摘要
博物馆是收集、保护和展示各种重要文物的场所,但展览空间有限,在实现资源共享、传播古老中国文化方面存在诸多限制。随着计算机和网络的发展,用数字化手段建成基于网络的数字化博物馆系统,可以较好地解决这些问题。而文物的三维重建技术则是数字化博物馆建设中最关键的技术之一。
旋转体是文物中比较常见的一类。在出土的大量文物中,许多陶器、瓷器、铁器、铜器等都是旋转体,旋转体的三维重建技术在文物保护、展示等方面具有重大意义。因此,研究如何从现实世界直接和快速地重建旋转体的三维模型,逐渐成为计算机图形学和计算机视觉领域的研究者努力的目标。
根据使用设备的不同,旋转体的三维重建主要可以分为两类方法:基于三维扫描的方法和基于二维图像的方法。前者能够精确获取旋转体的三维点云模型,但具有代价昂贵、使用方法具有侵略性、适用范围受限、数据采集不完善等缺点,并不适用于旋转体文物。后者则以使用摄像设备的方法为主,具备成本低、设备灵活、彩色纹理获取简单等特点。基于二维图像的方法也主要有两类:一类是由二维图像重建出三维点云模型,再进行表面重建,不但适用于旋转体对象,也适用于其它现实物体,但获取的点云模型通常存在噪声多、数据点稀疏不完整等问题。另一类充分利用了旋转体特有的几何结构,通过二维图像直接估计旋转体的旋转轴和母曲线,是旋转体三维重建研究的主流方法,但通常难以获取完整的纹理信息。
本文在前人工作的基础上,研究设计基于多幅图像的旋转体文物重建系统。该系统使用价格较低的照相机设备取代三维扫描仪来获取三维点云模型,既可以解决三维扫描仪设备价格昂贵的问题,还能获取到完整的纹理信息。同时充分利用旋转体固有的几何特性,通过计算旋转轴和母曲线重建三维表面模型,比通用的重建方法更有针对性,能有效处理点云模型获取过程中遗留下来的噪声多、数据点稀疏不完整等问题。
本文的主要工作包括以下几个方面:
1)利用旋转体结构上特有的几何特性,研究并实现基于图像的旋转体文物重建系统,并对系统中采用的关键技术进行研究,分析实验结果
2)采用SDM方法拟合母曲线。为了降低点云稀疏和噪声点多等问题对算法的影响,在进行曲线拟合之前,引入移动最小二乘法进行预处理,改善重建结果。
The museum is a place to collect, protect and display a variety of important archaeological heritages, but its space is limited. There are many limitations in the realization for sharing resources and disseminating the ancient Chinese culture. With the development of computers and network, these problems can be well solved through building a network-based digital museum system in using digital means. The three-dimensional reconstruction of archaeological heritages is one of the most critical technologies of building a digital museum.
Revolutions are the most common archaeological heritages. A large number of cultural relics, such as pottery, porcelain, iron, bronze, etc. are all revolution. The three-dimensional reconstruction technology of the revolution surface has great significance in the protection of cultural relics and archeology. Therefore, it becomes a goal of many researchers in the field of computer graphics and computer vision to reconstructing three-dimensional model of revolution directly and immediately.
According to the different equipments, the three-dimensional reconstruction of revolution can mainly be divided into two methods:to use the three-dimensional scanner as the representative approach and two-dimensional image-based approach. The former can accurately obtain the three-dimensional point cloud of revolution, but it has expensive, aggressive, limited in the scope of application, incomplete in collecting data and other shortcomings. So, this approach is not suitable to archaeological revolution. The latter mainly uses the camera equipment, which is low-cost, flexible and able to access color texture directly. This method also has two types. One type is reconstructing a three-dimensional point cloud model firstly, and then reconstructing three-dimensional surface models. It not only can be applied to the revolutions but also other real objects. But there are some shortcomings, for example, the noise points are more and the point cloud is sparse and incomplete. Another type takes advantage of the inherent geometric properties of revolution and estimates directly the rotational axis and the profile curve from two-dimensional images. It is common but often difficult to obtain complete texture information.
Based on the basis of previous work, this paper researches and designs an image-based reconstruction system of archaeological revolution. The system uses lower-priced camera equipment to capture 3D point cloud data in replace of the three-dimensional scanner, which not only can solve the problem of expensive equipment, but also can get full texture information. At the same time, this system takes full advantage of the inherent geometric properties of revolution to reconstruct the final three-dimensional surface model by calculating the rotational axis and profile curve. This method is can process some problems such as more noise, sparse or incomplete points and so on.
