Automatic LiDAR-based lighting inventory in buildings

• Automatic lighting inventory using a LiDAR dataset.
• Use of RANSAC algorithm for ceiling detection.
• Point cloud rasterization using nearest neighbor algorithm.
• Application of image processing algorithms to detect the centroid of the lighting.

Construction industry is a large contributor in terms of energy consumption for all stages of the building life-cycle. Among building features, lightning management is a crucial element for energy saving. In this paper, an algorithm for the automatic detection of ceiling lightings is developed and tested. The main sections of the algorithm consist of ceiling extraction, point cloud to image conversion, and luminaires detection. Ceiling extraction is performed using RANSAC algorithm for plane detection. Point cloud conversion uses nearest neighbor rasterization and image binarization. The final step deals with luminaires detection and considers two types of lightning present in the dataset: fluorescent lightings are distinguished using a refined Harris corner detector while a Hough transformation is applied to find circular low energy bulbs.The algorithm results reflect a completeness of 100% with a geometric accuracy of 5.8 cm in the centroid determination of fluorescent lighting and 3.0 cm in low energy bulbs. The computing time ranges from 148.8 s in the detection of the fluorescent lighting to 105.9 s for the case of low energy bulbs, with point clouds of 90 and 60 million points, respectively.