Face recognition using hierarchical neural networks

Face recognition technology has numerous commercial and law enforcement applications such as human machine interface, security, and surveillance. The techniques used in these applications include images processing, pattern recognition, and computer vision. This paper presents a hierarchical neural network based system for recognizing faces with different expressions. The hierarchical neural network used consists of modular subnets and each subnet consists of some nodes to represent different facial expressions. Two phases are used during the training of the network: locally unsupervised phase and globally supervised phase. Using different facial expressions as database, the hierarchical neural network achieves a recognition rate of 86.3% for faces outside the training set.