ARTICLE IN PRESS Pattern Recognition ( ) –.pdf

U N C O R R E C T E D P R O O F PR1890 pp: 1--11 (col.fig.: Nil) PROD. TYPE: COM ED: Pallabi PAGN: Vidya -- SCAN: Venki ARTICLE IN PRESS Pattern Recognition ( ) – /locate/patcog 1 A combination model for orientation eld of ngerprints Jinwei Gua, Jie Zhoua ;?, David Zhangb3 aDepartment of Automation, Tsinghua University, Beijing 100084, China bBiometrics Technology Centre, Department of Computing, Hong Kong Polytechnic University, Kowloon, Hong Kong5 Received 6 December 2002; accepted 14 May 2003 Abstract7 Orientation eld is a global feature of ngerprints that is very important in automatic ngerprint identication systems (AFIS). Establishing an accurate and concise model for orientation elds will not only improve the performance of orientation9 estimation, but also make it feasible to apply orientation information in the matching process. In this paper, a novel model for the orientation eld of ngerprints is proposed. We use a polynomial model to approximate the orientation eld globally and11 a point-charge model at each singular point to improve the approximation locally. These two models are combined together by a weight function. Experimental results are provided to illustrate the fact that this combination model is more accurate and13 robust with respect to noise compared with the previous works. The application of the model is discussed at the end. ? 2003 Published by Elsevier Ltd on behalf of Pattern Recognition Society.15 Keywords: Fingerprint identication; Orientation eld; Polynomial model; Point-charge model; Weighting factor 1. Introduction Among various biometric techniques, automatic nger- print identication systems (AFIS) are the most popular19 and reliable for automatic personal identication. During the last years, ngerprint identication has received increasing21 attention and the performance of ngerprint identication systems has reached a high level. However, it is still not23 satisfactory for a large database or ngerprints with poo