点云
人工智能
计算机科学
分割
形状上下文
计算机视觉
分类器(UML)
模式识别(心理学)
数据挖掘
图像(数学)
作者
Jingdao Chen,Zsolt Kira,Yong Cho
出处
期刊:Journal of Computing in Civil Engineering
[American Society of Civil Engineers]
日期:2019-07-01
卷期号:33 (4)
被引量:96
标识
DOI:10.1061/(asce)cp.1943-5487.0000842
摘要
Construction progress estimation to ensure high productivity and quality is an essential component of the daily construction cycle. However, using three-dimensional (3D) laser-scanned point clouds for the purpose of measuring deviations between as-built structures and as-planned building information models (BIMs) remains cumbersome due to difficulties in data registration, segmentation, annotation, and modeling in large-scale point clouds. This research proposes the use of a data-driven deep learning framework to automatically detect and classify building elements from a laser-scanned point cloud scene. The point cloud is first converted into a graph representation, in which vertices represent points and edges represent connections between points within a fixed distance. An edge-based classifier is used to discard edges connecting points from different objects and to form connected components from points in the same object. Next, a point-based object classifier is used to determine the type of building component based on the segmented points and augmented with context from surrounding points. Finally, each detected object is matched with a corresponding BIM entity based on the nearest neighbor in the feature space.
科研通智能强力驱动
Strongly Powered by AbleSci AI