Semiautomatic generation of three-view drawing of building using terrestrial laser scanning

35th International Symposium on Remote Sensing of Environment

地球科学;生态环境科学

Liang, Yubin^1 ; Zhan, Qingming^2 ; Che, Erzhuo^1 ; Chen, Mingwen^3

School of Remote Sensing and Information Engineering, Wuhan University, Wuhan, China^1

School of Urban Design, Wuhan University, Wuhan, China^2

State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, Wuhan, China^3

关键词: Cultural Heritage documentations;    Orthographic projections;    Pattern recognition method;    Semi-automatic generation;    Structural information;    Terrestrial laser scanning;    Three-dimensional data;    Three-dimensional laser scanning;   

Others  :  https://iopscience.iop.org/article/10.1088/1755-1315/17/1/012230/pdf
DOI  :  10.1088/1755-1315/17/1/012230

学科分类：环境科学（综合）

来源: IOP

【 摘 要 】

Terrestrial laser scanning is an effective and efficient technique for acquisition of three dimensional data of indoor and outdoor environment in a short period of time. Precision of laser scanning data are usually within millimetres, which is satisfactory for building surveying and mapping. In recent years terrestrial laser scanning has been widely used in historical building preservation and cultural heritage documentation. Three-view drawing (plan, front and section views) is standard and important presentation of building surveying and mapping. However, generation of three-view drawing of a building using terrestrial laser scanning data often entails much human intervention. In this paper we present a methodology for semiautomatic generation of three-view drawing of a building. Three-view drawing of a building is often made on virtual planes which are perpendicular to the axis directions of the building. We define the projection plane using interactively selected laser points of the building surface and project point cloud to the determined projection plane. We project point cloud data to such a virtual plane defined by interactively selected points on the surface of building. A depth image is generated based on the distance between points and the virtual plane. The generated depth image is orthographic projection of three-dimensional laser scanning scene, which preserves the structural information of a building. Then segmentation and pattern recognition methods are exploited to extract the features (geometric primitives) from the depth image. The extracted features can be further refined to generate three-view drawing of a building. The presented methodology greatly reduces volume of data in operation and experimental results show the effectiveness of the methodology.