earticle

영어

This paper presents a report on the experimental study of the impact of false minutiae on the performance of fingerprint matching systems. A 3-tier algorithm comprising of pre-processing, minutiae extraction and post-processing stages formed the backbone of the experiments. The pre-processing stage enhanced the fingerprint image, the minutiae extraction stage used the minutiae properties to detect and extract true and false minutiae points while the post-processing stage eliminated the false minutiae points. The experiments were performed on the four datasets in each of the three standard fingerprint databases; namely FVC2000, FVC2002 and FVC2004. The completion times for the minutiae extraction and the post-processing algorithms on each dataset were measured. A standard fingerprint matching algorithm was also implemented for verifying the impact of false minutiae points on FAR, FRR and the matching speed. Analysis of the obtained results revealed that for reliable and optimal performance of fingerprint matching systems, false minutiae points must be eliminated as much as possible from their operations.