The future of biometrics depends upon the development of accurate and repeatable methods of identifying or authenticating human characteristics. Many of our human physical attributes have been identified as potential unique vehicles to achieve this but presently each method has its own set of limitations. For widespread usage, and to succeed, these technologies must become foolproof, since the sheer nature of their intended applications demands that as a matter of course they should expect attempted breaches.


Facial Recognition
Visual recognition is frequently used today as a form of identification but it is mainly dependent upon human intervention to clarify that, a face matches a card for example. In terms of turnover, facial scanning is the second largest revenue provider after fingerscan technology for the biometrics industry according to the figures produced by IBG in 2003.

Software driven biometric facial recognition aims to take away the necessity for human presence in the authentication process. This usually relies upon devised algorithmic measurements that interpret the characteristics and features of a presented face to determine the possibility of a match to either a Photo ID/Pin card or to a photographic database. In simple terms, a unique multi-character facial "password" is created for each image. Technically, this is achieved by the development of algorithmic calculations through software that extracts and interprets the information for each face using varying scientific methods, such as Feature Mapping using Local Feature Extraction or Automatic Face Processing, or by the Eigenface or Neural Networks methods. Data is collected through 2D or 3D images from either static cameras or live video images, and in either monochrome or colour. Thermal imaging techniques are also used.

Facial recognition is probably one of the most socially acceptable biometrics, as visual recognition and interaction is the method that we all use in our everyday lives. The technology is relatively simple to integrate into other existing systems as photos form the basic authentication format for our driving licences, passports and other forms of access control ID. Image acquisition is considered to be generally none intrusive since we are all aware of constant camera presence around us today in offices, shopping centres and many other public areas. The challenges involved in developing facial recognition systems, as with most biometrics software, are complex and involve both human and technical aspects. The human face is subject to change due to a whole variety of reasons, such as ageing, skin tone, religious attire, illness, wearing glasses, facial hair and expression, all of which in terms of access control could cause “nuisance” problems and could require either a data base with several images of the same subject and/or a regularly updated file image. Lighting levels, weather conditions, angle of image and degree of subject cooperation will also have an effect on the performance of the software. In terms of covert use for this type of technology there are many objections with regard to protection of privacy where individuals could be identified without their knowledge or consent. However, under these circumstances with such an array of potential barriers to recognition, if an individual did not wish to be recognised then an automated facial recognition system alone would probably not be the most of effective tool.


Voice Scan Techniques
The voice is probably the only feasible biometric for recognition over the telephone, and the capture and recognition methods are certainly non-obtrusive and socially acceptable. It is thought that each individual has a unique voice pattern determined by sex, physical build, such as differences in vocal chord length, mouth shape, nasal cavities etc, and characteristics such as frequency, amplitude, rhythm and harmonics. In a similar way to the facial recognition techniques the voiceprint of an individual is analysed and allocated a unique multidigit pin or password by means of a specifically derived software programme. Live speech is then compared to the filed voice-print for authentication. This method would be simple to incorporate into already existing telephonic or audio password based security or access procedures as both the combined spoken password and the voiceprint can be authenticated simultaneously.

As we discovered above, facial recognition has both positive and negative aspects with regard to its reliability and accuracy, and so to do voice recognition techniques. Although some individuals are gifted at mimicking others, manufacturers claim, that generally, voice verification systems cannot be fooled by recordings or mimics. However there are still environmental and technical issues to consider, such as interference from background noise, equipment tone variation, and of course the human element such as vocal characteristics changing due to illness, stress, mood and age.


Focus on the Eyes
For many technological and sociohuman reasons, we have seen from the above that facial recognition and voice scan technologies are not ideal solutions alone. On the other hand iris recognition systems to date, given cooperation of the individual, are proving to be one of the most reliable forms of biometric identification under industry tests, for use when imaging can be done at distances of under approximately 1 metre. Although small about 11mm diameter, and sometimes difficult to photograph, the pattern variability of the iris among different people is said to be enormous, as each individuals iris is calculated to have an average of around 260 unique features. Together with some very clever algorithmic software this has contributed to the highly accurate test results for the technology and has also given it a significant advantage when there is necessity to search particularly large databases without encountering false matches despite the high volume of possibilities.

In addition, the iris is easily and uniformly located in all individuals, and although the iris is externally visible it is well protected from the environment making it a very stable organ over our lifetime with very few if any changes. In terms of image capture it is only minimally affected by angle of illumination, and performance is not limited by different characteristics through racial differences such as very dark eyes, or narrow eyes. Furthermore, identification is equally as good through glasses, contact lenses and goggles and it can be used in dark environments with the use of infrared illumination.

