This paper describes an active vision system based reverse engineering approach to extract the 3D geometric information from dental cast/teeth and transfer this information into computer-aided design/computer-aided manufacture (CAM) systems to improve the accuracy of the constructed units for patient care. The benefits of our approach are a non-contact measurement, low cost, and capturing all the surface data at the same time. Our system involves the development of a vision rig prototype accommodating up to three charged couple device cameras, image processing, and shape recovery from a series of convex profile (2D) images of complex objects morphology (e.g. incisor or sunflower seed). The rig is designed using engineering design methods such as a concept selection matrix and a weighted objectives evaluation chart. The rig concept design is based on the human/stereo vision principles in computer vision. The occluding boundary technique and its derived method, the slicing method, are described and employed by the vision system to generate (3D) models of two typical teeth using a dental cast. Comparison of the vision system and the Renishaw Triclone dental system is demonstrated on these two models with one of them being manufactured using CAM technology. The results have shown that our vision system is capable of supporting the advanced reverse design applications in restorative dentistry. It is concluded that the new system could encode the shape of small complex objects and the device could be useful in the characterization of natural materials as a basis for biology-based design.