結合光達點雲及航照影像重建直線與圓弧輪廓建物

NCUIR > Engineering College > Graduate Institute of Civil Engineering > Electronic Thesis & Dissertation > Item 987654321/1147

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/1147

Title:	結合光達點雲及航照影像重建直線與圓弧輪廓建物
Authors:	謝其亨;Chi-Heng Hsieh
Contributors:	土木工程研究所
Keywords:	建物重建;建物模型;光達;建物偵測;數位空照影像;building reconstruction;building model;aerial image;LIDAR;building detection
Date:	2006-06-19
Issue Date:	2009-09-18 17:21:41 (UTC+8)
Publisher:	國立中央大學圖書館
Abstract:	數碼城市(Cyber City)是都市規劃，建設，及管理之未來重要發展方向。而三維建物模型是數碼城市中不可或缺的部分。本研究以資訊融合的方式，結合光達及航照資料，重建建物模型。考量建物之多樣性，重建的重點將包括直線及圓弧輪廓之建物。先利用高程及光譜資訊，偵測房屋區塊，再由網格化光達資料以切片法得到建物概略輪廓線及階梯輪廓線。將概略輪廓線反投影至影像空間中建立工作區域，進行圓弧偵測及直線偵測。接著將圓弧及直線線段結合建物頂共面分析結果投影至物空間，得到物空間三維階梯線段及建物外圍輪廓線，並再次進行共面分析求取屋脊線。最後利用SMS法建立二維位相關係並模塑產生建物模型。本研究測試區位於新竹工業技術研究院。航照影像之空間解析力為50 cm，光達點雲密度為2.15(點/平方公尺)。建物區塊偵測成功率為91%。在偵測成功的建物區塊內，建物模型重建完全正確及部分正確之成功率為90%。建物輪廓部分均方根誤差在X方向為0.73 m，Y方向為0.71 m。模塑誤差為0.12 m。 Cyber city modeling is an important work in the city planning, construction, and management. Three dimensional building models are the indispensable component in the cyber city. Selecting the approach of data fusion, this investigation combines LIDAR point clouds and aerial images to reconstruct building models. Considering the variety of buildings, the reconstruction includes buildings with line and arc outlines. By the integration of geometric and radiometric information first, the building regions are extracted followed by the method of elevation slicing to obtain initial building outlines and step edges. Through the back projection of the outlines to the image space, the arcs and straight lines are obtained. Then the arcs and the straight lines are projected to the object space. Employing roof analysis, we determine the three dimensional edges and building outlines. Finally, the SMS method is applied to generate the building models. The test site is in the Industrial Technology Research Institute of Hsinchu. The spatial resolution of aerial image is 50 cm, and the point density of LIDAR data is 2.15(point/m2). The success rate of building detection is 91%. Among the successfully detected buildings, 90% are fully or partially reconstructed. The RMSE of building boundaries are 0.73 m and 0.71 m in X and Y directions, respectively. The shaping error is 0.12 m.
Appears in Collections:	[Graduate Institute of Civil Engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Size	Format

社群 sharing

Loading...