Polygonal Reconstruction of Building Interiors from Cluttered Pointclouds

Abstract

In this paper, we propose a framework for reconstructing a compact geometric model from point clouds of building interiors. Geometric reconstruction of indoor scenes is especially challenging due to clutter in the scene, such as furniture and cabinets. The clutter may (partially) hide the structural components of the interior. The proposed framework is able to cope with this clutter by using a hypothesizing and selection strategy, in which candidate faces are firstly generated by intersecting the extracted planar primitives. Secondly, an optimal subset of candidate faces is selected by optimizing a binary labeling problem. We formulate the selection problem as a continuous quadratic optimization problem, allowing us to incorporate a cost function specifically for indoor scenes. The obtained polygonal surface is not only 2-manifold but also oriented, meaning that the surface normals of each polygon are consistently oriented towards the interior. All adjacent and coplanar faces that were selected, are merged into a single face in order to obtain a final geometric model that is as compact as possible. This compact model of the room uses less memory and allows for faster processing when used in virtual reality applications. The method of L. Nan et al. was used as a starting point for our proposed framework. Finally, as opposed to other state-of-the-art interior modeling approaches, the only input that is required, is the point cloud itself. We do not rely on viewpoint information, nor do we assume constrained input environments with a 2.5D or, more restrictively, a Manhattan-world structure. To demonstrate the practical applicability of our proposed method, we performed various experiments on actual scan data of building interiors.

Normal Classification of 3d Occupancy Grids for Voxel-Based Indoor Reconstruction From Point Clouds

“…The cells of the resulting cell complex are then partitioned into building interior and outside space. Storey-wise 2D cell complexes (Li et al, 2018;Tran and Khoshelham, 2020) as well as fully threedimensional cell complexes can be used (Coudron et al, 2018;. Other reconstruction methods make use of trajectory information of the mobile mapping system if available (Cui et al, 2019;Nikoohemat et al, 2020;Lim and Doh, 2021) or operate in a discretized voxel grid (Fichtner et al, 2017;Flikweert et al, 2019;Gorte et al, 2019).…”

PC2WF: 3D Wireframe Reconstruction from Raw Point Clouds

“…Learning Structured Models. Many methods have been proposed to abstract point clouds into more structured 3D models, e.g., triangle meshes (Lin et al, 2004;Remondino, 2003;Lin et al, 2004), polygonal surfaces (Fang et al, 2018;Coudron et al, 2018), simple parametric surfaces (Schnabel et al, 2007;Li et al, 2019). From the user perspective, the ultimate goal would be to reverseengineer a CAD model (Gonzalez-Aguilera et al, 2012;Li et al, 2017b;Durupt et al, 2008).…”

“…Specifically, many man-made objects are (approximately) polyhedral and can be described by corners, straight edges and/or planar surfaces. Roughly speaking there are two ways to abstract a point cloud into a polyhedral model: either find the planar surfaces and intersect them to find the edges and corners, e.g., Schnabel et al (2007); Fang et al (2018); Coudron et al (2018); or directly find the salient corner and/or edges, e.g., Jung et al (2016); Hackel et al (2016).…”

Semantic Extraction of Permanent Structures for the Reconstruction of Building Interiors from Point Clouds

“…In Figure 14, a qualitative comparison is shown between the reconstruction results with and without prior semantic completion. For the geometric reconstruction of the room layout from the extracted planar primitives, we used a similar pipeline as proposed in [48]. As we can see the semantic completion is required to filter out the clutter as it gives rise to erroneous indentions.…”