Automated deterministic model-based indoor scan planning

Scan planning describes the process of choosing equipment and locations for reality capture with laser scanners. By contrast to the traditional, expert-based method usually conducted in the field, automated approaches aim to solve this task exclusively with pre-existing data in the form of plans or 3D models of the scene. Existing approaches for automation are mostly either limited to 2D or based on simulations of laser scans, which oversimplifies respectively complicates the process to the degree that makes them inapplicable for practitioners. We aim to solve both problems by basing our solution on a 3D representation of the target scene and a deterministic approach. Thus, the workflow remains computationally feasible while the complexity of real-world scenes is sufficiently represented. We present a literature review on related research and technical guidelines for scan planning to define realistic requirements for scan planning, including point density, field of view, and depth of field limitations. To develop valuable strategies, we create a static set of candidate locations on a grid in the scene. We then perform visibility and coverage analysis and evaluate each candidate’s fitness for the overall strategy based on its contribution to our pre-defined scan requirements. Finally, selected locations are combined to form an optimized strategy to fulfill these requirements following two versions. We apply two basic methods for candidate selection and investigate their implications in a descriptive experiment.