A Survey of Methods for Detecting Indoor Navigation Elements from Point Clouds

With the popularity of large-scale public facilities, and increasing human indoor activities, people get an urgent demand for indoor refined navigation models. In recent years, 3D reconstruction technology such as 3D laser scanning, photogrammetry and computer vision grows fast. They can acquire high-precision data quickly and efficiently, and consequently provide rich data source for indoor refined navigation. However, the methods to extract indoor navigation elements available for indoor pathfinding like rooms, doors and windows, stairs, corridors have been one of difficult and attractive fields. For this purpose, aiming at the problems in indoor navigation, this paper summarizes and evaluates various algorithms and theories for indoor navigation elements extraction from point cloud, and proposes a new idea for indoor navigation elements extraction and navigation network generation from point cloud which combines geometric and statistical methods on the basis of summarized advantages and disadvantages, and thus offers the reference for the same trade or occupation.