Moving face recognition through dynamic vector field neural networks
Zhizhai Hu, Ming Zhang · 2002
In the real world, human faces are non-rigid and moving. There are still no systems or models which can perform moving face recognition effectively. In this paper, moving face discrimination is treated as a dynamic system. A dynamic vector field neural network (DFN) model is developed for mapping alternations of the gray level face image with the binary vector of n/sup 2/ dimensions. Vector operators, such as translating T(t) and rotating R(t) operators, are used to recognise the moving face. Our experiment results show that DFN models can recognize moving faces with higher accuracy.