Logo

Faced with shortcomings in the existing biometric systems in identifying criminals, researchers at IIT-Kanpur have developed a new system which can record multiple traits of people, giving an almost 100 pc matching score.

Experts claim that this system will help the sleuths nab criminals with accuracy and speed and is likely to overcome the flaws of the current biometric system being used by intelligence and security agencies in the country.

The new system can simultaneously integrate five traits -- face, fingerprint, iris (eye), ear and signature -- and the IIT-K has already tested the system on a sample of 1831 individuals.

The results have shown a success rate of 99.24 percent, an expert at the IIT-K's Computer Science and Engineering Department said.

"The multi-modal biometric system uses multiple traits to capture different types of biometrics and fuses them in order to meet stringent performance requirements," he said.

The project sponsored by the Union Ministry of Communication and Information Technology is tipped to be used soon by the Centre for creating a database of criminals, the official said.

"A group of 20 officials from police, CBI, Forensic department and National Crime Records Bureau are undergoing training for operating the system," he said.

The system is likely to reduce the time for matching criminal records manifold as it uses various traits for identity verification.

"Uni-modal biometrics system suffers from an overhead of large number of comparisons with other existing records in the database. As database size increases, the data retrieval and search time increases," the official said.

The multi-modal system is able to meet strict performance requirements imposed by high security applications such as accessing a nuclear facility, he said.

A Multi-modal biometric verification system can be considered as a classical information source that combines evidence provided by different biometrics to improve the overall decision-accuracy.

In Fusion at Matching Score Level, feature images are created independently for each sensor and are then compared to the enrollment templates which are stored for each biometric trait.

Based on the proximity of feature image, each sub-system computes its own matching score.

These individual scores are finally combined into a total score, which is passed to the decision module.

"The system is a coupled architecture where each sub-system performs like a single biometric system," he said.

The system takes advantage of the proficiency of each individual biometric and deters spoofing since it would be difficult for an impostor to spoof multiple biometric traits.