Face Recognition Based on Features Measurement Technique

摘要

We present a new technique to infer dimensions that can be used in biometric face recognition. The methodology is centered on inferring unique dimensions from human ears which provides unique physical biometric features. The process of determining the distance is done by harvesting the real actual dimensions from 2D faces images. This is achieved by using specific point to point distances on the two ears in human face. The points chosen give dimension information which enables discrimination for face recognition. The empirical results confirm that an accuracy of 94% recognition rate is achievable. The different positions of measurement points on the ears have a powerful impact to reduce the error of face recognition. Hence, our new measurement dimensions technique is precise and nodal facial points can be reflected as a robust face recognition method.