K-means clustering-based approach for face recognition

In recent years, computer-based automatic face recognition technology has been widely concerned. At present, the majority of face recognition systems are using supervised machine learning techniques. These systems require a training step to learn a model for human face recognition task. This paper uses the face recognition method which is based on non-supervised learning to realize a non-supervised Face Recognition System. In this system, non-supervised k-means clustering algorithm replaces the Classification algorithm used in the previous face recognition system. Finally, the proposed method was tested by the standard test and the experiments show that under certain conditions the system has a high recognition rate. Copyright © 2010 Binary Information Press.