Researchers seek to improve identification of palm prints

Partial palm prints are said to constitute approximately 30 per cent of all prints recovered from crime scenes.

One objective of the three year Automatic Palm Print Identification project is to establish a small-scale simulated dataset that is comparable to the IDENT1 police database, which contains the prints of over two million individuals.

At present, palm prints are rarely used in criminal trials because they can’t be matched to the IDENT1 database, or the case closes before a match can be made.

Prof Ian Sillitoe, from Wolverhampton University’s Department of Engineering and technical collaborator on the project, told The Engineer: ‘When a palm or finger is put down, you’re left with a series of spurs and thin lines, which correspond to ridges and dips in the skin.’

In current technology, the methods used to analyse fingerprints are also applied to palms, which are approximately 40 times larger than fingerprints and contain a wider variety of features, such as creases.

Dr Raul Sutton, an expert in forensic science at Wolverhampton University and also a collaborator on the project, claimed this often proved problematic because criminals always leave partial and not whole prints and that these partial prints are often smudged, making identification more difficult.

‘The difficulty lies not just processing the images, but also the computational complexity of matching one partial print against a number of others,’ added Sillitoe.

‘We’re going to look at novel techniques for filtering and enhancing palm print images,’ he said. ‘We will scan a palm print onto a computer and process it so that the palm’s key features are recorded as a set of numbers, which will represent the different patterns on the palm.’

These numbers will then be matched against an equivalent set of numbers from the police database.