Automatic and complete optical 3D digitization strategy with adaptive global view planning.

This paper presents a practical automatic 3D scanning pipeline aimed at achieving comprehensive 3D digitization of complex unknown objects in real-world settings. Addressing the challenge of selecting minimal yet effective viewpoints for complete surface coverage while reducing redundancy, we introduce an adaptive global viewpoint planning algorithm that models a viewpoint energy field based on spatial and placement constraints, and applies a field-guided optimization strategy to maximize coverage efficiency. We built an autonomous scanning system that integrates multi-resolution optical sensors with a robot and motor to support efficient and high-fidelity 3D reconstruction. To achieve high accuracy, a low-cost and high-precision two-step calibration method is proposed for the entire sensor-actuator setup. Experimental results on public and real-world models demonstrate the system's capability to achieve accurate, complete, and fully automatic 3D reconstruction, highlighting its practical potential for optical metrology and real-scene digitization applications.