International Journal of Engineering and Technology

Biometrics refers to the recognition of individuals based on their physiological and behavioral traits. The ear is one such biometric cue because it is found to be unique even among twins. The proposed technique authenticates, a person using the ear as a biometric by exploiting its inherent geometric and structural details and is rotation and scale invariant. This paper aims at developing a modular, robust, biometric authentication system which utilizes an ear profile, by the use of Hough Transform. The Hough Transform maps the curved features of the ear onto straight lines .This makes it accurate in detecting connected components in outer ear images which have been captured under different environmental conditions like illumination or noise. Due to such variations curved edges are often only partially detected as broken edges. Hough Transform detects the contours of such outer ear images effectively. The curved edges of the outer ear are mapped onto straight lines called Hough Line. The length of the longest Hough Line, called max-line, is used for matching. Matching is done using Manhattan Distance between the max-lines of two images. This system has produced an accuracy of 98.26% when tested on the IIT Delhi database, which is greater than accuracy achieved by other edge based mapping techniques.