A novel approach to fingerprint pore extraction

Automated fingerprint identification systems (AFIS) have become a popular tool in many security and law enforcement applications. Most of these systems rely on the matching of fingerprints using the position and orientation of ridge endings and bifurcations within the fingerprint image. While this information is sufficient for matching fingerprints in small databases, it is not discriminatory enough to provide good results on large collections of fingerprint images. This paper presents a means of obtaining additional discriminatory information from fingerprint images by demonstrating a novel method to extract the locations of sweat pores from the grayscale fingerprint image. This is achieved through the implementation of a modified minimum squared error approach. The proposed algorithm is capable of obtaining good results even from images obtained from basic 500 dpi optical live-scan devices despite the common belief that images obtained at this resolution are not of high enough quality. Results of the proposed method are demonstrated on fingerprint images taken both from a common live-scan device and the inked prints of the NIST 4 database. An explanation of the approach is presented, the results are discussed, and future research possibilities are put forth.