A study of a reverse engineering system based on vision sensor for free-form surfaces

Reverse engineering can quickly create a CAD model of a new product, in which, the sensor, sampling planning and surface reconstruction are three crucial elements. In this paper, a reverse engineering system involving a new vision sensor, an improved sampling planning module and a fine surface reconstruction module is developed. A characteristic of the proposed sensor is strong linearity between output and input, obtained by the structure optimization when a simple lens replaces the asperic lens. Back propagation (BP) neural network error compensation heightens accuracy. To increase efficiency of digitization, an improved sampling planning approach is proposed; it is based on surface curvature and tangent line slope of a measured point. In surface reconstruction, a new adaptive extracting approach based on curvature of surface reconstructs the non-uniform rational B-spline (NURBS) surface for the scattered data. The accompanying reverse engineering experiment proves the proposed system to be reliable and efficient.