以最小二乘平面套合法進行空載與車載光達點雲套合

Registration of Airborne Lidar and Mobile Lidar Point Clouds Using Least Squares 3-D Surface Matching

黃世涵(Shih-Han Huang)

指導教授 ： 張智安

國立交通大學/工學院/土木工程系所/碩士(2012年)

https://doi.org/10.6842/NCTU.2012.00914

摘要

光達(Light Detection and Ranging, LIDAR)為近年來快速發展的測距儀器，光達利用短時間內發射大量雷射脈衝對空間中的物體進行測距，並獲取雷射脈衝在物體表面反射點的三維坐標，大量的三維坐標點位集合為光達點雲資料(Point Clouds)。光達可裝置在不同載具，將不同載具或不同測站間的資料轉換至同一坐標系統，並使重疊區有一致的幾何特性，以整合點雲資料進行後續應用的行為即稱為光達點雲套合。
本研究採用最小二乘平面匹配法(Least Squares 3-D Surface Matching)進行空載及車載光達點雲套合。研究中使用共軛平面特徵建立坐標轉換關係，並以最小二乘法求解轉換參數達到點雲套合的目標。為建立平面特徵間的共軛關係，本研究採用資料網格結構化並使用特徵向量分析將網格結構中的點雲轉換為平面特徵，最後利用角度及距離門檻產生候選共軛平面。
本研究中使用一組模擬資料以及兩組真實資料進行點雲套合，為了解點雲密度比與套合成果之關係，先使用模擬點雲資料在已知轉換參數環境下進行套合分析；真實點雲資料套合部分使用空載光達點雲以及車載光達點雲，空載光達點雲由Optech ALTM 30/70掃描，資料範圍為台北市政府周邊都市區；車載光達點雲由Optech Lynx掃描，資料範圍為台北市光復南路之部分路段。真實點雲資料須先以航帶間套合消除資料內航帶間誤差，接著進行車載光達與空載光達之間套合。
成果分為模擬資料成果以及真實資料成果，模擬資料部分顯示出資料密度影響套合成果，將點雲密度調降為1/10時，套合誤差將放大為兩倍；在真實資料套合實驗成果中，進行航帶間點雲套合可將空載光達航帶間之誤差由原始資料的9cm改善為4.7cm，而車載光達資料航帶間誤差則由原始資料之40cm降低為4cm；於空載光達對車載光達點雲
II
套合中，原始點雲誤差為84cm也於套合處理後降低為5cm。經過套合處理後，資料間誤差皆明顯降低。

關鍵字

最小二乘平面匹配法 ； 套合 ； 特徵匹配 ； 光達點雲

並列摘要

LIDAR (Light Detection and Ranging, LIDAR) is a rapid development technology. Lidar emits a large number of laser pulses in a short time to measure the three-dimensional coordinates of the object. These three-dimensional coordinates acquired by lidar called point clouds. Lidar can be equipped with different platforms. In order to combine lidar data from different platforms, point clouds registration is applied to improve the geometrical consistence results between lidars.
In this study, we proposed a least squares 3D surface registration process to register two different point cloud datasets. First, we divide the point clouds into a 3-D voxel structure. Then, we use Principle Component Analysis to calculate the normal vector of plane in eachvoxel. The distance and angle between planes are analyzed to construct the candidate conjugate planes, and finally the unknown transformation parameters are solved by least squares adjustment.
This study use a simulation data and two real data in point clouds registration. In order to analyze the effect of point density in registration, the simulation data simulates different conditions and estimates the errors. The real data are airborne and mobile lidars which were acquired by Optech ALTM 30/70 and Optech Lynx in Taipei city.
In the experiment result, the simulation analysis indicates that registration error is increasing when the density of lidar is decreasing. For real dataset, the registration error
IV
between airborne lidar strips is improved from 9cm to 4.7cm. The registration error of mobile lidar between different paths is improved from 40cm to 4cm. Finally, the registration error of airborne and mobile lidars is improved from 84cm to 5cm.

並列關鍵字

least squares 3-D surface matching ； registration ； point clouds

參考文獻

Akca, D. (2007). "Matching of 3D surfaces and their intensities." ISPRS Journal of Photogrammetry and Remote Sensing 62(2): 112-121.

Barnea, S. and S. Filin (2008). "Keypoint based autonomous registration of terrestrial laser point-clouds." ISPRS Journal of Photogrammetry and Remote Sensing 63(1): 19-35.

Besl, P. J. and N. D. McKay (1992). "A method for registration of 3-D shapes." IEEE Transactions on pattern analysis and machine intelligence 14(2): 239-256.

Gruen, A. and D. Akca (2005). "Least squares 3D surface and curve matching." ISPRS Journal of Photogrammetry and Remote Sensing 59(3): 151-174.

Habib, A., K. I. Bang, A. P. Kersting and D. C. Lee (2009). "Error budget of LiDAR systems and quality control of the derived data." Photogrammetric engineering and remote sensing 75(9): 1093-1108.