Door State Recognition Method for Wall Reconstruction from Scanned Scene in Point Clouds

Doors are important elements of building façades in scanned point clouds. Accurate door detection is a critical step reconstruction and indoor navigation. However, recent methods may often obtain incomplete information can only detect doors with single state (open or closed). To improve this, recognition method proposed based on corner straight-line fitting. Firstly, plane segmentation local features introduced to structural division from the raw data extract wall. Next, bounding box each calculated points, which then combined feature constraint classify Then, boundary extracted by normal vector, disordered discontinuous points fitted projection. Finally, obtained through analysis angle between straight-lines wall door. The effectiveness tested evaluated Livingroom ICL-NUIM House Room datasets. Furthermore, comparative experimental results indicate that our recognize different states effectively robustly scenes.