have appealing properties for personal identification; they have a rich structure that appears to be consistent with age from a few months after birth. Clearly, ears are not affected by facial expressions. Images of ears can be acquired without the subject’s participation and ears are big enough to be captured from a distance. However there exists a big obstacle - the potential occlusion by hair and earrings, which is almost certain to happen in uncontrolled environments. Up-to-date surveys of ear biometrics have recently been provided by Hurley et al. [2] and Choras [3]. Non-invasive biometrics such as gait, face and ear are often acquired with standard equipment, and they include unnecessary background information. The input samples of an ear biometric system are often head profile images with eyes, mouth, nose, hair and etc. as background. Enrolment (or registration) in an automatic biometric system is the process of detecting and isolating the area of interest. Yan et al. [4] developed an automatic ear biometric system which inputs 2D colour images and 3D range information of the human head profiles. Their enrolment uses a multistage process which uses both 2D and 3D data and curvature estimation to detect the ear pit which is then used to initialize an elliptical active contour to locate the ear outline and crop the 3D ear data. Chen et al. [5] G. Bebis et al. (Eds.): ISVC 2007, Part II, LNCS 4842, pp. 549–558, 2007. c Springer-Verlag Berlin Heidelberg 2007
550
B. Arbab-Zavar and M.S. Nixon
also use 2D and 3D face-profile images and the ears are detected by a two step alignment procedure, which aligns edge clusters with an ear model based on the helix and anti-helix of one ear image. Alvarez et al. [6] fits an ovoid model to the ear contour. However, their algorithm initializes with a manual estimate of the ear contour. These studies have not presented an evaluation of occlusion tolerance. Yan et al. merely suggest that their method is capable of handling small amounts of occlusion b
Data Loading...
