Surely at some point, we've been fascinated by fingerprints and how they can be used to solve crimes. Human fingerprints are unique and are fairly consistent over time. Thanks to technological advancements, fingerprint pattern recognition systems are becoming automated and are increasingly used in applications such as identity management and access control .
All humans are born with flow-like pattern of ridges and valleys on each finger. No two people have the same pattern – not even identical twins . Light injuries on the finger surface only damage the pattern temporarily, and hence, the ridge will reappear when the injury heals .
Fingerprint ridges can be categorised into three levels :
- Level 1 features (Patterns) refer to macroscopic details such as ridge flow and pattern type. Generally, there are three basic pattern types: arches, loops, and whorls. These types are still insufficient in fingerprint identification.
- Level 2 features (Points) are the minutiae such as when ridges split into two, ridge endings, "eyes", and "hooks" etc. These features are sufficient in establishing the individuality of fingerprints.
- Level 3 features (Shape) include all dimensional attributes of the ridge and other permanent details such as line shape, creases, pores, breaks, and scars etc. These features are what forensic experts look for.
Level 2 and 3 features can provide quantitative and qualitative information in identification, which are useful in latent or partial fingerprint examinations .
Fingerprints can be sampled using the following methods :
- The traditional "ink and paper" method involves applying ink to the finger surface, rolling the finger on a card, and scanning the card to generate a digital image.
- The manual "lifting" method usually refers to the dusting technique used to sample latent fingerprints during crime scene investigations.
- The automated "live-scan" method produces a digital image which is obtained by placing the finger on the surface of an electronic fingerprint reader/scanner.
For Automated Fingerprint Identification Systems (AFIS), there are two main types of "live-scan" sensors: optical and solid-state [1,5].
Optical sensors require the finger is to be placed on the top side of a glass prism (part of the sensor). Afterwards, one side of the prism is illuminated via a diffused light. The fingerprint valleys which have no contact with the glass platen reflect the light, whereas ridges that touch the platen absorb the light. This differential property of light reflection allows the ridges (which appear dark) to be discriminated from the valleys .
Solid-state sensors use silicon-based, direct contact sensors to convert the physical information of a fingerprint into electrical signals . Differences in physical properties, such as capacitance and conductance of the ridges and valleys are detected by pressing or sweeping a finger against the solid-state sensor .
Automated fingerprint recognition systems usually have two stages of operation: enrollment phase and identification phase .
During the enrollment phase , the sensor scans the user's fingerprint and converts it into a digital image. A minutiae extractor processes the fingerprint image to identify minutia points that are unique for every user. The system saves the minutiae information, such as location and direction, along with the user's information in the enrollment database.
During the identification phase , the user touches the same sensor, generating a new fingerprint image called a query print. Minutiae matching is one of the most common approaches used in fingerprint-matching algorithms: minutia points are extracted from the query print, and the matcher module compares the extracted set with stored data in the enrollment database to find the number of common minutia points. Minutiae that have similar location and directions are deemed to be matched. The match score between two fingerprints should be high for the same finger and low for those from different fingers.
Things become more difficult and complicated when altered, fake, or severely damaged fingerprints  come to play, but this should hopefully get you interested in fingerprint matching.
This is a great example of how technology can aid identity management, forensics and law enforcement.
- A. K. Jain, J. Feng, and K. Nandakumar, "Fingerprint Matching", IEEE Computer Society: Computer Magazine: February 2010, pp. 36-44, Feb. 2010. Web Link.
- A. K. Jain, S. Prabhakar, and S. Pankanti, "On the Similarity of Identical Twin Fingerprints", Pattern Recognition, vol. 35, no. 11, pp. 2653-2663, 2002.
- D. Braggins, "Fingerprint sensing and analysis", Sensor Review, vol. 21, no. 4, pp. 272-277, 2001.
- A. K. Jain, Y. Chen, and M. Demirkus, "Pores and Ridges: High-Resolution Fingerprint Matching Using Level 3 Features", IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 29, no. 1, pp. 15-27, Jan. 2007.
- T. Harris, "How Fingerprint Scanners Work", HowStuffWorks.com. Web Link.
- J. Feng, A. K. Jain, and A. Ross, "Fingerprint Alteration", MSU Technical Report, MSU-CSE-09-30, Dec. 2009.
About the author
Our monthly email newsletter keeps you up-to-date with our activities around the world.