Automatic identification of window regions on indoor point clouds using LiDAR and cameras

摘要: In this paper, we propose an algorithm to automatically identify window regions on exterior facing facades of buildings using interior 3D point cloud resulting from ambulatory backpack sensor system, outfitted with multiple LiDAR sensors and cameras. We develop a set discriminative features for the task, namely visual brightness, infrared opaqueness, occlusion indicator, within Markov Random Field (MRF) framework provide structured prediction or glass regions. A preprocessing classifier is trained produce node potentials, large margin parameter training used boost performance. Our has been data taken at 3rd floor Cory Hall UC Berkeley, total facade area 269.1 m2, tested walls 2nd Hall, Walgreens, office building in San Francisco, 454.6 m2. Window are successfully identified 85.5% F 1 -score 94.2% accuracy.