Adaptive Agent System for Ear Recognition Using Mathematical Model Approach

Abstract

Person’s identification or recognition is a first problem step in all security
systems. In these systems, Ear biometric was suggested to be a crucial tool for
recognizing people in critical, important organizations and government
security agencies. It is unique to each individual and relatively unchanging
during lifetime. The proposed system in this study presents a new algorithm
for ear recognition. It offers a mathematical model based on centroied and
related cluster analysis which tries to find a reciprocal twin sample to the
nearest neighbors and then they are grouped as a cluster. System is trained on
individual subjects with the same conditions. Each enrolled image subject is
converted to a gray scale. Experimental results reflect that the proposed
system can be offered 93% accuracy, less complexity and can make more
accuracy with subjects who are those have a short hair.