hands

This page looks at hand-based biometrics, including fingerprints and hand geometry devices.

It covers -

• introduction
• fingerprints
• palmprints
• hand geometry
• thermograms

    introduction

For most people fingerprints are the biometrics. In contrast to retina and iris scans their use as a biometric identifier is so embedded in legislation, police practice and popular culture as to be perceived as 'normal' or 'benign'. However, research and implementation has encompassed a range of hand-based biometrics. Some US states for example have established whole-of-government palmprint registers for access to data networks. A range of organisations have used different hand-based technologies for managing access to particular facilities.

The following paragraphs discuss those technologies and highlight points of entry to the literature regarding research and application.

    fingerprints

As the preceding pages noted, fingerprinting quickly displaced bertillonage as the default biometric in all major jurisdictions, with establishment of large-scale fingerprint registers and recognition by courts.

Use of the biometric is based on the uniqueness of patterns - often characterised as whorls and loops - on the surface of an individual's fingers. Those patterns are stable and generally readily detectable after early childhood (major exceptions relate to people whose work or hobbies involve sustained exposure to acids that scar the finger tips or to ink that impedes capture of an image). They thus meet the essential test for a biometric, ie they can be recorded, used for reference purposes and uniquely identify a person for life.

Aficionados of film noir or crime fiction will be familiar with use of fingerprints in law enforcement -

• contact between the finger/s and many substances leaves an image of the pattern that can often be discerned (eg by use of a powder or other technique that aids identification by the naked eye and recording by a camera)
• that image that can be compared with an existing record of the print/s (eg obtained when the individual was previously incarcerated or provided a set of prints for identification purposes as a government employee)
• the matching can be used to confirm or disprove the individual's presence at a particular location, participation in a particular activity or other matter.

The configuration of lines on an individual print can be mapped, with relationships between points in that geography being identified by naked eye (the basis of most searching of fingerprint registers prior to the 1970s) or by machine-based vision. Reduction of those relationships to mathematical expressions enables large-scale automated searching of electronic databases, in essence matching one set of set of expressions with another in order to match two records.

Fingerprints were initially recorded by inking the person's finger's and then rolling the selected finger or all fingers over a pad, with the resulting image being photographed and often categorised with a numerical code for easy sorting. Recent years have seen use of camera-like devices to capture high-resolution images directly from the fingers and use of contact sensors (including electronic field imaging, surface capacitance and thermal imaging).

Those sensors might be used as a gateway restricting access to a campus, building or room. They might instead guard access to a particular device: vendors have released thumbprint readers that a standalone or are incorporated in laptops, desktop personal computers and even mobile phones. Some security solutions involve comparison of data on a smart card with a scan of the individual's finger, typically compared with a master register.

The effectiveness of readers and the underlying software for analysis of images and matching with the 'reference' image on a timely basis varies considerably. As you might expect, the performance of a specialist perimeter control device that is tied to a dedicated server (with sophisticated software and considerable processing power) is typically better than an off-the-shelf device sold at your local IT shop for attachment to your personal computer.

Although there is disagreement about claimed (and achieved) performance the poorest systems appear to offer a false accept rate of around 1:1,000 and a false reject rate of around 1:100. False accept rates for superior systems are around 1:1,000,000.

Some perimeter control systems feature tests for 'liveness', reflecting incidents where researchers have subverted the technology by using gelatine impressions of fingers and gummi bears or by using a severed finger.

The number of devices in use is not known. It is clear that national and provincial governments have built up large fingerprint databases (the Australian registers are highlighted here and here), including collections relating to criminal convictions, criminal investigation and identification of government employees/contractors. Some private organisations have also accumulated substantial collections in identifying workers.

For introductions see Simon Cole's Suspect Identities: A History of Fingerprinting and Criminal Identification (Cambridge: Harvard Uni Press 2001), Handbook of Fingerprint Recognition (Berlin: Springer 2005) by Davide Maltoni, Dario Maio, Anil Jain & Salil Prabhakar, Automatic Fingerprint Recognition Systems (Berlin: Springer 2003) edited by Nalini Ratha & Ruud Bolle and Salil Prabhakar & Anil Jain's 2002 'On the Individuality of Fingerprints' in 24 IEEE Transactions on PAMI 8 (PDF).

For the symbolic changes fingerprinting brought to policing see: Dean Wilson's 'Traces and transmissions: techno-scientific symbolism in early twentieth-century policing' in Crime & Empire 1840-1940: Criminal justice in local and global context (Cullompton: Willan Press 2005) edited by Barry Godfrey & Graeme Dunstall.

    palmprints

The same principles can be used to electronically map and compare the patterns on palms, with some major organisations using palmprint readers as the basis of perimeter control schemes.

The rationale is typically that -

• a palmprint offers a larger area for identification and is less likely to be obscured, of importance in rapid capture of information
• palmprint recognition is perceived as more benign than fingerprints, ie supposedly does not have negative connotations of fingers being recorded by a provincial policeforce or national security agency

The size of palmprint databases across the world is unknown. Most systems appear to be isolated, ie are used for perimeter control by a particular organisation rather than being systematically collected for sharing by a group of government agencies and deployment as part of passport or other ubiquitous identity schemes.

The salient work on palmprinting is D Zhang's Palmprint Authentication (New York: Kluwer 2004)

    hand geometry

Hand geometry schemes are an echo of bertillonage, aiming to uniquely identify individuals by mapping the shape of a person's hand and supplying a mathematical expression for the relationship between different features of that hand (eg length of each finger, distance to wrist and so forth).

Early systems involved caliper-style measurement or even x-rays on the basis that the best data would be obtained by deriving relationships from a measurement of bone. Current systems, which have essentially remained within the laboratory, seek a less threatening measurement. They have been criticised as providing poorer results than fingerprint or iris/retina recognition and as being commercially uncompetitive. They are unlikely to be adopted on a large scale.

A point of entry to the literature is provided by Leena Ikonen's 2003 Hand Geometry-based Biometric Systems paper (PDF).

    thermograms

Enthusiasts have proposed use of thermograms, ie infrared images of vascular patterns in the back of an individual's hand (in contrast to the palm patterns noted above).

It has been claimed that images of the location and thickness of the veins in a hand are sufficiently stable and unique to enable verification of a person's identity on an ongoing basis. Vascular patterns vary from left to right hand (and are not shared by twins).

Capture typically involves placing the hand on the surface of a reader that takes an infrared scan, with the resulting image being compared to a reference database to verify the identity.

Commercial application of the technology has been limited, with organisations apparently preferring palmprint readers. There have been no national rollouts and none appear likely in the near future.

An introduction is provided by Satu Alaoutinen's 2003 Biometrics in infrared/near-infrared band (PDF).

Hitachi began introducing a finger vein biometric system in 2006, which as usual was claimed to be "the fastest and most secure biometric". It verifies a person's identity based on blood vessels under the skin in a finger.

The pattern of blood vessels is captured by transmitting near-infrared light at different angles through the finger, typically the middle finger. That light is partially absorbed by haemoglobin in the veins, with the pattern being captured by a camera as "a unique 3D finger vein profile" which is converted to a digital code that is matched with an existing profile to verify the individual's identity.

Hitachi claims that because the veins are invisible to the eye the image is "difficult to forge and impossible to manipulate" and that "it is extremely unlikely that finger vein profiles can be taken" without individuals being aware of that capture.

   next part  (kinetics)