Facial clustering model upon principal component analysis databases

Recently, the advancement of face recognition technology, which manifests itself in the variety of applications such as ATM machines, CC cameras, personal identification, etc., brings about a new-fashioned surveillance situation to distinguish and identify any person. This paper tries to improve the face recognition process by introducing a new model, Face Clustering, in which the face angles are clustered using AP-clustering approach so that memory space is saved and search accuracy as well as speed are boosted. Besides, a novel sensor device and program are proposed for measuring the real face angles and the face angles from face images, respectively. Hence, measuring the greater angles (>40°) using sensor device, which might not be achievable by the program, can become facilitated to be used in face recognition process. The angle values are then compared and the results demonstrate that Face Clustering model outperforms the PCA approach.