Multimodal Biometrics -- Is Two Better Than One?

Multimodal Biometrics

Multimodal Biometrics -- Is Two Better Than One?
September 2006

By Imran Khan, Research Associate, Auto ID & Security

Introduction

Biometrics has been adopted in a variety of large-scale identification application -- ranging from border control to voter ID issuance. While the technology is conceptually adept, in reality there are numerous challenges associated with enrolling large populations using just single (unimodal) biometrics. These challenges can be overcome by deploying multimodal biometrics systems.

The Problems with Unimodality

The shortcoming of unimodal biometrics is that no one technology is suitable for all applications. Therefore, the presence of a multimodal biometric system helps compensate for the following limitations:

The usage of certain biometrics makes it susceptible to noisy or bad data, such as inability of a scanner to read dirty fingerprints clearly. This can lead to inaccurate matching, as bad data may lead to a false rejection.
Unimodal biometrics is also prone to inter-class similarities within large population groups. In case of identical twins, a facial recognition camera may not be able to distinguish between the two.
Some biometric technologies are incompatible with a certain subset of the population. Elderly people and young children may have difficulty enrolling in a fingerprinting system, due to their faded prints or underdeveloped fingerprint ridges
Finally, unimodal biometrics are vulnerable to spoofing, where the data can be imitated or forged.

Definition of Multimodality

Multimodality is the usage of more than one physiological or behavioral characteristic to identify an individual. It involves the fusion of two or more technologies such as fingerprint, facial recognition, iris scanning, hand geometry, signature verification, or speech recognition.

The fusion is done by running the two (or more) biometric inputs against two (or more) different algorithms, to arrive at a decision. This technique is useful in large-scale civil ID applications, where the identity of thousands of people need to be authenticated at a time. To have an additional method of verification as a backup reduces the possibility of inconveniences caused by the malfunctioning of the primary biometric.

Advantages of Multimodality

It is estimated that approximately 5 percent of any population has unreadable fingerprints, either due to scars or aging or illegible prints. In a civil ID scenario, where millions of people have to be enrolled in the system, the segment of the population who are un-enrolable will face inconveniences. Having multimodal biometric technology can overcome this restriction and ensure lower failure to enroll rate (FTE).

Multimodality can also address the problem of aversion to fingerprinting, found in certain parts of the world. Sometimes people associate fingerprints with criminal activity, and are reluctant to submit prints. By having an additional biometric available, a greater number of people can be enrolled into the system

Using multiple biometrics solves the problem of inter-class similarity and the resultant high false acceptance rate (FAR). If people with similar hand sizes or similar looking facial features can gain false acceptance, the presence of another biometric such as signature verification can distinguish between the samples.

Another advantage of using multimodality is that it solves the problem of data distortion. If the quality of one of the biometric samples is unacceptable, the other can make up for it. If a fingerprint has been scarred and the scanner rejects the distorted sample, having another modality like facial recognition can prevent high false rejection rates (FRR).

Unimodal Biometrics can be easily spoofed. Placing a high-resolution picture of a fingerprint under the scanner can deceive some systems. However, by using multiple biometrics, even if one modality could be spoofed, the person would still have to be authenticated using the other biometric. Besides, the effort required for forging two or more biometrics is a deterrent to those who wish to do so. Diagram

Disadvantages of Multimodality

Some of the disadvantages of multimodality include:

Interoperability and Standardization: The technology is at its early stages, and there are no universal standards set for storing templates and having all biometrics technologies seamlessly work together.
Cost: The addition of another biometric technology can drive up the price of the solution. Critics of multimodal systems say it does not offer real value, and that it is just a marketing gimmick geared toward increasing sales.
Reduced Matching Level: Calculations provided by John Daugman, the originator of the iris algorithm, claim that if a stronger biometric is used in conjunction with a weaker biometric, the result is not necessarily a stronger combined system. The error rate (FAR or FRR) of the weaker biometric can bring down the overall effectiveness of the system.

Major biometrics companies have released products that combine multiple biometrics, mainly for large-scale civil ID usage. Motorola's biometrics unit offers a 'multi-biometric' enrollment and verification solutions with support for fingerprinting, 2D/3D facial recognition, and signature verification. ImageWare Systems also has a middleware product called 'Biometric Engine' that can capture finger, face, and iris data to be used for passport and national ID issuance. Viisage was selected to implement an integrated finger and face biometric system for Iceland's e-passport program. Rather than relying on paper-based images, the facial recognition technology will be used to store a photograph in the passport chip. This will offer higher quality and accuracy levels. The state of Qatar in May 2006, announced the rollout of a national identification project, which will store fingerprint, face, and iris biometric data on a smart card. Similar integration of multiple biometrics with smart card-enabled ID documents is being implemented around the world.

Future Trends and Conclusion

So far, the prominent biometrics combinations have been fingerprint, facial, and iris recognition. It is possible that the patent issue surrounding hand geometry has made it less accessible for union with multimodal systems. An interesting trend has been the acquisition of companies focusing on a particular biometric technology by companies that make other biometrics. Identity solutions company Viisage, in January 2006, entered into an agreement to merge with Identix, who makes fingerprint scanners. Viisage also acquired the iris recognition company Iridian Technologies, thus becoming competent in facial, fingerprint, and iris biometrics. Similarly, Cross Match Technologies bought out the facial recognition technology company C-VIS, in order to provide multi-modal biometric offerings to their customers. It will be interesting to see if offering a multimodal product triggers the next wave of development within the market, or if it is just a short-term marketing gimmick.

back