| dc.contributor | 地政系 | |
| dc.creator (作者) | Chio, Shih-Hong | |
| dc.creator (作者) | 邱式鴻 | zh_TW |
| dc.date (日期) | 2005 | |
| dc.date.accessioned | 21-七月-2015 17:23:13 (UTC+8) | - |
| dc.date.available | 21-七月-2015 17:23:13 (UTC+8) | - |
| dc.date.issued (上傳時間) | 21-七月-2015 17:23:13 (UTC+8) | - |
| dc.identifier.uri (URI) | http://nccur.lib.nccu.edu.tw/handle/140.119/76804 | - |
| dc.description.abstract (摘要) | Terrain information is implied in airborne LIDAR data, therefore algorithms should be developed to extract meaning information from them for subsequent application. Especially the building roofs in airbore LIDAR data are very important for 3-D building reconstruction. The difficulty of building roof extraction lies on how to exclude the irrelevant data and to exact them automatically. Therefore, this paper will present an approach to automatically acquiring the 3-D building roofs from airborne LIDAR data based on the data snooping theory. Firstly, coarse and fine TIN structures are constructed based on pyramided LIDAR data that are constructed from original airborne LIDAR data with different spacing distances. On the assumption that roofs are composed of either horizontal or slope planes, some better plane information, called TIN planes, are extracted by means of least squares fitting among coarse TIN structure where cover some fine TIN structures. Forward selection algorithm is used in data snooping, therefore a airborne LIDAR dataset of one best fitting plane is selected from the extracted TIN planes. Afterwards, the neighboring fine TIN structures of this dataset are selected and verified TIN by TIN if they could be merged into this dataset by the data snooping theory. If any one could be included into the dataset, new plane information is calculated by the least squares fitting. Using this new plane information again, neighboring fine TIN structures of the new dataset are selected and verified by the same approach. Until no fine TIN structure could be included, the entire plane information is extracted completely. By utilizing the same procedure, another better dataset is selected from unprocessed TIN planes. The same procedure is used to merge the fine TIN structures into a same plane. After all planes are extracted, the related planes should be merged into more complete planes. Then the object knowledge of 3-D building roofs are employed to differentiate the building roofs with other terrain objects or terrain surfaces. From the experiments, this paper will show the efficiency and feasibility of proposed approach. | |
| dc.format.extent | 176 bytes | - |
| dc.format.mimetype | text/html | - |
| dc.relation (關聯) | Asian Association on Remote Sensing - 26th Asian Conference on Remote Sensing and 2nd Asian Space Conference, ACRS 2005, Volume 3, Pages 1841-1849 | |
| dc.subject (關鍵詞) | Airborne IDAR data; Airborne LiDAR; Airborne lidar data; Automatic extraction; Building model; Building reconstruction; Building roof; Data sets; Data snooping; Forward selection; Least squares fitting; LIDAR data; Terrain surfaces; Algorithms; Buildings; Optical radar; Remote sensing; Roofs; Space optics; Tin; Three dimensional | |
| dc.title (題名) | Automatic extraction of 3-D building roofs by data snooping from airborne LIDAR data | |
| dc.type (資料類型) | conference | en |
