TR2016-150

User-Guided Dimensional Analysis of Indoor Building Environments from Single Frames of RGB-D Sensors


    •  Xiao, Y., Feng, C., Taguchi, Y., Kamat, V.R., "User-Guided Dimensional Analysis of Indoor Building Environments from Single Frames of RGB-D Sensors", Journal of Computing in Civil Engineering, DOI: 10.1061/​(ASCE)CP.1943-5487.0000648#sthash.qf8wYMXI.dpuf, Vol. 31, No. 4, November 2016.
      BibTeX TR2016-150 PDF
      • @article{Xiao2016nov,
      • author = {Xiao, Yong and Feng, Chen and Taguchi, Yuichi and Kamat, Vineet R.},
      • title = {User-Guided Dimensional Analysis of Indoor Building Environments from Single Frames of RGB-D Sensors},
      • journal = {Journal of Computing in Civil Engineering},
      • year = 2016,
      • volume = 31,
      • number = 4,
      • month = nov,
      • doi = {10.1061/(ASCE)CP.1943-5487.0000648#sthash.qf8wYMXI.dpuf},
      • url = {https://www.merl.com/publications/TR2016-150}
      • }
  • Research Areas:

    Computer Vision, Robotics

Abstract:

In many construction, facility management, and inspection tasks, dimensional analysis of geometric features and artifacts is significant for spatial analysis and decision making. Tasks such as as-built geometry modeling and robotic workspace generation need to efficiently interpret critical dimensions of specific objects (e.g., diameter of a pipe, width of an opening) in a potentially cluttered environment based on data gathered from various positions. Thispaper presents a user-guided dimensional analysis approach to automatically acquire geometric information from a single frame of an RGB-D sensor. In the first step, an RGB-D sensor is used to capture three-dimensional (3D) point clouds of building environments. Then, by extracting planes and performing geometric analysis, the dimensional information of objects of interest is obtained from a single frame. The designed user guidance system evaluates the completeness of the acquired data, and then provides interactive guidance for moving the sensor to acquire complete data, from which stable and accurate geometric measurements can be obtained. The proposed method has been tested on hallways, door frames, and stairs in a building environment. The experimental results demonstrate that the method offers significant promise in enabling dimensional analysis in a wide variety of realtime measurement contexts.