基于三维激光扫描的房屋尺寸质量智能化检测方法

基于三维激光扫描技术,提出了一种智能化、全流程的房屋尺寸质量检测方法,包括点云数据配准、点云数据轻量化、房屋逆向建模及尺寸质量检测。通过点云数据映射全景图、基于YOLO v5神经网络模型的标靶纸目标检测以及基于模板匹配方法的标靶中心估计等步骤,可实现多站点云数据之间的自动配准; 通过可分解图滤波算法进行点云数据重采样,实现数据轻量化并提高运行速度; 针对房屋整体点云数据,提出了集点云数据分割、表面重建、尺寸质量检测于一体的综合算法。结果表明:基于标靶纸的点云配准方法能够自动完成各站点云数据的配准,得到完整房屋点云数据; 点云数据分割技术能够分离不同墙面、楼板底面和顶面的点云数据; 表面重建算法能够生成房屋的实体模型; 尺寸质量检测技术能够自动计算出表面的平整度和垂直度; 提出的房屋尺寸质量检测方法全面、可行,且能够适用不同的户型,研究成果以期替代人工测量完成房屋的平整度与垂直度的检测。

Intelligent Inspection Method for Dimensional Quality of Houses Based on 3D Laser Scanning

An intelligent and full-process method for dimensional quality inspection of houses was proposed based on three dimensional(3D)laser scanning technology, including point cloud data(PCD)registration, PCD simplification, reverse modeling and dimensional quality inspection of houses. PCDs obtained by multi-site were automatically registered through panoramas mapped by PCDs, checkerboard targets detection using YOLO v5 neural network model and target centers estimation using template matching algorithm. In order to accomplish the simplification and accelerate runtime, decomposed graph filtering algorithm was adopted to resample the PCDs. A comprehensive algorithm integrating PCD segmentation, surface reconstruction and dimensional quality inspection was proposed for the registered PCD of houses. The results show that the checkerboard targets-based registration method can register the PCDs of each site automatically and obtain the complete PCD of houses. The PCDs of wall, floor and ceiling can be separated by the PCD segmentation technique and solid model can be built using the surface reconstruction algorithm. Flatness and verticality can be computed by the dimensional quality inspection technique. The proposed intelligent inspection method for the dimensional quality of houses is comprehensive, feasible and applicable to different house types. The research achievements contribute to replace manual measurement to achieve the inspection of flatness and verticality.