A practical iris acquisition system and a fast edges locating algorithm in iris recognition
Yuanning Liu, Senmiao Yuan, Xiaodong Zhu, Qingliang Cui · 2004
Absfrod -A simple ondproeficol iris acquisition syxtem is established by scporcltimg light unif for iris rceognitiorr Based on directions of gradient vectors in on iris image. iris edges ore obfainel Thisprrper inlnodmedNor- mol Line concept, improved operations of fhe trudifiond Hough fmnslofion with the deJnition andgiven a fasf iris edges locafingproceeding. The U- perimenf results proved the validity ond e~riectne~~ of tho iris acquisition system and fhe lowtion algorifhm. enhancement algorithm should not suppose that it is easy to extract the local frequency and orientation information. This article presents a fast fingerprint enhancement algorithm. It can adaptively enhance the ridge and the furrow structure by use of the local section frequency information and the ridge orientation information, then the ridge and the furrow can be sorted out easily. At the same time, the key step of fingerprint identifica- tion is pattern matching, using the character description to tell whether the two graphs belong to the same person. Although thcre are various kinds of ways of fingerprint minutia descrip- tion, the existing problems in the pattem matching process are based on this different condition. For example, the observing point or sensor is different, or it is with the same sensor but in the different time. The two images will be different because of rotation, translation and scale. In the process of feature extrac- tion and model matching, this kind of difference is undefined. The position of trait point will be changed no matter what kind of way we choose. It is known that most pattern matching al- gorithms are based on the matching (4,5,6). These algorithms allow to lose some minutiae to a degree. The algorithm (7) is to allow that the minutia position is within a definited held, but these field could not be too large. And another algorithm (SI repeatedly improves each point which is correspondent with model fingerprint image until it meets ending condition. Al- though the algorithm bas been improved a lot in order to brief the complicated matching, it takes a long time to run because of its repeated operation features. According to above algo- rithms shortcomings that the speed is slow, and the precision is poor, this paper gives a algorithm which is based on three-order spline curve fitting curve of point matching, creation some def- initions, lemmas. The results indicate that the algorithm shows out excellent features in good velocity and precision. 11. PROPOSED APPROACH A. Iris Detection Hardware Now the known iris image acquisition equipment mainly include: the light refraction method put forward by Dr. Daug- man in Cambridge University and the light dispersion method put forward by Dr. Wildes in U.S.A. Compared with the above methods, our acquisition equipment system is shown as be- low (graph l), the features are described as followings. Its construction is simple and small in cnbage. Only one lamp is