The big bang-up

Updated Tuesday, 1st August 2006
Dr Michael Hobson reveals how Crimewatch led his research in a new direction - using astronomy techniques to improve CCTV

This page was published 15 years and 5 months ago. Please be aware that due to the passage of time, the information provided on this page may be out of date or otherwise inaccurate, and any views or opinions expressed may no longer be relevant. Some technical elements such as audio-visual and interactive media may no longer work. For more detail, see our Archive and Deletion Policy.

Have you ever wondered how astronomers produce those wonderful pictures that appear in the papers or on television from time to time?

They certainly look very impressive, but in many cases it has taken a lot of work to turn the signal received by the telescope into a clear image of a star cluster or distant galaxy.

The basic images produced by telescopes often bear only a passing resemblance to the final pictures presented in the media or scientific journals.

These 'raw data' (as they are called) usually suffer from 'blurring' and 'noise' caused by the inevitable imperfections inherent in any measurement process.

Therefore, an important part of modern astronomy is to develop ways of processing the raw data to remove these unwanted artefacts, and produce a clear picture, worthy of a slot on the Ten O'Clock News.

As an astronomer working in the Cavendish Astrophysics Group in Cambridge for the last 10 years, I have been very closely involved in developing a wide variety of 'image reconstruction' techniques.

These are basically mathematical algorithms that run on a (usually very large) computer. The program reads in the blurred, noisy data and attempts to reconstruct the actual image the telescope observed (which we generally call 'the truth').

It is likely that I would still be working solely in the analysis of astronomical images if I hadn't sat down one evening with a cup of tea and happened across an episode of Crimewatch on television. In one particular item, a man had walked into a Post Office, threatened the cashier and made off with a considerable sum of money.

Fortunately, the man has been caught on the CCTV camera. Unfortunately, as is often the case, the picture was terrible. It was so blurred that it was impossible to make out the man's features.

Suddenly, I realised that the techniques I had developed for reconstructing astronomical images might work equally well at enhancing pictures like those taken by the CCTV camera.

Guess Who? game [Image: bslavinator under CC-BY-NC licence]
Guess Who? game [Image: bslavinator under CC-BY-NC licence]

Astronomy is a subject that holds a general interest for the public, but is often criticised for having no practical applications. In fact, this is far from true, with numerous modern technologies owing their genesis to 'blue skies' research of a most fundamental nature.

Nevertheless, I was very excited by the idea of making seemingly esoteric research provide real practical benefits in the everyday world.

At that point, I decided to pursue the notion of creating a general image enhancement method built on the ideas of a generation of astronomers around the world.

I began working on the problem the very next day. Despite having considerable existing research and teaching committments, I managed to to find a couple of hours a day to work on adapting my astronomical software to enhance everyday images.

To make the process more entertaining, I downloaded a picture of Marilyn Monroe's face to use as my test image! I first put the image through a distortion algorithm to simulate the effect of a cheap CCTV camera such as those used in many shops. Then I attempted to recover her features.

As is often the case in research, my first few attempts were terrible and resulted in images that resembled Quasimodo more than Marilyn.

The problem of reconstructing an everyday image such as a face presented me with a multitude of new challenges that do not occur in astronomical images.

On the other hand, some aspects of everyday images are easier to deal with than telescope data. After a few weeks of producing rather monstrous visages, the reconstructions started to take shape and Marilyn's features began to emerge on my computer screen.

This improvement continued until, about two months after seeing Crimewatch, I had a mathematical algorithm that would have been able to put Marilyn Monroe in the dock if she had ever robbed a Post Office!

At this point, however, the teaching term began in Cambridge and I had no time to pursue these ideas further. Fortunately, it was also around this time that I was made aware of the Fellowship programme offered by the National Endowment for Science, Technology and the Arts (NESTA for short).

NESTA is a Government agency set up using National Lottery money, with the aim of promoting innovation in a wide variety of areas. In particular, NESTA aims to encourage interaction between different areas in the Arts and Sciences.

Given the interdisciplinary nature of the work I had begun, I decided to apply for a Fellowship and was fortunate enough to be awarded one.

The Fellowship runs for 3 years and has provided me with sufficient funds to take on an excellent young researcher and programmer, called Charlie McLachlan, and to buy the necessary computing equipment to develop the image enhancement algorithms further.

Having just completed the first year of my NESTA Fellowship, and with Charlie's help, the algorithms have improved enormously. We have also begun a series of collaborations with others in the field of image enhancement, in order to share ideas and take advantage of the latest developments around the world.

One of the most exciting aspects of our work has been the wide range of problems we are now able to tackle.

Aside from the original application in CCTV images, we have found that the same basic principles can be applied to the enhancement of images such as those taken by medical scanners, police cameras, satellite photographs and many others.

There are even applications to the speeding up of computer generated animation and the real-time clean up of video footage.

Although the development work still continues, we now have a range of techniques that we feel should prove useful in many areas. Indeed, we have recently begun the process of meeting with representatives of police forces, health authorities and other interested parties, with a view to supplying them with our new technology.

So, next time you visit the Post Office, look up into the CCTV camera and smile. At least now we should be able to tell it's you!




Ratings & Comments

Share this free course

Copyright information

For further information, take a look at our frequently asked questions which may give you the support you need.

Have a question?