CLASSIFICATION OF SURFACE CURVATURE FROM SHADING IMAGES USING NEURAL-NETWORK

This paper proposes a new approach to recover the sign of local surfac e curvature of object from three shading images using neural network. The RBF (Radial Basis Function) neural network is used to learn the ma pping of three image irradiances to the position on a sphere. Then, th e learned neural network maps the image irradiances at the neighbor pi xels of the test object taken from three illuminating directions of li ght sources onto the sphere images taken under the same illuminating c ondition. Using the property that basic six kinds of surface curvature has the different relative locations of the local five points mapped on the sphere, not only the Gaussian curvature but also the kind of cu rvature is directly recovered locally from the relation of the locatio ns on the mapped points on the sphere without knowing the values of su rface gradient for each point. Further, two step neural networks which combines the forward mapping and its inverse mapping one can be used to get the local confidence estimate for the obtained results. The ent ire approach is non-parametric, empirical in that no explicit assumpti ons are made about light source directions or surface reflectance. Res ults are demonstrated by the experiments for real images.