Fingerprint sample quality assessment via ridge line count using Laplacian of Gaussian edge finding

We propose a new fingerprint sample quality measurement algorithm based on local fingerprint area analysis and ridge line counts. The proposed method determines ridge orientation of local areas with Principal Component Analysis, finds edges using the Laplacian of Gaussian and zero-crossings approach, and counts valley to ridge transitions across the center of the area to assign a local score. The local scores are aggregated to assign a single quality value for the fingerprint. We assess the performance of our method with Error versus Reject Curves obtained from comparison scores from three fingerprint comparators on CASIA Fingerprint V5.0 and FVC 2004 DB 1A datasets. The quantitative evaluation shows that the proposed method achieves predictive performance on level with NFIQ while achieving higher fidelity.