Abstract
Abstract. We focus on a region-based point clustering to extract a polygon from a massive point cloud. In the region-based clustering, RANSAC is a suitable approach for estimating surfaces. However, local workspace selection is required to improve a performance in a surface estimation from a massive point cloud. Moreover, the conventional RANSAC is hard to determine whether a point lies inside or outside a surface. In this paper, we propose a method for panoramic rendering-based polygon extraction from indoor mobile LiDAR data. Our aim was to improve region-based point cloud clustering in modeling after point cloud registration. First, we propose a point cloud clustering methodology for polygon extraction on a panoramic range image generated with point-based rendering from a massive point cloud. Next, we describe an experiment that was conducted to verify our methodology with an indoor mobile mapping system in an indoor environment. This experiment was wall-surface extraction using a rendered point cloud from some viewpoints over a wide indoor area. Finally, we confirmed that our proposed methodology could achieve polygon extraction through point cloud clustering from a complex indoor environment.
Highlights
IntroductionModel-based clustering requires CAD models to estimate simple objects or point clusters from the point cloud
The point cloud taken from Trimble Indoor Mobile Mapping System (TIMMS) was rendered from these viewpoints
We have proposed a method for panoramic rendering-based polygon extraction from indoor mobile LiDAR data
Summary
Model-based clustering requires CAD models to estimate simple objects or point clusters from the point cloud. The CAD model preparation approach is unsuitable for modeling unknown objects. Edge-based and region-based clustering are often used to model unknown objects (Tsai et al 2010). These approaches focus on geometrical knowledge (Pu, et al 2009) and 2D geometrical restrictions, such as the depth from a platform (Zhou, et al 2008) and discontinuous point extraction on each scanning plane from the mobile mapping system (Denis et al 2010) to extract simple boundaries and features in urban areas
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have