-Caveat Lector- from: http://www.newscientist.com/ns/19990619/newsstory9.html <A HREF="http://www.newscientist.com/ns/19990619/newsstory9.html">New Scientist: Tread carefully </A> ----- [Archive: 19 June 1999] ------------------------------------------------------------------------ Tread carefully Duncan Graham-Rowe SHOE PRINTS are the most common clue left at the scene of a crime, but matching crooks and their footwear is not an easy task. Forensic scientists have to painstakingly sort through photos of different types of shoes, boots and trainer prints. Now, however, computer scientists in Northern Ireland have turned to fractals to take the pain out of classifying footwear and to speed up crime detection. Although some police forces have access to shoe-print databases, these all rely on people to do the classifying. Inconsistencies are inevitable because the system is subjective, says Avian Alexander, who developed the fractal database at Queen's University in Belfast. The new system, however, is fully automated. According to Alexander, this means you need little training to use it. After the shoe-print impression is entered into a computer using a scanner or a digital camera, the system will search through hundreds of thousands of impressions and give you the nearest matches. It does this by breaking down each image into a list of fractal coefficients. Fractals are sets of complex geometric shapes that look the same over a wide range of scales. Alexander's mathematical technique makes it possible to describe a shoe tread according to its shape. In much the same way that complex fractals can be described by simple formulae, so a mathematical expression that describes a shoe-print can be produced from an image, making it easy to compare with others. The number of shoes coming on the market each year is vast. Some sports shoe companies such as Nike and Adidas churn out hundreds of types annually, while maintaining a stock of a few hundred at any one time. While some lines may be withdrawn, some people keep their footwear for years, making a large database all the more important, says Alexander. Not only is the fractal image database much faster at finding matches than a human, it can even find matches for partial prints or prints with faded treads. This can be a real problem with visual identification according to James Thorpe, a forensic scientist at Strathclyde University, in Glasgow. "As shoes are worn down, the individual characteristics change." The main advantage of Alexander's system is that it does away with the need to classify shoes into different categories according to the type of shoe, tread, p15heel and pattern. The subjective nature of this task, he says, makes it prone to error: after all one person's zigzag is another's squiggle. >From New Scientist, 19 June 1999 � Copyright New Scientist, RBI Limited 1999 ----- Aloha, He'Ping, Om, Shalom, Salaam. Em Hotep, Peace Be, Omnia Bona Bonis, All My Relations. Adieu, Adios, Aloha. Amen. Roads End Kris DECLARATION & DISCLAIMER ========== CTRL is a discussion and informational exchange list. Proselyzting propagandic screeds are not allowed. Substance�not soapboxing! These are sordid matters and 'conspiracy theory', with its many half-truths, misdirections and outright frauds is used politically by different groups with major and minor effects spread throughout the spectrum of time and thought. That being said, CTRL gives no endorsement to the validity of posts, and always suggests to readers; be wary of what you read. CTRL gives no credeence to Holocaust denial and nazi's need not apply. Let us please be civil and as always, Caveat Lector. ======================================================================== Archives Available at: http://home.ease.lsoft.com/archives/CTRL.html http:[EMAIL PROTECTED]/ ======================================================================== To subscribe to Conspiracy Theory Research List[CTRL] send email: SUBSCRIBE CTRL [to:] [EMAIL PROTECTED] To UNsubscribe to Conspiracy Theory Research List[CTRL] send email: SIGNOFF CTRL [to:] [EMAIL PROTECTED] Om
