Project involving usage of state-of-art algorithms to first create a coarse reconstruction of indoor environments and then using that coarse mesh to create a planar mesh that can then be used for BIM. This was created using either depth images from depth sensors, or depth images created by using Surface from Motion and Multivision Stereo pipelines on sets of images.
We start with panoramic Point Clouds of Rooms
Register the different captures to get a full point cloud of the interior
We analyze the data to get a coarse estimation of room segmentation
Use the "Free Space" between the camera and the point to understand where the walls start
We use the previous data to make planar maps of the rooms
2018
