BIOMETRIC IDENTIFICATION SYSTEM USING HUMAN IRIS PATTERNS

Abstract

Biometric systems will become ubiquitous in the near future, especially for security and transactions requiring identification of a person. A plethora of biometric technologies have emerged but iris recognition has proved to be the most reliable and accurate. Daugman patented this technology in 1994. Afterwards many other systems have appeared but we have observed that they are all variants of Daugman's algorithm. The iris recognition system consists of four modules- segmentation, normalization, feature encoding and matching. In segmentation, the annular iris region - which contains the unique patterns, is localized and noise is removed. We have proposed a novel method for the detection of iris/pupil and iris/sclera boundaries. In normalization the extracted iris region is converted into doubly dimensionless polar coordinates to account for imaging and physiological inconsistencies. The discrimination information for each iris is extracted by quantization of the phase data from the 1-D log Gabor filters. For matching hamming distance was employed. Two templates were found to be matched if a test of statistical independence failed. Our results prove that iris recognition is a reliable technology.