A Hybrid-reflectance-modeled and Neural-network-based Shape from Shading Algorithm

This paper introduces a shape from shading algorithm which adopts a hybrid reflectance model and neural networks. Different from traditional neural-network-based algorithms, this algorithm considers both of the diffuse and specular reflections; on the other hand, also different from traditional hybrid-reflectance-modeled algorithms, this algorithm can adjust the surface albedo and the combination ratio between the diffuse and specular components according to a training algorithm. With the hybrid reflectance model, the algorithm can deal with various kinds of surfaces; with the neural network, the consistency can be improved and the reconstruction time can be greatly reduced. Experimental results at the end of this paper also prove the superiorities.