Digital Image Processing

Open Access Open Access  Restricted Access Subscription Access
Open Access Open Access Open Access  Restricted Access Restricted Access Subscription Access

Fingerprint Image Quality and Prediction of Matching Performance


Affiliations
1 Vishvakarma Institute of Information Technology, Pune, India
2 Public Company in White Plains, NY, United States
3 Electrical and Computer Engineering Department at San Diego State University, San Diego, United States
4 Clarkson University, NY, United States
     

   Subscribe/Renew Journal


Due to their high reliability, fingerprints have been extensively used as a biometric identifier. The performance of an automatic fingerprint authentication system relies heavily on the fingerprint image quality as seen in several studies. In this work, we present a new method to quantify fingerprint image quality which is relevant to matcher performance. The ultimate goal of this research is to determine and overcome the underlying causes of the poor match. Newly developed wavelet features and previously developed spatial features, as inputs to a fuzzy c-means classifier, are used predict matcher performance. Results are obtained for two different matchers, namely NFIS bozorth3 and Verifinger (Neurotechnologija), for two different optical sensors,Crossmatch Verifier 300 and Secugen Hamster III.


Keywords

Fingerprints, Image Quality, Fingerprint Matchers, Performance Prediction, Fuzzy Classifier, Kullback-leibler Distances, Principal Component Analysis, Energy Distribution Analysis.
User
Subscription Login to verify subscription
Notifications
Font Size

Abstract Views: 225

PDF Views: 3




  • Fingerprint Image Quality and Prediction of Matching Performance

Abstract Views: 225  |  PDF Views: 3

Authors

Aditya Abhyankar
Vishvakarma Institute of Information Technology, Pune, India
Nilesh Kulkarni
Public Company in White Plains, NY, United States
Sunil Kumar
Electrical and Computer Engineering Department at San Diego State University, San Diego, United States
Stephanie Schuckers
Clarkson University, NY, United States

Abstract


Due to their high reliability, fingerprints have been extensively used as a biometric identifier. The performance of an automatic fingerprint authentication system relies heavily on the fingerprint image quality as seen in several studies. In this work, we present a new method to quantify fingerprint image quality which is relevant to matcher performance. The ultimate goal of this research is to determine and overcome the underlying causes of the poor match. Newly developed wavelet features and previously developed spatial features, as inputs to a fuzzy c-means classifier, are used predict matcher performance. Results are obtained for two different matchers, namely NFIS bozorth3 and Verifinger (Neurotechnologija), for two different optical sensors,Crossmatch Verifier 300 and Secugen Hamster III.


Keywords


Fingerprints, Image Quality, Fingerprint Matchers, Performance Prediction, Fuzzy Classifier, Kullback-leibler Distances, Principal Component Analysis, Energy Distribution Analysis.