单体建筑物点云数据的CSF净化处理

从场景整体点云数据中提取单体建筑物的点云是建筑物单体三维建模的基础。然而,现有点云提取方法在提取建筑物点云数据时往往包含部分植被、地面等非建筑数据点,不利于建筑物对象建模。针对该问题,本文提出使用CSF方法对初步提取的建筑物点云数据进行净化处理。该方法首先将场景点云数据投影生成点云图像,根据图像特征初步提取单体建筑物点云数据;然后对获得的单体建筑物点云数据采用CSF方法进行净化处理,可以获得较为纯净的单体建筑物点云数据。本文以南京师范大学仙林校区部分区域为研究对象对该方法进行了验证。结果表明,该方法可以较好地对建筑物点云数据进行净化,得到较为纯净的单体建筑物点云数据,为基于点云数据的建筑物单体模型构建打下了良好的基础。

Purification of single building point cloud data using cloth simulation filter

Extracting the point cloud of single building from the whole point cloud data of the scene is the basis of 3D modeling of a single building. However, the existing methods of point cloud extraction often include some non-building data points such as vegetation and ground, which is not conducive to building object modeling. To solve this problem, this paper proposes to use CSF to purify the preliminarily extracted point cloud data of a building. Firstly, the scene point cloud data is projected to generate a point cloud image, and then the point cloud data of single building is extracted according to the image features. Secondly, the point cloud data of single building is purified with CSF method, andpure point cloud data of single building can be obtained. The method is validated in a part of Xianlin Campus of Nanjing Normal University. The results show that the method can purify the building point cloud data efficiently which facilitates the 3D model construction of a single building based on the point cloud data.