Dr. Ted Dunstone.
Biometrics has always been an interesting field because it is at the intersection of a number of disciplines; biology, statistics, computer science and engineering. However, the field can be simply defined as the use of a distinctive and relatively stable physical or behavioural characteristic of a person for the purpose of recognition. Given this wide scope for what can be considered a biometric, the mainstream biometrics (fingerprint, iris, face, hand, voice, signature recognition) only scratch the surface of the possible techniques.
There are many alternative biometrics technologies that that various organisation have developed over the years, some of which are complementary (assisting an existing biometric system, such as skin type recognition for facial recognition), while others are essentially orthogonal – providing additional biometric checks leading to improved resistance to attack (such as pulse recognition). There are a few that propose using exotic sensor types for recognition, such as brain wave detection.
These biometric systems are related to alternative biometric types, not a conventional type with a different means of capture. Fingerprint capture can, for instance, be achieved in several different ways, many of them quite niche, including ultrasonic reading and special light-emitting polymers. These capture systems, however, still use the same or similar fingerprint information as the more common optical or capacitance readers.
These less mainstream biometrics provide several useful benefits when combined with the mainstream techniques. The additional information gathered can help ensure inclusiveness for those people who are unable, or find it difficult, to use one or more of the traditional biometric identifiers, and it can also be used to improve security and accuracy.
As biometric systems protect more critical infrastructure or are used in more places (for example, for passports), bio-hackers will continually evolve new ways to break a biometric system. The more esoteric biometrics techniques will help to keep ahead of these hackers. It is likely that in the future, these types of multi-sensory systems will become the norm, where a variety of sensors will be used to come to a decision on identity.
One of the biggest problems with the more esoteric biometrics is the lack of data both for development and testing. In many cases, these biometrics may only have been examined on a small sample of people (sometimes even just the people from the lab) and this is seldom indicative of real world performance. Neither stability nor uniqueness of a biometric can be established until a properly controlled (and ideally independent) trial has taken place under real world conditions. Many other issues, including susceptibility to environmental effects and the durability of sensors, also need to be considered. The low cost and vastly improved performance of traditional biometric systems creates an increasingly high barrier for the entry of new techniques; however, there is always an opportunity for a new technology where significantly enhanced levels of performance are achieved.
There are a diverse collection of techniques for alternative biometrics. The following brief descriptions, in no particular order, provide a selection of some of the more obscure techniques.
- Pore recognition. Even the smoothest skin is covered with minute pores. The pattern produced by these pores is distinctive for every patch of skin, leading to a technique called pore recognition. It has been used for improving fingerprint recognition; however, reading the pores consistently requires much higher resolution than existing scanners provide.
- Ear recognition. This can be used to recognize people from the side in cases where the face may be obscured and can be used as an adjunct to face recognition. However, the ear is not always visible, depending upon people’s hairstyle or earrings.
- Infrared detection and recognition. This technique depends upon the body heat emitted, generally from the face. The heat signature of an individual can vary significantly depending on what they are wearing or doing, or from different environmental conditions. However, the structure of the branching of blood vessels in the face supplies a unique pattern of heat flow to the face that will remain consistent independent of make-up or minor cosmetic surgery.
- Skin patterns. A recently developed technology that is in commercial use analyses the skin patterns of individuals. The algorithm uses the skin regions around the cheeks and exploits the high-frequency information found there to carry out recognition. Previously, this information was just thrown away as being noise. Its strength also lies in the fact that it can be easily combined with existing face recognition systems.
- Skin composition. The optical properties of skin have also been explored recently by a partnership including Johnson and Johnson. Using spectroscopic techniques, the structure and composition of the skin can be determined from its optical pattern. This has been used to enhance fingerprint recognition systems by providing additional recognition information.
- DNA matching. DNA matching is not esoteric in its current widespread use for forensics, where the sample can be examined in a lab over a number of hours. What is new is the capability to analyse the DNA in real time on the surface of a chip. The technology for undertaking this has yet to be perfected for recognition purposes. However, continuing advances, such as the rapid screening of genetic illnesses or biological pathogens, make it certain to be practical in the near future. DNA has the drawback that identical twins share exactly the same information, although this could be compensated for by also using another biometric. Another problem is that we leave samples of our DNA everywhere, so potential contamination, either intentional or otherwise, will be a real issue.
- Gait recognition. The way we move (our gait) is also relatively unique as it depends upon the length of our limbs and the conditioning of our muscles. At one time or another we have all identified someone when their back is turned, mainly by the way they are walking. The technology has been developed primarily for military surveillance purposes as it can be used from a long distance away when other visible cues are masked.
- Heartbeat recognition. Despite our pulse varying in speed, there are a number of features that make each persons heartbeat distinct. These features can be used to validate an identity and are particularly useful for fingerprint systems where it is possible to both read the fingerprint and take the pulse simultaneously. The finger, however, needs to be pressed for some duration for a sensor to obtain an accurate identity reading.
- Brain imaging. Advances in scanning technology will eventually improve to the point where brain imaging is possible without the use of cumbersome and expensive NMI equipment (although it will still probably require the user to wear some kind of headset). Under these circumstances, the way the brain responds to stimulus will be highly specific to a particular individual and so could be certainly be used for recognition.
- Odour recognition. Despite our attempt to scrub and clean away body odour, a residual and unique smell is always present and can be detected with sensitive equipment. This is substantiated by the fact that many animals will learn to respond to their owner based on smell alone. Odour recognition has also advanced significantly recently and it is being used increasingly for the detection of explosive substances in air transport.
- Footprints. Shoe recognition does not really fall into the category of a biometric since a shoe can be given to someone else (however, it is a useful forensic technique). However, footprints are relatively unique (just like the palm) and are sometimes used (as just an ink impression) in hospitals on born infants as a form of identity.
- Fingernail recognition. The ridges of fingernails are claimed to be usable for recognition purposes. Again, this technology can act as a complementary recognition system along with fingerprint or hand geometry readers.
Of those techniques proposed above that have made it off the drawing board, most have suffered significant commercialization problems – often due to accuracy or user inconvenience factors. Some, however, will become more relevant in the future as the technology improves and as the importance of multi-sensory systems becomes better understood. And a few are now making in-roads into enhancing more mainstream recognition technology.
Dr. Ted Dunstone is the CEO of Biometix (www.biometix.com), a leading provider of biometric security assessment services and tools. Over the past fifteen years, he has provided consultancy services to many government agencies and is a frequent contributor to government policy and debate on the application of biometrics. He also founded the non-profit Biometrics Institute in 2001 (a leading Asia-Pacific policy body) and is currently the technical chair. In 2003, he was the recipient of the NSW Pearcy Award, which is given for innovative and pioneering achievement and contribution to research and development within the IT&T industry.
