本篇論文目的是利用MICROSOFT所發行的色彩深度攝影機KINECT來達到對真實物件快速建立其三維立體模型的目的。由於KINECT本身俱備深度資訊與色彩資訊，因此我們可利用深度資訊建立模型的網格；利用色彩資訊來進行材質貼圖，使得建立出來的模型更為擬真。
本論文主要使用Iterative Closeset Points(ICP)的技術來使得KINECT的深度影像可以互相對齊，利用對齊大量的深度影像來補強單一深度影像視角有限的不足，資料預先處理的方式是運用金字塔式的架構來降低點集合的數量，這個步驟可使得ICP的速度得以獲得提升，並從低解析度的影像中發現物件整體的運動方向，從高解析度的影像中找尋物件各個部分的運動變化。
ICP演算法可在兩個不同時間取得的深度影像中找尋彼此座標點之間的對應關係，這些具有對應關係的點稱之為對應點。在對應點中找尋距離較大的座標點，將其提至較高解析度的影像中再次進行的ICP對齊，直到所有對應點之間的距離皆小於一個可容忍範圍的閥值為止。在每次ICP的過程中會產生一個轉換矩陣，將各個座標點與對應的轉換矩陣進行矩陣運算後可將輸入的兩個深度影像對齊，接著合併兩個深度影像的三維座標點集合並建立網格，最後將網格依KINECT同時間所獲得的色彩資訊進行渲染，即完成動態物件的三維立體建模程序。

The purpose of this thesis is to make use of Microsoft’s KINECT for a rapid 3D model construction of objects in the real word. As KINECT is equipped with both the capability of capturing both color and depth, it can be used to construct a 3D mesh, together with the color and texture of mesh.
The main technique involved is Iterative Closest Point, or ICP in short, to align the point clouds captured from multiple viewing angles by KINECT, so that a more complete model can be obtained. As ICP is a time-consuming process, and the data for alignment is huge, a preprocessing step is performed to build a multi-resolution pyramid to reduce the data size and to speed up the ICP computation as a result. In addition, the global motion can be detected in the low-frequency component, while the local motion in the high-frequency component.
ICP can be used to align two point clouds captured at different times, and the goal is to find the correspondence in between. We start with the low-frequency component in the pyramid for ICP, and gradually refine to high-frequency component for the points which do not match well, until the correspondence of all points can be found. For each alignment, the derived transformation is performed to align a group of point cloud to the other. Finally the color/texture information acquired previously can be applied to achieve the 3D model construction.

[1]. DEY, T.K., SUGIHARA, K., AND BAJAJ, C. L. Delaunay triangulations in three dimensions with finite precision arithmetic. Comput. Aided Geom. Design 9 (1992), 457-470.
[2]. S. Izadi, D. Kim, O. Hilliges, D. Molyneaux, R. A. Newcombe,P. Kohli, S. Shotton, J. Hodges, D. Freeman, A. J. Davison, andA. Fitzgibbon. KINECTFusion: Real-time 3D reconstruction and interaction using a moving depth camera. In Symposium on User InterfaceSoftware and Technology (UIST), 2011.
[3]. R. A. Newcombe et al. Real-Time Dense Surface Mapping and Tracking with KINECT. In ISMAR, 2011.
[4]. Lu Xia, Chia-Chih Chen and J. K. Aggarwal. Human Detection Using Depth Information by KINECT. ACCV 2010
[5]. I. Oikonomidis, N. Kyriazis, and A. A. Argyros, Efficient Model-based 3D tracking of Hand Articulations using KINECT. in BMCV, 2009.
[6]. P.J. Besl, N.D. McKay, A method for registration of 3D shapes. IEEE Transactions on Pattern Analysis and Machine Intellinegce 14 (1992) 239–254.
[7]. D. Simon, Fast and accurate shape-based registration, Ph.D. Thesis, The Robotics Institute, Carnegie Mellon University, November 1996.
[8]. Dorai, C., Weng, J.. and Jain, A. “Registration and Integration of Multiple Object Views for 3D Model Constrution,” Trtrns. PAMI, Vol. 20, No. I . 1998
[9]. Jost, T., Huegli, H., 2003. A multi-resolution ICP with heuristic closest point search for fast and robust 3D registration of range images. IEEE International Conference on 3D Digital Imaging and Modeling, Banff, October 6–10, pp. 427–433.
[10]. RUSINKIEWICZ, S. AND LEVOY, M. 2001. Efficient Variants of the ICP Algorithm. Proc. 3DIM 2001
[11]. Mark Pauly , Markus Gross , Leif P. Kobbelt, Efficient simplification of point-sampled surfaces, Proceedings of the conference on Visualization '02, October 27-November 01, 2002, Boston, Massachusetts
[12]. Shaffer, E., Garland, M. Efficient Adaptive Simplification of MassiveMeshes. IEEE Visualization 01, 2001.
[13]. P. J . Burt and E. H. Adelson, “The Laplacian pyramid as a compact image code,” IEEE Trans. Commun., vol. 31, pp. 532-540, 1983.
[14]. Leonidas Guibas , Jorge Stolfi, Primitives for the manipulation of general subdivisions and the computation of Voronoi, ACM Transactions on Graphics (TOG), v.4 n.2, p.74-123, April 1985.
[15]. FORTUNE, S. 1987. A sweepline algorithm for Voronoi diagrams. Algorithmica 2, 153-174.
[16]. Garcia, M. J. and Steven, G. P. (1999). Fixed grid finite elements in elasticity problems. Engineering Computations, 16, 2, pp. 145–164.
[17]. Garcia, M. J., Ruiz, O., Ruiz, L. M., Querin, O . (2004). Fixed Grid Finite Element Analysis for 3D Linear Elastic Structures. VI World Congress on Computational Mechanics, Beijing, China, Sept.
[18]. Lempitsky, V. S., and Ivanov, D. V. 2007. Seamless mosaicing of image-based texture maps. In CVPR
[19]. BURT, P. J., AND ADELSON, E. H. 1983. The Laplacian pyramid as a compact image code. IEEE Transactions on Communication 31, 4, 532–540.
[20]. msdn KINECT for Windows blog.
website : http://blogs.msdn.com/b/KINECTforwindows
[21]. wikipedia. Delaunay Triangulation
website : http://en.wikipedia.org/wiki/Delaunay_triangulation
[22]. Julang Jiang , Zhong Huang , Feng Xue , Chunsheng Ou . 2009. Fast Algorithm for Texture Synthesis on Surfaces. Image and Signal Processing, 2009. CISP '09. 2nd International Congress on
[23]. Kenshi Takayama , Makoto Okabe , Takashi Ijiri , Takeo Igarashi, Lapped solid textures: filling a model with anisotropic textures, ACM Transactions on Graphics (TOG), v.27 n.3, August 2008
[24]. Fredo Durand , Julie Dorsey, Fast bilateral filtering for the display of high-dynamic-range images, ACM Transactions on Graphics (TOG), v.21 n.3, July 2002