地面光達利用雷射掃描技術可獲取地物表面大量的三維點位，由於每一掃描站具有掃描範圍的限制，通常須以多測站的方式進行場景掃描，再將所有測站點雲拼接成完整掃描場景的資料。然而因各測站的點雲資料均屬於不同的局部坐標系統，必須透過轉換將各站點雲整合至統一的坐標系統中。目前常見的方法為以點雲為基礎(Point cloud-based)之反覆最鄰近點法(Iterative Closest Point)，但此法對初始值的要求較高且容易受到粗差(Outliers)的影響，故本研究以特徵為基礎(Feature-based)的方式，來完成點雲的套合。

每一測站點雲中通常分佈著許多可利用的點、直線與平面特徵，共軛的特徵經坐標轉換至地面坐標系統時，應維持相同特徵性質，依據此特性可建立坐標轉換之觀測方程式，進行坐標轉換參數之整體平差計算。因此本研究從各測站點雲中萃取出共軛的點、直線及平面特徵，並於場景內佈設至少三個控制標點，利用所有共軛特徵及控制點進行多測站多特徵之聯合平差，解算各測站與地面坐標系統間的轉換參數，完成所有測站點雲的套合。

由實驗數據得知，利用點特徵進行傳統單站互相銜接之整體套合RMSE精度為0.752公尺，而在最小約制的條件下，點特徵多測站平差的整體套合結果則為0.097公尺，另外結合點線面特徵之聯合平差方面，其分別在E、N、H方向及整體上較點特徵多測站平差提升了44.2%、7.2%、20%與18.6%之精度，證實多特徵多測站的聯合平差可減少誤差傳播的影響且有助於增加點雲套合之準確度。

Ground-based LiDAR (Light Detection and Ranging) is capable of collecting a large amount of 3D point coordinates for landscape reconstruction. However, the observed point cloud of a single ground-based LiDAR station has limited scanning coverage. Combining point clouds obtained from several scanning stations is usually required to obtain a complete coverage of a landscape. It means that each point cloud defined its own local coordinate system should be transformed into a common coordinate system, which is usually the ground coordinate system. Iterative Closest Point (ICP) is one of popular approaches for point cloud registration, but it suffers from the low tolerance for outliers and the high requirement of initials. A feature-based adjustment using the point, line and plane features is proposed in this research to accomplish the point cloud registration.

Features of points, lines and planes can be found easily in a LiDAR point cloud. Conjugate features identified in different point clouds will maintain the invariant properties, when they are transformed into other coordinate systems. This enables the formation of observation equations using point, line and plane features to solve the parameters of coordinate transformation. Therefore, the adjustment computation of combining various features and ground control points can be performed for the registration of multi-station point clouds.

The experiments show that the overall RMSE of using point feature to register point clouds station by station is 0.792m, and under the minimum constraint, the multi-station registration result of using point feature is 0.097m; besides, the directions of E, N and H and overall RMSEs of combining point and line with plane features can be reduced 44.2%, 7.2%, 20% and 18.6% than only adopting point feature. As results reveal the multi-station point clouds adjustment of applying multi features can not only reduce the effect of error propagation but increase the accuracy of registration.

李姝儀，『從地面雷射掃描點雲萃取物面角特徵供多測站資料連結之研究』，國立成功大學測量及空間資訊學系碩士論文，2005。

郭朗哲，『地面光達於崩塌地地形製圖之應用』，國立交通大學土木工程學系碩士論文，2004。

莊子毅，『以三維直線特徵進行地面光達點雲套合』，國立台灣大學土木工程學系研究所碩士論文，2006。

曾義星、史天元，『三維雷射掃描技術及其在工程測量上之應用』，土木水利，2002。

湯凱佩，『以共軛平面特徵進行光達點雲資料結合之平差模式』，內政部辦理LiDAR之高精度極高解析度數值地形測繪、資料庫建置與應用推廣工作案成果發表暨應用研討會，pp. 15-24, 2005。

劉燈烈，『地面光達點雲資料的平差結合與影像敷貼』，國立成功大學測量及空間資訊學系碩士論文，2004。

劉嘉銘，『光達點雲資料特徵萃取之研究』，國立成功大學測量及空間資訊學系碩士論文，2005。

賴志凱，『地面雷射掃描儀的精度分析與檢定』，國立成功大學測量及空間資訊學系碩士論文，2005。

Acka, D., “Full Automatic Registration of Laser Scanner Point Clouds”, Optical 3D Measurement Techniques VI, pp. 330-337, 2003.

Besl, P.J. and N. D. Mckay, “A method for Registration of 3-D shapes”, IEEE Transection on Pattern Analysis and Machine Intelligence, Vol.14, pp. 239-256, 1992.

Boehler, W. and Marbs, A. “Investigating Laser Scanner Accuracy”, CIPA Symposium, Turkey, October, 2003.

Cobby, D.M., Mason, D.C. and Davenport, I.J., “Image Processing of Airborne Scanning, Laser Altimetry Data for Improved River FloodModeling”, ISPRS Journal of Photogrammetry and Remote Sensing, 56: pp. 121-138, 2001

Dold, C., and Brenner, C., “Registration of Terrestrial Laser Scanning Data Using Planar Patches and Image Data”, The ISPRS International Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences, 36(5), pp.78-83, 2006.

Gruen, A. and Acka, D., “Least Squares 3D Surface Matching”, International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol. 34, part 5/W16, 2005.

Habib, A., Ghanma, M., Morgan, M. and Al-Ruzouq, R.,“photogrammetric and Lidar Data Registration Using Linear Features”, American Society for Photogrammetry and Remote Sensing, Vol. 71, No. 6, pp.699-707, 2005.

Konno, T., Konno, K., Fujimoto, T., and Chiba, N., “Point Cloud Registration Based on Features Lines Derived from Depth Difference”, Proc. IWAIT2005, pp.499-504, 2005.

Maas, H.G., “Fast Determination of Parametric House Models From Dense Airborne Laser Scanner Data”, International Archives of Photogrammetry and Remote Sensing, Bangkok ,Thailand, 1999.

Point of Beginning, 2009 3D Laser Scanner Hardware and Software Surveys, referred on Nov 19, 2009. http://laser.jadaproductions.net/

Shah, T.R., “Automatic Reconstruction of Industrial Installations Using Point Clouds and Images”, Nederlandse Commissie voor Geodesie Netherlands Geodetic Commission(NCG), 2006.

Vosselman, G. and Maas, H.G.,“Adjustment and Filtering of Raw Laser Altimetry Data”, OEEPE Workshop on Airborne Laser Scanning and Interferometric SAR for Detailed Digital Elevation Models, Stockholm, pp.62-73, 2001.

Vosselman, G. and Dijkman, S.,“3D Building Model Reconstruction from Point Clouds and Ground Planes”, International Archives of Photogrammetry and Remote Sensing, Annapolis, Maryland. pp.37-43, 2001.

Wehr, A. and Lohr, U., “Airborne Laser Scanner – An Introduction and Overview”, ISPRS Journal of Photogrammetry and Remote Sensing, Vol. 54, Issues2-3, pp. 68-82, 1999.