Wednesday, June 16, 1999 Published at 18:20 GMT 19:20 UK
Criminals will be caught red-footed
New fingerprint technology has also helped police
The headache of trying to finger criminals by studying their footprints will be eased with the arrival of a new computer identification system.
Shoe prints are the most common clue found at crime scenes, but as footwear manufacturers design new styles, identifying them is increasingly difficult.
Forensic scientists now have to search through thousands of photographs of different types of shoe, boot and trainer to identify the perpetrator's foot.
But computer scientists at Queen's University in Belfast have developed a fractal image database, which finds matches much more quickly.
Although some police forces have access to shoe print databases, these all rely on manual classification.
Student Avian Alexander, who developed the database, said that with human-led databases, inconsistencies are inevitable because the system is subjective.
He told BBC News Online: "The problem of using image databases is that the majority use annotation where the user has to describe the shoe pattern in words.
"Shoes have to be categorised according to the type of shoe, tread, heel and pattern and what looks like a wavy line to one person may look like a zig-zag to another."
Mr Alexander said the new fractal system is fully automated and that detectives will need little training to use it.
The shoe print impression is entered into a computer using a scanner or digital camera, and the system will search through hundreds of thousands of impressions.
"The database works by breaking down each image into a list of fractal coefficients. The user isn't required to make any judgements," Mr Avian said.
Fractals are sets of complex geometric shapes that look the same over a wide range of scales.
Mr Alexander's mathematical technique makes it possible to describe a shoe tread according to its shape.
In the same way that complex fractals can be described by simple formulae, a mathematical expression that describes a shoe print can be produced from an image, making it easy to compare with others.
Identifying old shoes
Another advantage of Mr Alexander's system is that it can easily identify shoes with worn soles or get matches from partial imprints.
Mr Alexander got the idea for the system from his girlfriend, Glenda Smart, who has encountered the problems of identification in her work as a forensic scientist.
He said the database is still in the pilot stage, and will be developed further by other students.
The New Scientist has highlighted the results of Mr Alexander's work in its current issue.