Facial Recognition Technology Has Improved

Facial recognition technology has improved drastically in recent years, and is becoming a viable method for augmenting traditional security measures. It has been implemented and tested (although with varying degrees of success) by a variety of government agencies and private institutions in an effort to provide more reliable security and identity confirmation, as it is not as prone to fraud like traditional, non-biometric identification. Facial recognition is a type of biometric that relies on recording "the spatial geometry of distinguishing features of the face" (John D. Woodward, Horn, Gatune, & Thomas, 2003, p. 3). In terms of security, biometric authentication affords the strongest method of identity verification, since ID cards and/or passwords can be transferable between persons, or forged by malicious third parties.

Facial recognition relies on the capturing of facial images, which are then referenced and compared against an existing database of images. It is in this regard that facial recognition has an advantage, since it can effectively utilize legacy image databases, and does not need, for the most part, a new infrastructure. This type of biometric security can be helpful in supplementing traditional surveillance measures, as it can reliably and quickly locate criminals or terrorists with more accuracy and speed than even the most highly trained security personnel. The advantage of facial recognition technology over other forms of biometric authentication are that facial recognition surveillance relies on public images, the capturing of which is a non-intrusive and contact-free process, and integrates with existing camera surveillance equipment (John D. Woodward et al., 2003, p. 7).

II. Past Uses

Biometric-based surveillance relying on facial recognition technology has been employed in the past, with varying degrees of success. In 2002, the NIST conducted a large-scale test of the accuracy of facial recognition technology at the U.S. Naval facility in Dahlgren, VA. The test was primary used to gauge the reliability of facial recognition technology for access control, with successful results. In an NIST brief, it was noted that the probability of correct verification with a sample of 33,000 individuals was roughly 89%, with a false alarm rate of 1%. The difficultly in interpreting this result to support the viability of wide-scale surveillance, however, arises from the controlled nature of the test. When facial recognition technology was used to test the viability of biometric access control with no ambient light control, such as outdoor conditions, the accuracy for facial recognition is only 47% (Summary of NIST Standards for Biometric Accuracy, Tamper Resistance, and Interoperability, 2002, pp. 15-16).

Another recent field test assessing the viability of facial recognition technology was conducted by the German government, to determine the accuracy for photo identification at points of entry. This test was started in April 2003, and ended in July 2003 at the BKA (Federal Criminal Police Office) in Wiesbaden, Germany. The results were similar to the NIST test, as it was concluded that facial recognition provides a high degree of accuracy only when environmental factors are controlled. They also found that there has been no adequate study on the effects of ageing and facial recognition. Despite these concerns, the report did acknowledge that biometric facial scanning was suitable for use when verifying photo documents, given that the preconditions, such as controlled ambient lighting, are met (cf. An investigation into the performance of facial recognition systems relative to their planned use in photo identification documents - BioP I, 2004).

III. Viability for Use in Large-Scale Venues

The success of facial recognition technology for wide-scale security augmentation is hindered by the inherent limitations of current technology, as well as environmental and physiological factors. Given that venues for major sporting events traditionally have security measures already in place, such as video surveillance and gated access, facial recognition technology might be useful in augmenting other, more traditional forms of security. For facial recognition to be successfully implemented, it is necessary to capture facial images when visitors go through gated areas, and then compare that image against government databases. There are, however, severe limitations with this type of electronic surveillance, since for accurate scanning, visitors must look directly into a camera, not be wearing any articles of clothing that might obstruct facial features (such as hats or glasses), and ambient lighting must be controlled. Additionally, facial recognition technology can only provide warning for people who have photographic records in government databases; as such, it is entirely ineffective against new, unknown threats.