A neural-learning-based reflectance model for 3-D shape reconstruction

In this letter, the limitation of the conventional Lambertian reflectance m odel is addressed and a new neural-based reflectance model is proposed of w hich the physical parameters of the reflectivity under different lighting c onditions are interpreted by the neural network behavior of the nonlinear i nput-output mapping. The idea of this method is to optimize a proper reflec tance model by a neural learning algorithm and to recover the object surfac e by a simple shape-from-shading (SFS) variational method with this neural- based model. A unified computational scheme is proposed to yield the best S FS solution. This SPS technique has become more robust for most objects, ev en when the lighting conditions are uncertain.