of indoor structures often leads to incorrect modeling; 2) obstacles such
as chairs, desks, tables, flowerpots, and decorations represent occluded
areas; 3) the huge size of scanned data often leads to system slowdown or failure during the modeling process. For these reasons, fully
automated modeling is not a feasible solution at the present stage. The
viable alternative, rather, is improved productivity of manual operation.
This paper originated from our preceding study on productive highcomplexity 3D city modeling of building exteriors. This time, in the
present, follow-up study, we shifted our focus to a productive semiautomatic approach to as-built BIM creation for indoor environments.
The remainder of this paper is organized as follows. Section 2 briefly
summarizes as-built data acquisition trends, from conventional field
surveying to recent 3D laser scanning, and reviews the literature on
as-built BIM creation and its limitations. Based on the review of existing
approaches, we propose a semi-automatic approach via Subsection 3.1
(Overview), Subsection 3.2 (Automatic 3D geometric drawing) and
Subsection 3.3 (Manual modeling in BIM software). The performance
of the proposed approach is demonstrated in Subsections 4.1 and 4.2
with two experimental results. Finally, in the concluding Section 5, the
strengths and limitations of the proposed approach are discussed.