Research on the reconstruction method of B-spline surface based on radius basis function neural networks

Surface reconstruction is the key technology in the geometry reverse engineering. In order to obtain the object´s geometrical model, we have to construct surface by large numbers of measured data points. This paper introduces a new method for the reconstruction of free-form surface. Firstly, it profits scattered and measured data points which come from free-form surface archetype by radius basis function neural networks algorithm. Secondly, it maps the mathematical model of free-form surface by the linear combination of radius basis function and the weights of the hidden layer. Finally it transforms the mathematical model to bicubic B-spline surface. This paper also commentates the feasibility of the above idea that resolves the problems of surface fitting by radius basis function neural networks