There are of course some potential problems, such as medical conditions like the formation of cataracts, and interference in performance would be caused through mirrored sunglasses and different coloured contacts, but the latter two are really dependent upon a degree of cooperation from the user.

In addition to iris recognition, retinal scanning is a further option as a biometric identification method. Retinal scanning is based on the characteristics of the patterns of blood vessels at the back of the eye. This has been used in the past for particularly high security applications, and is probably less likely to become more widely used than iris scanning as it is considered to be somewhat intrusive, and, in former years before the technology was more advanced, there were arguments indicating that the scanning method could be damaging to the eye. The capture can be quite time consuming compared other biometrics capture methods, and involves total cooperation of the user. The individual must look through a specifically designed device where a low-intensity light source is directed through the pupil to illuminate the retina and performs a 306 degree circular scan. Due to the location of the retina and the complexities involved this technology carries a high degree accuracy and stability. The obtrusive nature of the image capturing equipment guarantees good focus and alignment, and more or less eliminates fraudulent access attempts.

Neither Iris recognition technology nor retina scanning are in such wide use as some of their counterparts because the equipment used needs to be very specific and dedicated and therefore potentially very cost restrictive. However, developments are moving forward for both methods, but as iris recognition is faster and less intrusive and developments are moving to enable it to be combined in the same equipment as used for facial recognition, it is thought that this will be the most successful of the two in the future.


Biometrics in Use Today
High Street banks and shops are starting to implement biometrics technologies into their businesses for verification of usage of debit and credit cards for example. The most commonly used is dynamic signature verification, which measures and assesses the characteristics of an individuals shape, speed, pressure, pen angle and sequence of signature. This method is oriented more towards verification rather than identification, and is particularly compatible with monetary transactions, documentation signing, patrolling and proof of delivery etc. It is very unobtrusive and socially acceptable as people are used to signing as confirmation of their identity, although as forgery is potentially possible it is probably not an effective option to use alone without an additional authentication method.

Hand geometry is also a nonintrusive easy to use biometric technique, and can easily be used for physical access control applications. The algorithms are designed to measure and analyse the dimensions of the hand, incorporating the shape and length of the fingers. The equipment to date tends has been quite bulky and expensive but that is now changing as potential new applications bring the devices into higher demand.

Again using the hands, a behavioural biometric, Keystroke dynamics, is a method of analysing the hypothesis that we all interact uniquely with a computer keyboard. However, it has to be said, that more accuracy is achieved with this method for touch typists or regular keyboard users, as others tend not to have developed peculiar or unique keyboard habits. This technology is dependent upon consistency of key depression, measuring variables such as, duration of key depression, time between strokes, error frequencies, and pressure of stroke. This is a particularly low cost biometric option and is an ideal software solution for PC access combined with a password.


Future developments – Watch this space
In many ways all biometrics are unusual, making them, hopefully unique and capable of identifying an individual through a particular human trait. There are however, some biometrics still in the development pipeline, which are intriguingly interesting and will be useful for very particular and specific applications rather than for the many applications that the more commonly known biometric technologies are suited to.


Gripping Stuff!
The University of Twente in the Netherlands is developing a new concept in Dynamic grip recognition. They have coined a new phrase, the TAPgun, and have taken up the challenge to devise a new method of protection for the police. The phrase TAPgun, stands for "Take-Away Protection" for police guns. The figures for 1999 demonstrated that in the United States 17% of police officers killed in the line of duty were shot with their own guns. It has been proposed that dynamic grip recognition could provide a unique effective solution, compared to the other suggestions of a mechanical lock, proximity token or fingerprint recognition, which in certain circumstances could work against the individuals they are meant to protect. Dynamic grip recognition will enable the gun to "recognise" its handler through a pre-programmed sensor built into the grip area of the gun. The individual characteristics of the authorised officer, including dynamic grip, hand geometry, pressure image and pressure dynamics will be incorporated in the memory of the sensor, disabling the gun from use if handled by another party.


Walking Tall
Often people can recognise others by the way they stand or walk, without having to see their face. Although in its preliminary stages, scientists are beginning to acknowledge that gait recognition may well become a feasible biometric for the future, as positive results are showing that one individuals gait is sufficiently different from another. Gait recognition could have many advantages over existing biometric methods, as the subject in question need not be aware of an attempted identification as it could be carried out at a distance and without the need for the individual to be facing the analytical equipment as with most visual recognition methods. Current research is concentrating on putting together the largest database specifically geared towards gait recognition, providing for further scope into image sequences with respect to the skeleton, the upper part of the body and how different clothes, footwear and carrying different objects effect our gait. There is obviously a long way to go, and much more research to be undertaken before anything concrete will materialise in terms of gait recognition, but it is interesting to see that there are still new biometrics methods materialising as the industry evolves.