Lidar点云数据中建筑物的快速提取

基金项目:

国家十五“863”计划: 面向网络海量空间信息的大型GIS 2001AA135170

Extract Buildings Quickly from Lidar Point Cloud Data

Faculty of Information Engineering , China University of Geosciences, Wuhan 430074, China

摘要: Lidar技术可快速获取地表的高精度三维点云数据, 目前对此类数据的分类却是速度慢、精度低, 尤其是城市区域建筑物和树木靠得较近时更是难以准确提取建筑物.介绍了一种基于点云数据生成距离影像, 而后引入对比度纹理辅助的点云数据建筑物快速提取方法.结果证明, 该方法不需要其他辅助数据就能实现点云数据中建筑物的快速提取.
- Lidar点云数据 /
- 距离影像 /
- 高程纹理 /
- 灰度共生矩阵
Abstract: Lidar can capture 3D geographical information quickly, and form a massive discrete point cloud.There are some shortcomings in the existing classification algorithms, such as low precision of classification and slow processing speed, especially in processing buildings close to trees. In this article, a new building extraction algorithm is introduced. The range image is abstracted from the original data and then the height texture is used for assisting classification. The authors tested the algorithm with Lidar data. The result of the experiment shows that this algorithm can distinguish adjacent buildings and trees. Also, the algorithm is good at classification precision and high speed. It has obvious advantages in the urban Lidar points cloud classification.
- Lidar point cloud /
- range image /
- height texture /
- co-occurrence matrix

图 1 距离影像

Fig. 1. Range image of Lidar point cloud

下载: 全尺寸图片幻灯片

图 2 规则化DSM

a.原始DSM; b.地面模型DTM; c.规则化后的DSM

Fig. 2. Determination of the normalized DSM

下载: 全尺寸图片幻灯片

图 3 在水平方向和垂直方向统计灰度共生矩阵

Fig. 3. Calculated co-occurrence matrix space with horizontal and vertical directions

下载: 全尺寸图片幻灯片

图 4 对比度纹理分类处理流程

1.规则化DSM; 2.垂直方向的对比度测量; 3.水平方向的对比度测量; 4.取2个方向的最小值; 5.滤去建筑物角点

Fig. 4. Flow chart of the contrast measurement classification

下载: 全尺寸图片幻灯片

图 5 高程对比度纹理分类算法流程

Fig. 5. Flow chart of the algorithm

下载: 全尺寸图片幻灯片

图 6 原始点云和分类结果

a.原始点云数据; b.分类结果, b中深灰色为形态学滤波得到的地面点, 浅灰色为建筑物

Fig. 6. Original points cloud and classification result

下载: 全尺寸图片幻灯片

[1]	Ben, G., 2002. Segmentation of TIN-structured surface models. ISPRS WG IV/6, joint conference on geo-spatial theory, processing and applications. Ottawa, Canada.
[2]	Elberink, S. O., Maas, H. G., 2000. The use of anisotropic height texture measures for the segmentation of airborne laser scanner data. IAPRS, 33: 678 -684.
[3]	Filin, S., 2002. Surface classification from airborne laser scanning data. IAPRS, 34: 119 -124.
[4]	Roggero, M., 2002. Object segmentation with region growing and principal component analysis. IAPRS, 34: 289 -294.
[5]	Zhang, K. Q., Chen, S., Whit man, D., et al., 2003. Progressive morphological filter for removing nonground measurements from airborne Lidar data. IEEE Transactionson Geoscience and Remote Sensing, 41 (4): 872 -882. doi: 10.1109/TGRS.2003.810682
[6]	Zhang, X. C., Huang, Z. C., Zhao, Y. H., 1996. Remote sense digital image process. Zhejiang University Press, Hangzhou. 291 -294 (in Chinese).
[7]	章孝灿, 黄智才, 赵元洪, 1996. 遥感数字图像处理. 杭州: 浙江大学出版社. 291 -294.