Human Recognition using Ear Biometrics By Computer Vision Approach

Abstract

The introduction of a class of ear-based biometrics for use in the creation of passive identifying systems. The theoretical uniqueness and long-term measurability of the ear are used to demonstrate the feasibility of the proposed biometric, and the practical applicability of a computer vision-based system is demonstrated. The ear of each subject is represented as an adjacency graph constructed from the curve segments of its diagram. We provide a unique authentication approach based on graph matching that accounts for incorrect curve segments that may appear as a result of changes (lighting, shadowing, and occlusion) in the ear picture. Because the traits in this class of biometrics are reliable and can be reliably derived from a distance, they are excellent for passive identification.