By Mark Ward
Technology correspondent, BBC News website
Researchers in the UK are working on a computer system that aims to make shoeprints at crime scenes as useful as fingerprints and DNA.
The system could help the work of crime scene investigators
When finished, the automated system will search through records of the patterns on soles to identify the footwear used by a criminal.
It aims to increase the usefulness of shoeprints by using a common labelling systems for features found on soles.
The pilot system can already identify 85% of samples.
The realisation that criminals would find it almost impossible to avoid leaving shoeprints behind at a crime scene prompted a change in the law, said Professor Nigel Allinson of the University of Sheffield, who is leading the work on the identification system
The Serious Organised Crime Police Act 2005 now gives the prints the same legal status as DNA and fingerprint evidence.
"We all leave footprints, and though they are not as unique as DNA or fingerprints, they are good intelligence and can be good evidence," said Professor Allinson at a London conference on science and forensic work.
Home Office statistics show that 14.8% of crime scenes attended by crime scene investigators in 2004-2005 yielded shoe print evidence.
The Serious Organised Crime Police Act 2005 also gives police powers to take shoe prints from suspects. Before now, relatively few forces have routinely taken such prints.
Although some forces do take shoe prints and compare what they find at crime scenes with what is on file, much of the analysis comes down to human interpretation of the features found on soles, said Professor Allinson.
This is the reason why little of the data gathered by individual forces was consistently labelled.
By contrast, the system Professor Allinson and his colleagues are working on aims to automatically classify the features found on shoe soles, so crime scene evidence can be easily compared with records on file.
This classification should be straightforward, said Professor Allinson, because criminals favour shoes that have many distinct features on their soles.
"Luckily criminals wear trainers," he said. "If they all wore Oxford brogues we would be in a very difficult position."
The identification system works by subjecting shoeprints to several different image processing techniques to reveal the features on the sole.
The first stage cycles through levels of brightness in the image to see which features, such as logos, circles or ridges, persist at extremes of light and dark.
The captured features are then processed so they can be recognised at different rotations and scales.
Finally, these distinct features are analysed to see how they lie in relation to each other.
The prototype system can match prints taken in custody suites with those in the database 85% of the time, said Professor Allinson.
He emphasised that it has yet to be tried with partial shoeprints found at crime scenes.
Now work is being carried out to improve the system's detection rate and to make the whole system easy to use.
"It has to be a pretty much automatic system if its going to be successful," said Professor Allinson.
He presented information about his research at an event in London organised by the Engineering and Physical Sciences Research Council to highlight how science can aid forensic work.