HOW many of you are like me. You have a feeling that you have seen that face before but you cannot put a name to it?

How pleased we would be if we could look up our "memory database" of faces and come up with an instant identification and name.

That is also what security agencies would like to do, but their job is vastly more complex with potentially millions of faces to scan.

Yet major progress is being made by a York-based company to solve the technically awesome problem of matching an unknown face with one from a large database of recorded images.

Cybula is a spin-out company, created in 2000 by Professor Jim Austin, of the Department of Computer Science, at the University of York.

The company was formed to exploit the university's AURA technology, which is a powerful set of tools for accessing complex datasets. It currently employs seven staff, has a turnover of around £500,000 and, crucially, is backed up by around 35 more staff in Prof Austin's research group.

Initially, the company was provided high-performance software and hardware for making searches on large unstructured text datasets, such as in address-matching or looking for matching molecules in the enormous drug databases of pharmaceutical companies.

More recently, Cybula's technology has been put to an even more challenging and important use in the rapid recognition of faces for security applications.

Cybula developed a system, called FaceEnforce, after AC Technology, a small high-tech company based in Washington, US, asked it to look into facial recognition systems.

Earlier approaches had used only two dimensional images, which give relatively poor results. Cybula's FaceEnforce uses 3-D images which yield far better performance but require immensely greater computing power.

Prof Austin's company not only provides some pretty amazing software, but also the very powerful computer hardware to allow the ultra-rapid processing that is necessary if the system is going to be of any practical use.

It would be less helpful to know that a recognised terrorist had gone through a check point half an hour ago than if identification could be established within seconds.

FaceEnforce is also faster than other biometric methods, such as iris or fingerprint recognition.

Cybula's 3-D approach is one of only three such commercial systems world-wide and it is now being evaluated by the US National Security Agency as part of the $18 billion homeland security programme.

FaceEnforce has already made it through the first two rounds of evaluation and the company should learn later in the year whether it has been successful in becoming the adopted system.

Prof Austin's pragmatic approach to the commercialisation of high technology is refreshing.

When asked why he extended his successful career as a leading academic researcher to the riskier field of entrepreneurial, commercial exploitation, he responds that "the rest of my family is already in business, so why not me".

He also appreciates the crucial need to have a key US partner who can take his ideas forward in the intended market.

AC Technology is ideal in this respect, as a young US company that "functions as a systems aggregator, anticipating and solving tomorrow's challenges with today's best of breed technologies".

Once again, we see a powerful application of York's university-based research with the potential for great importance to the international security industry and the capacity to grow another strong and successful business.

We will keep a close watch on this one and remember the next time you can't put a name to that face you half recognise - we know a man who can!

Updated: 10:37 Wednesday, October 12, 2005