LEARNING IDENTITY WITH RADIAL BASIS FUNCTION NETWORKS

Radial basis function (RBF) networks are compared with other neural ne twork techniques on a face recognition task for applications involving identification of individuals using low-resolution video information. The RBF networks are shown to exhibit useful shift, scale and pose (y -axis head rotation) invariance after training when the input represen tation is made to mimic the receptive held functions found in early st ages of the human vision system. In particular, representations based on difference of Gaussian (DoG) filtering and Gabor wavelet analysis a re compared. Extensions of the techniques to the case of image sequenc e analysis are described and a time delay (TD) RBF network is used for recognising simple movement-based gestures. Finally, we discuss how t hese techniques can be used in real-life applications that require rec ognition of faces and gestures using low-resolution video images. (C) 1998 Elsevier Science B.V. All rights reserved.