CREU 2016-2017 Final Report: Towards an Ear Classification Scheme for Improved Biometric Recognition

Ears have been used to identify people since the early 1900s. The first example of ears being used to recognize humans was when Iannarelli, an American police officer, used ear measurements to keep track of prisoners. His work lead to the creation of the discipline called biometrics. Biometric systems can be used to confirm an individual’s identity by comparing a sample physical feature to a previously examined template. The process of developing a biometric system consists of present and analyzing a physical feature, extracting the most distinguishing parts of the presented feature, then lastly creating and storing a template of the feature. In our project we focused solely on collecting biometrics relating to the structures that comprise the ear. Ear biometrics is the measurement and analysis of the physical attributes of the ear to identify an individual. In order to improve ear recognition and identification in ear biometrics, the research conducted intends to determine the efficiency of an ear classification scheme in improving the biometric recognition of ears. Questions that we aimed to address in our study were: 1. Can ears be categorized to their structures such as the anti-helix, tragus, and lobule? 2. Can a Haar-based classifier be developed using OpenCV’s computer vision library that is able to categorize the anti-helix, tragus, and lobule from pictures? 3. Can this classifier be used on an existing open-source ear database to categorize the ears in the database? 4. Can the rank-1 recognition rate or accuracy of existing biometric algorithms be improved by through this categorization step? The research was conducted on the helix and lobule found on the outer structure of the ear.