Ceiling Equipment Extraction from TLS Point Clouds for Reflected Ceiling Plan Creation

Abstract

The reflected ceiling plan (RCP) is a two-dimensional drawing of facilities with ceiling equipment, such as lighting, fire alarms, sprinklers, and inspection holes. The RCP is often created from existing facilities for safety standard verification, renovation, and inspection. However, the creation of RCPs of large-scale facilities requires significant time and effort. In this study, a method for extracting ceiling equipment information from point clouds acquired using a terrestrial laser scanner (TLS) was developed for RCP creation. The proposed method is based on footprint detection for ceiling equipment and involves three steps. First, circular and quadrilateral footprints of the ceiling equipment from point clouds of each scan are detected. Next, the footprints are merged for multiple scans and clustered using their dimensions and point distributions. Finally, the labels of pieces of equipment are interactively assigned to each cluster. The performance of the proposed method was evaluated for four facilities using TLS point clouds. The experimental results showed that the detection rates of footprints (recall) exceeded 90% within a scan distance of 6 m, and the labeling accuracy was also more than 90%. For 79 scans (point clouds) of a facility, the time for extracting 80% of equipment information for RCP creation was approximately 25 min, which corresponds to 2% of the manual RCP creation time of the facility. This demonstrates that the proposed method achieves efficient RCP creation.