Fingerprint
Recognition
Of all the biometrics, fingerprint based personal identification (PI) is the most mature, proven, and accepted technology [1] [2] [3] [4]. It's not surprising then that fingerprint based PI is the most active area of biometric research, development, and applications. This is due (at least in part) to: (a) fingerprints are the primary means of identification used by Governments and law enforcement agencies the world over, and (b) finger-scan technology has matured to the point where it is relatively inexpensive, easy to integrate, manage, and use. Moreover, finger-scan (often called live-scan) technology's replacement of the messy ink-and-roll fingerprint acquisition procedure has reduced the criminal stigma associated with fingerprints. A fingerprint is the pattern of ridges and valleys on the surface of the finger. The two primary template matching technologies used in fingerprint based PI are minutia matching (minutia are local ridge discontinuities) and global matching (correlation of global ridge patterns). In addition to these, there are some promising techniques purported to incorporate the best discriminative features of both [5] [6] [7]. Fingerprints have traditionally been classified according to their global ridge patterns using the Henry System [4]. Well over one hundred years old, the primary Henry System ridge pattern classifications include: the left loop, the right loop, the arch, the tented arch, and the whorl. If a fingerprint based PI system is operating in "identification mode" (as with most law enforcement and forensics applications), the template database can be sub-divided into ridge pattern classifications to help reduce the identification time. For example, if a live-scan fingerprint is classified as an arch, the identification time would be reduced if only the arch portion of the template database were searched. This is important when dealing with huge databases (like the FBI's) containing millions of fingerprints. If a fingerprint based PI system is operating in "verification mode," ridge pattern classifications typically offer little benefit and are rarely used .
A fingerprint is the pattern of ridges (usually appearing dark in an image) and valleys (usually appearing light in an image) on the surface of the finger. Fingerprints are fully formed around the seventh month of gestation and adhere very well to the four biometric characteristics described above. A person's fingerprints will remain essentially constant throughout their life unless their hands are exposed to excessive or repetitive abrasions such as those encountered by people who perform certain kinds of manual labor. For example, the work performed by a brick mason can temporarily alter the "permanent" characteristic of their fingerprints, but the original ridges soon return if the labor is stopped. Injury appears to be the only method by which the "permanent" characteristic can be compromised. A fact that emphasizes the uniqueness of fingerprints is that they are unique even when considering genetics. That is to say that, not only do identical twins have unique fingerprints, the ten fingerprints of an individual are each unique. This is the main reason that fingerprints are an excellent biometric. The primary discriminating features that make each fingerprint unique are their minutia (also called Galton features). Minutiae are local discontinuities in the otherwise smooth flow of ridges. Of these, ridge endings and bifurcations can be considered a basis set for all other minutia. That is, all other minutia can be constructed using ridges endings and bifurcations. For example, a lake can be thought of as two bifurcations, an independent ridge is merely two ridge endings, etc. Because of this, most minutia matching fingerprint based PI systems restrict themselves to these two minutia types. Additional discriminating features include the core and the delta(s) (also called singular points). The core of a fingerprint can be thought of as the center of the pattern. A core is defined as the topmost point on the innermost upward recurving ridge. A recurving ridge is one that u-turns. A delta is a triangular series of ridges that are normally found at the lower right and/or left of a fingerprint. A delta is defined as the center of a triangular region where there is a convergence of ridges that flow from three different directions [5]. It should be noted that (unlike minutia) not all fingerprints possess these features. Moreover, even when present, these features may be corrupt or truncated in the image capture process.
The fingerprint is the most prevalent biometric used in PI systems to date. One reason for this is that fingerprints have been (for many years) the primary means of PI used by law enforcement agencies the world over. Fingerprint based PI systems work well in user "identification mode," although the manageable template database volume may be smaller than eye-scan based technologies. Fingerprint technology requires physical contact and usually a small amount of user training. Once educated, users typically have few problems and accept the technology readily. Of all the biometric technologies, fingerprints account for not only the largest market share, but the largest research and development budget as well. Because of this, fingerprint technology is flourishing thus making PI implementation cheaper, faster, and easier. So much so that it is expanding beyond corporate to personal use. With the large offering of inexpensive image capture devices and associated software development kits (SDKs), fingerprint technology makes an ideal starting point for the PI system developer. However, before a significant investment is made, application specific performance capability should be verified.
[1] Julian Ashbourn, "Biometrics: Advanced Identity Verification, The Complete Guide," Springer, London, 2000.
[2] A. Jain, R. Bolle, S. Pankanti, editors, "BIOMETRICS Personal Identification in Networked Society," Kluwer Academic Press, Boston, 1999.
[3] D. Zhang, "AUTOMATED BIOMETRICS Technologies and Systems," Kluwer Academic Publishers, Boston, 2000.
[4] L. Jain, et al, editors, "Intelligent Biometric Techniques in Fingerprint and Face Recognition," CRC Press, Boca Raton, 1999.
[5] A. K. Jain, S. Prabhakar, L. Hong, and S. Pankanti, "Filterbank-based Fingerprint Matching," IEEE Transactions on Image Processing, Vol. 9, No.5, pp. 846-859, May 2000, http://biometrics.cse.msu.edu/publications.html. Last accessed: 31 July 2001.
[6] A. K. Jain, S. Prabhakar, L. Hong, and S. Pankanti, "FingerCode: A Filterbank for Fingerprint Representation and Matching", Proc. IEEE Conference on CVPR, Colorado, Vol. 2, pp. 187-193, June 23-25, 1999, http://biometrics.cse.msu.edu/publications.html. Last accessed: 31 July 2001.
[7] A. Saleh, R. Adhami, "Curvature-based matching approach for automatic fingerprint identification," Proceedings of the 33rd Southeastern Symposium on System Theory, pp. 171-175, March 18-20, 2001.
