Face recognition for smart rooms

Abstract

In this thesis, automatic face recognition systems for smart room access control are presented. To find the appropriate approaches applied in smart rooms, different face detection and recognition algorithms are studied, and then their theoretical and applicable suitability to construct the automatic access control system is analyzed. At first, an accurate face detection/recognition system by Support Vector Machine (SVM) is proposed. In the following, an unsupervised real-time face recognition system by a combination of NFA and NED is initially presented, during which, skin color models, unsupervised clustering and face recognition techniques are utilized. Finally face recognition from multiple perspectives based on Gabor wavelets is analyzed, and experimental results are reported.