A neural network approach for 3-D face shape reconstruction

In this paper, a novel approach for 3-D face shape recovery based on neural network is presented. A learning vector quantization network architecture based on varying parameters and eliminating is developed that learns the correction of gender patterns and recognizes facial expressions of human faces. To achieve robustness in viewing, the network is trained with a wide range of illumination and conditions. A method of merging recovered 3-D surface regions by minimizing the sum squared error. Hence we measure the average absolute percentage error per pixel (AAPEPP) for each recovered face part. The new algorithms for data driven, stable, update the surface slope and height maps are proposed. This approach significantly reduces the residual errors. Experimental results illustrate the good performance of our approach.