Demand for high-volume 3D scanning of real objects is rapidly growing in a wide range of applications, including quality-control for manufacturing, online retailing, entertainment with virtual reality, as well as archaeological documentation and reconstruction. Fully realizing the potential of 3D acquisition requires scanning large numbers of objects with high quality and at reasonable cost. Although mature technologies exist for high-fidelity 3D model acquisition, deploying them at scale continues to require non-trivial manual labor.
This dissertation focuses on studying practical 3D acquisition for large numbers of objects. The problem is challenging, because it is hard to automatically find a proper set of scanner views that can not only completely cover the surface of multiple objects with different shapes, but also capture high fidelity surface model of the objects. Furthermore, it is non-trivial to position a 3D scanner at each of the desired views accurately and efficiently.
We propose a prototype system for multi-object 3D acquisition, which allows non-expert users to scan large numbers of physical objects within a reasonable amount of time, and with greater ease. Our system uses novel planning algorithms to control a structured-light scanner mountedon a calibrated motorized positioning system. We demonstrate the ability of our prototype to safely, robustly, and automatically acquire 3D models for large collections of small objects.
We propose an objective function for automated view and path planning, taking into account both accuracy and efficiency of the scanning system. We analyze different approaches to optimize for the objective, and discuss their performance and practicality.
In addition, we address the problem of surface inaccessibility to further refine our multi-object 3D acquisition system. We explore solutions for improvement from both the hardware and software ends.