基于激光雷达的城市数字化建模技术研究

发布时间：2018-05-18 11:12

本文选题：车载激光雷达 + 点云滤波　；参考：《长春理工大学》2017年硕士论文

【摘要】：车载激光雷达技术作为一种新兴的城市测量技术,为重建逼真的建筑物三维模型提供了一种新途径。本文在研究车载激光雷达对地定位原理、点云数据特点及各类激光雷达建筑物建模方法的基础上,确定车载激光雷达建筑物建模方案。首先,依据城市点云数据特点,提出一种基于坡度差值的点云滤波方法,通过对比实验,验证该方法滤波的高效性与实用性。此外,设定参数阈值并利用随机抽样一致性算法,对建筑物立面点云进行分割并通过实验分析距离阈值设定对处理结果的影响。然后,通过改进基于语义识别的建筑物立面外轮廓提取方法并结合基于投影点密度格网化的建筑物内轮廓提取方法完成建筑物轮廓特征提取。最后,利用提取出的建筑物轮廓结合面片拓扑关系,构建建筑物几何框架并将其中的特征数据编写成DXF文件,导入AtuoCAD中,自动生成模型并利用3DMAX软件对重建的建筑物模型进行纹理贴图使其更加形象逼真。本文以车载激光雷达系统获取的街区原始点云数据为基础,以城市建筑物数字化建模为主要研究内容,对车载激光雷达点云数据滤波、建筑物立面点云分割、建筑物立面轮廓提取及建筑物建模等方面进行了着重研究,在算法与理论方面取得了一定进展,为城市数字化建模提供了可靠的理论依据,具有一定的参考价值。
[Abstract]:As a new urban measurement technology, vehicle-mounted lidar provides a new way to reconstruct realistic 3D building models. Based on the study of the principle of vehicle lidar location, the characteristics of point cloud data and the modeling methods of various lidar buildings, the building modeling scheme of vehicular lidar is determined in this paper. Firstly, according to the characteristics of urban point cloud data, a point cloud filtering method based on gradient difference is proposed. The effectiveness and practicability of the filtering method are verified by comparative experiments. In addition, the parameter threshold is set and the random sampling consistency algorithm is used to segment the elevation point cloud of the building. The effect of the distance threshold setting on the processing results is analyzed experimentally. Then, building contour feature extraction is accomplished by improving the method of extracting the outer contour of building elevation based on semantic recognition and combining with the contour extraction method based on projection point density mesh. Finally, the geometric frame of the building is constructed by using the extracted building contour and the topological relation of the facet, and the characteristic data is written into DXF file and imported into AtuoCAD. The model is generated automatically and the reconstructed building model is textured by 3DMAX software to make it more vivid. Based on the original point cloud data obtained by vehicle lidar system, the digital modeling of urban buildings is the main research content in this paper. The point cloud data of vehicle lidar is filtered, and the elevation cloud of buildings is segmented. In this paper, the extraction of building facade contour and building modeling are emphatically studied, and some progress has been made in algorithm and theory, which provides a reliable theoretical basis for urban digital modeling and has certain reference value.
【学位授予单位】：长春理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P225;TN958.98

【参考文献】