Conclusions
In these pages, a semi-automated approach to productive as-built
BIMcreationhasbeenintroduced. Theapproachentailstwosteps:automated 3D geometric drawing, and manual creation of the as-built BIM.
In the 3D geometric drawing step, the plane components are extracted
by RANSAC segmentation and refined. Then, the boundaries of complex
indoor facilities are traced. The plane features and remaining points are
imported into commercialBIM software and usedasa reference to facilitate the manual construction of the as-built BIM.
The major contributions of this research to as-built BIM production
with point cloud data are as follows. First, the huge data size of
point clouds is drastically reduced without loss of information. The
experimental results showed that the data size was decreased to 3.8%
of the original point cloud for a corridor and to 4.3% for an atrium,
both at Yonsei University. This reduction enabled the modeler to efficiently manipulate the geometric drawing in BIM software without
systemslow-downor failure. Second, theproposed approach can be utilized in the creation of as-built BIM for huge and complex structures.
Most conventional approaches are used only limitedly under highly
favorable conditions (e.g. a simple rectangular room). The proposed approach, by contrast, can be used in real applications to complex indoor
environments such as the above-noted atrium. The extracted boundaries enable the modeler to easily identify facility structures and
model them. Additionally, the remaining points provide information
on un-segmented areas, making possible the modeling of complex
components.