Learning parametric specular reflectance model by radial basis function network

For the shape from shading problem, it is known that most teal images usual ly contain specular components and are affected by unknown reflectivity. In this paper, these limitations are addressed and a new neural-based specula r reflectance model is proposed. The idea of this method is to optimize a p roper specular model by learning the parameters of a radial basis function network and to recover the object shape by the variational approach with th is resulting model. The obtained results are very encouraging and the perfo rmance is demonstrated by using the synthetic and real images in the case o f different specular effects and noisy environments.