A novel shape recovery approach in unknown environment based on wavelet neural network

A framework for the recovery of smooth surface shapes from shading images is presented. The method is based on using a single hidden layer wavelet neural network to approximate complex nonlinear reflectance map function and reformulate the shape from shading problem as the minimization of an error function over the network weight. Due to the properties of the wavelet neural network, the training problem can be transformed into a convex optimization process, therefore the global minimum can be obtained and the learning speed also increases. Under the proposed model, the lighting parameters are not necessary. Object surface is recovered with recursive algorithm. And hierarchical implementation is employed to reduce the computations. Both synthetic and real images are used to validate accuracy and efficiency of the algorithm, experimental results demonstrate the new algorithm is more robust and accurate than the traditional ones.