1Department of Master of Computer Applications, Siddaganga Institute of Technology, Tumakuru, Karnataka, India
*Corresponding author e-mail id: jayaram_mca@sit.ac.in
In this article, we present the development of a personal identification system which is based on a unique methodology of recognition of geometry of the human ear. The case in point is the application of convex hulls to capture the shape of the ear in an optimal way with utmost precision. Around 300 images of right ear which were captured in vivo formed the database for the development of the system. All the 300 images were precisely fitted into convex hulls and 13 features namely area, aspect ratio, baricentric coordinate, convexity, concavity, eccentricity, circular equidiameter, Euler number, Ferret's diameter, farm factor, branch angle, perimeter and solidity were elicited from the images. The system so developed is evaluated for its generality and consistency. We found a very high recognition accuracy of 98%. The system performance measures are found to be in corroboration with international standards.
Person identification system, Ear biometrics, Convex hulls, Recognition accuracy, Performance measures
