由空載光達資料進行建物偵測與結構線萃取

Building Detection and Structure Line Extraction from Airborne LiDAR Data

王正楷(Cheng-Kai Wang)

指導教授 ： 徐百輝

國立臺灣大學/工學院/土木工程學研究所/碩士(2007年)

https://doi.org/10.6342/NTU.2007.00654

摘要

由於空載光達系統之盲訊號特性，其並不一定能直接獲取欲觀測物體的特徵點位座標，因此諸如建物的屋角點、邊緣線、屋頂面等特徵，皆須經過後續的處理才能得到。而本研究的主要目的是針對原始的光達資料，先偵測出建物位置，再進行建物結構線萃取。整個研究方法主要的概念為利用光達資料的顯著資訊─高程，與隱藏資訊─建物之幾何特徵及物理意義來進行。在建物偵測方面，利用小波分析的方法得到多解析度的邊緣，進而選擇合適尺度下的邊緣線，藉由搜尋封閉的建物輪廓得到初步的建物位置，並以建物的幾何特性─高程、面積、紋理等來提升偵測成果，最後得到獨棟建物之位置。而在結構線的偵測方面，主要將屋頂結構線分為兩個部份來進行，第ㄧ為外部輪廓線，萃取的方式是利用霍夫轉換及其他附加幾何條件來進行；第二為內部結構線，萃取的方式為利用相鄰屋頂面的交會來獲得；最後再將此二種特徵線合併在一起，進而重建完整的屋頂面形狀。根據本實驗的成果顯示，ㄧ些經典的屋頂面，例如平屋頂、山形屋和L形屋頂皆可在沒有任何輔助資訊的加入下，順利完成屋頂面的重建。

關鍵字

光達 ； 建物偵測 ； 結構線萃取 ； 小波 ； 外部輪廓線 ； 內部結構線

並列摘要

Because of the blind characteristics of laser signals from the airborne LiDAR system, we can not directly get the 3D coordinates of object’s features such as the roof corners, edges, and complete object’s faces of buildings, etc. which could be obtained by processing the LiDAR data. The goal of this research is to detect the locations of buildings and then extract the structure lines of buildings from the raw LiDAR data. The basic idea is to use the visible information, the elevation, and the hidden information, geometry and physic properties of buildings to obtain the locations and structure lines of buildings. In the aspect of building detection, the wavelet analysis was used firstly to build up the multi-resolution edges, and then a set of edges at an apposite wavelet scale was chosen. Secondly, the initial locations of buildings were obtained by searching the closed edges and the detection results were refined by the elevation, area and texture of buildings. Finally, the locations of independent buildings were obtained. In the aspect of structure line detection, the main idea is to separate the roof structure lines into two parts, one is for external contours and the other is for internal structure lines. The external contours were extracted using Hough transform and other geometric conditions. The internal structure lines were extracted by any two adjacent roof faces intersecting each other. Finally, the two kinds of structure lines were combined to construct the roof shapes. The experiment results showed that the classic roofs such as plane roofs, gabled roofs and L-shape roofs could be reconstructed successfully without any auxiliary information.

並列關鍵字

LiDAR ； Building Detection ； Structure Line Extraction ； Wavelet ； External Contours ； Internal Structure Lines

參考文獻

張智安、陳良健 (2005)。整合光達資料與高解析衛星影像於建物偵測。航測及遙測學刊，第十卷，第四期，第361-371頁。

Ackermann, F. (1999). Airborne laser scanning--present status and future expectations. ISPRS Journal of Photogrammetry and Remote Sensing 54(2-3): 64-67.

Axelsson, P. (1999). Processing of laser scanner data--algorithms and applications. ISPRS Journal of Photogrammetry and Remote Sensing 54(2-3): 138-147.

Baltsavias, E. P. (1999). A comparison between photogrammetry and laser scanning. ISPRS Journal of Photogrammetry and Remote Sensing 54(2-3): 83-94.

Dash, J., E. Steinle, Singh, R. P. and Bahr, H. P..(2004). Automatic building extraction from laser scanning data: an input tool for disaster management. Advances in Space Research 33(3): 317-322.