Ahmed

University of Calgary, Department of Geomatics Engineering

Submission name:

Ahmed et al.

Short description of the method:

Semantic indoor geometric modelling approach (SIGMA) is designed for modelling complex interior structures. The system consists of two modules, namely, the modeling module and the editing module. The first module is based on a five-step process, namely, preprocessing, 3D segmentation, layout reconstruction, wall-surface object modeling, and ceiling reconstruction. Downsampling and outlier removal are both used in the preprocessing step, while a hybrid approach of region-based and model-based segmentation is applied in the segmentation step. The layout reconstruction step is related to determining the main structural elements of the interior scene (wall, floor, and ceiling). Empty regions are detected and wall surface objects (doors and windows) are both modeled for each wall segment using object-based energy function. Ceilings are reconstructed from the intersection between floor surfaces and wall boundaries for each room, and floor level is defined as well.

Reference:

Shi, W., Ahmed, W., Li, N., Fan, W., Xiang, H., Wang, M., 2019. Semantic Geometric Modelling of Unstructured Indoor Point Cloud. ISPRS International Journal of Geo-Information, 8(1), 9.