Door recognition in cluttered building interiors using imagery and lidar data

Abstract

ABSTRACT:Building indoors reconstruction is an active research topic due to the importance of the wide range of applications to which they can be subjected, from architecture and furniture design, to movies and video games editing, or even crime scene investigation. Among the constructive elements defining the inside of a building, doors are important entities in applications like routing and navigation, and their automated recognition is advantageous e.g. in case of large multi-storey buildings with many office rooms. The inherent complexity of the automation of the recognition process is increased by the presence of clutter and occlusions, difficult to avoid in indoor scenes. In this work, we present a pipeline of techniques used for the reconstruction and interpretation of building interiors using information acquired in the form of point clouds and images. The methodology goes in depth with door detection and labelling as either opened, closed or furniture (false positive).

Bayesian graph-cut optimization for wall surfaces reconstruction in indoor environments

“…The selected walls include windows and doors, and our algorithm is evaluated in both cases without manual tuning, as opposed to other techniques in the literature which are particularly tuned to identify only one of them (e.g. such as [1,3]). The datasets were captured by a LiDAR range scanner and were processed on an Intel Core i7 2.8 GHz CPU computer, without any multithreading technique.…”

“…Moreover, the state of the reconstruction process of building interiors is less advanced than the more well-studied case of reconstructing the exterior outlines and façades of buildings [1,2]. This is because indoor environments often contain many objects and obstacles which prevent an ideal data acquisition and easy reconstruction of the structures of interest (see also Fig.…”

“…Authors in [10] present a graph-cut based reconstruction technique in order to reconstruct the wall surfaces from indoor scenes, without however identifying their openings. In [1] the authors combine information from 3D point clouds and images in order to reconstruct and classify the indoor scenes, without being able however to identify the windows on the walls.…”

“…Some of these rely on a priory knowledge obtained from training models and supervised learning algorithms in order to provide evidences for structural wall elements [2][3][4], while others combine indoor and outdoor scans in order to avoid heavy indoor occlusions [5], or they do not consider occlusion at all [6]. Furthermore, many techniques combine both 3D point and color image data from cameras [1] or stereo (depth) images [7] in order to classify the indoor objects and recover the architecture of the scene, while there are methods which try to keep in a minimum level the clutter and occlusion, scanning only indoor areas that have only essential interior objects and lack of commonly used furnitures [6].…”

Indoor Navigation From Point Clouds: 3d Modelling and Obstacle Detection

“…In addition, windows and doors could be confused with holes caused by the presence of other objects with similar size and shape such as cupboards, bookshelves, etc. In this case, candidates can be pruned by analysing the original 3D point cloud through histograms point-to-plane as in (Díaz-Vilariño, 2014).…”