A Multiple Indexes Quality Assessment for Fingerprint

Abstract

The performance of fingerprint recognition relies heavily on the quality of fingerprint images. In this paper, a multiple indexes fingerprint quality assessment is proposed for poor fingerprints. This method fuses seven indexes from three kinds of typical fingerprint features (gray features, local features and global features) by a support vector machine classifier. Experiment results on FVC2004 database show that our proposed method can identify poor quality fingerprint accurately (94.3%-98.7% for different sub databases).