Ear Biometrics- An Alternative Biometric

This paper is one of the parts of a biometric based identity verification security system development project. Today, the most successful biometric based identification technologies such as fingerprint and iris scan are used worldwide in both criminal investigations and high security facilities. Even though Face recognition is one of the developing biometric methods; illumination, makeup, posing, emotional expressions and face-lifting reduce the success of face recognition. A new biometric which is not effected by any of the factors above is needed. The alternative biometric should overcome the drawbacks of face recognition. Twins are identical but their ears differ from each other, ear is also 3-dimensional but it is simpler than face and emotional expressions do not affect the ear. In the light of this, ear is a good alternative to face, as a biometric. In this study, the methods presented in the literature are tested on ear images. These methods are linear classification algorithms that work on 2D image databases. It is found out that, PCA, FLD, modified FLD which is also known as DCVA and LPP has better results at ear recognition than face recognition. Ear recognition has higher hit rates, when compared with face recognition researches that are presented in the literature previously. The results of this study proved that ear is the best alternative to face at personal identification tasks.