期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2008
卷号：XXXVII Part B5
页码：523-526
出版社：Copernicus Publications
摘要：Mainly methods used currently for point cloud in the terrestrial three-dimensional laser scanner are to processing small industrial parts, or sculptures, in which the data format and the corresponding algorithms can not be directly to deal with the building. To solve the problem, the paper presents a multi-direction extracting method of the building contour based on point cloud of the terrestrial laser scanner. First the rude point cloud data is transformed into the three two-dimensional data matrices, which is composed separately of x, y or z coordinates. Secondly the noise points in each matrix such as x or y, are removed by weighted median filter. Using modulus maxima detection, the directional derivatives of each filter matrix can be obtained, and then after the threshold and binarization, the edges in each matrix are extracted. Lastly, based on the edge data of the three matrices, the original point cloud data are fused, and the building contour can be reconstructed. The Leica HDS3000 has been used to get a building point cloud and PTX format data. Results in the experiment show that the approach is very effect for point cloud extraction. Because of considering the characteristics of the building and using data in the directions of x, y and z, the accuracy of building contour points is fine
关键词：Terrestrial Laser scanning; Pointe Cloud Images; Multi-direction; Data Transformation; Noise Eliminated by Soft- ; threshold; Building Extraction
