Telling Crooks From Saints

Machines help us to check 
frauds and crimes, reports Biplab Das 

A machine's hawk eyes pin down a six-feet tall guy moving suspiciously in the airport lounge. It reads the facial lines, compares with thousand other faces and recognises the hardened criminal at large. As soon as the man signs his name on the boarding pass, the automatic handwriting expert supplies the clinching evidence that helps police nab the elusive serial killer. 

International experts of pattern recognition conjured up such a futuristic security zone at a recently-held conference by the Electronics and Communication Sciences Unit of Indian Statistical Institute. "Artificial intelligence provides the framework of pattern recognition," said Dr. Dwijesh Dutta Majumder, director of Institute of Cybernatics Systems and Information Technology. "Recent progress in computer science has enriched artificial intelligence, reducing the gulf between man and machine." According to him, we now have sophisticated computers that are very good at solving non-numerical problems, which make the machines indispensable in our day-to-day life. 

Nowadays, computers are being used to diagnose diseases, defend national boundaries, explore high energy physics, and peer into space. It has played a huge role in radio communication. "Computers read the minute details of patterns in the data beamed back by remote sensing satellites and spacecraft, and convert them into perceptible messages," said Dr. A. P. Mitra of National Physical Laboratory. 

However, the tremendous progress in machine intelligence has failed to check the rising cases of frauds and crimes. "To counter these, biometrics helps identify the patterns of an individual's physical properties such as fingerprint, iris, vein structure, and ear shape for authenticated access to any services," said Prof. Anil K. Jain from the department of computer science and engineering, Michigan State University in the US, who spoke on 'Biometrics Authentication'. 

"These days we have to use so many passwords, which can be easily forgotten, copied or misused," Jain said. "But an individual's information stored through biometrics, won't be lost." Sensors embedded in laptop computers, mobile phone, and desktops can distinguish fingerprints of genuine and fake users. Current biometric systems also use the characters of retina, face, signature, voice, palmprint, and gait to establish a person's identity, Jain added. 

Handwriting bears an unmistakable signs of any individual. Dr. Sargur N. Srihari from the department of computer science and engineering of State University of New York, Buffalo, US, spoke on the 'Quantitative Assessment of Handwriting Individuality.' 

Features like spacing between words, word shape, relative height between letters, legibility, and pictorial quality are used to develop a software system for visual handwriting matching," said Srihari. "Such a system is of great help to single out crimes involving secret documents."