3D object reconstruction from multiple controlled viewpoints

To reconstruct the complete model of a 3D target by robot vision, multiple viewpoints often need to be planned to obtain sufficient 3D surfaces for integration. This paper presents a novel method of viewpoint planning for incrementally building the models of unknown objects or environments by an active vision system. This method is based on the concept of trend surface, which is the regional feature of a surface for describing the global tendency of change. A new mathematical model is proposed for predicting the unknown area of the object surface. A criterion is defined to determine the exploration direction and a unique surface model is established by analyzing the surface curvature. Then an algorithm is developed for determining the sensor pose which satisfies the many placement constraints such as resolution, focus, and field of view. Experiments are carried out to demonstrate the proposed method.