A new edge detection scheme based on Radial Basis Function networks is proposed. The scheme operates on a two-tiered scheme where in the first stage each pixel in the input image is classified according to its potential for being part of an edge. The second stage then combines these pixels in to true edges in the input image. Both stages use radial basis function networks. The scheme illustrates how input space of edge patterns can be used to train the neural network with minimum parameters. Compared with other neural network paradigms, the proposed scheme is simpler in terms network size, computational requirements and provides better results even in low-contrast images. 0-7803-6278-0/00/$10.00 (C) 2000 IEEE

According to psychologists there are six types of universal facial expressions namely, "Fear", "Surprise", "Anger", "Sad", ''Disgust'' and "Happy". Holistic recognition of these facial expressions from static images requires nonlinear classifiers capable of operating on noisy highdimensional feature spaces. Often Radial Basis Function networks (RBFN) are used for classification in these applications. Conventional RBF networks however, in spite of their capabilities in working with high-dimensional feature spaces, often fail to deliver satisfactory performance in these scenarios due to small training sample sets, noisy features anUor features not following the required class smcture. This paper presents an improved RBFN architecture that overcomes these problems through asymmetrical scaling of feature axes according to specific requirements of the class structure of the classification problemThe scaling factors are computed automatically from the available mining samples, without any explicit analysis of their multi-variate statistical properties. The proposed network yielded an overall recognition rate of over 92% for the 6 expression classes, and a smaller network sue compared to other types of RBFN classifiers.