An Extended Moore-Neighbour Tracing Algorithm for Ear Segmentation in a Multi-Feature Ear Recognition System

Ear recognition is an example of a biometric system that uses human biological traits for recognition. This kind of recognition has been recently examined due to its distinctive properties such as its invariant shape. When performing analysis on image processing or pattern recognition, one of the major problems encountered is the number of features involved. It is necessary to extract a well-defined feature to make the classification process more efficient. Hence, this paper aims to propose an ear recognition system that extracts two features from the human ear: textural and geometrical features. This is aimed at improving the accuracy of the biometric trait. The extracted features were saved as a template and used for matching. The proposed system was evaluated with two online ear image datasets (AMI Ear database and USTB Ear database) and it produced an accuracy of 98.15.