The main works of this paper are:
1) Based on the special geometric structure of revolution, studying and implementing a reconstruction system from two-dimensional images to three-dimensional surface model of archaeological revolution, studying several key technologies of the system and analyzing experiment results.
2) Fitting profile curve using SDM. In order to reduce the impact of sparse point cloud and noise points to the algorithm and to improve results, this paper introduce moving least-squares before making curve fitting.
引文
[1]王若苏,任洪娥,邱兆文.文物三维重建技术综述.科技信息:学术版.2007,34:334,433.
[2]向辉,孟祥旭,杨承磊.山东大学考古数字博物馆设计与实现.系统仿真学报.2003,15(3):319-321.
[3]Nister D. An Efficient Solution to the Five-Point Relative Pose Problem. IEEE Transactions on Pattern Analysis and Machine Intelligence.2004,26(6):756-770.
[4]Hartley R, Zisserman A. Multiple View Geometry in Computer Vision. Cambridge university press,2003.
[5]Furukawa Y, Ponce J. Accurate, Dense, and Robust Multi-view Stereopsis. IEEE Conference on Computer Vision and Pattern Recognition.2007.
[6]Chunmei Duan, Xiangxu Meng, Lu Wang.3D Reconstruction from Un-calibrated Images Taken from Widely Separated Views. IEEE International Conferences on Cybernetics and Intelligent Systems.2008,58-62.
[7]Edelsbrunner H, Mucke E P. Three-dimensional Alpha Shapes. ACM Transaction on Graphics,1994,13(1):43-72.
[8]Bajaj C L, Bernardini F, Xu G Automatic Reconstruction of Surfaces and Scalar Fields from 3D Scans. In Cook R ed. Proceedings of SIGGRAPH'95. Danvers: Assison-Wssley Publishing Company,1995,109-118.
[9]N.Amenta, M.Bern and M.Kamvysselis. A New Voronoi-based Surface Recon-struction Algorithm. SIGGRAPH 98, (1998),415-421.
[10]Amenta N., Choi S., Kolluri R.. The Power Crust, Unions of Balls, and the Medial Axis Ttransform. Computational Geometry:Theory and Applications 19 (2001),127-153.
[11]N.Amenta, S.Choi, T.K.Dey and N.Leekha. A Simple Algorithm for Homeo-morphic Surface Construction. Internal J. Comput. Geom. Applications,12, (2002).
[12]T.K.Dey and S.Goswami. Tight Cocone:A Water-tight Surface Reconstruction. J.Computing Informat, Sci.Engin,13(2003),302-307.
[13]T.K.Dey, S.Goswami. Provable Surface Reconstruction from Noisy Samples. In Proc.20th Annu.Sympos. Comput. Geom. (2004),330-339.
[14]Terzopoulos D, Platt J Barr, et al. Elastically Deformable Models [J]. Computer Graphics,1987,21(4):205-214.
[15]Ruperecht D, Nagel R, Muller H. Spatial Free-form Deformation with Scattered Data Interpolation Methods [J]. Computers & Graphics,1995,19(1):63-71.
[16]Turk G Re-tiling Polygonal Surfaces [J]. Computer Graphics,1992,26(4): 55-64.
[17]Pentland A, Sclaroff S E. Closed Form Solutions for Physical Based Shape Modeling and Recovery [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,1991,13(7):715-729.
[18]James D L and Pai D K. ARTDEFO:Accurate Real Time Deformable Objects [J]. Computer Graphics,1999,33(4):65-72.
[19]E.Bittar, N.Tsingos, M.-P.Gascuel. Automatic Reconstruction of Unstructured 3D Data:Combing a Medial Axis and Implicit Surface. Computer Graphics forum,1995, 14(3):467-468.
[20]Alexa M., BEHR J., Cohen-or D., Fleishman S., Levin D., Silva C.. Point Set Surfaces. In Proc. of the Conference on Visualization'01 (2001),21-28.
[21]Ohtake Y, Belyaev A., Alexa M., Turk G, Seidel H.. Multi-level Partition of Unity Implicits. TOG (2003),463-470.
[22]KAZHDAN M.. Reconstruction of Solid Models from Oriented Point Sets. SGP (2005),73-82.
[23]Kazhdanl M., Bolitho M. and Hoppe H.. Poisson Surface Reconstruction. Eurographics Symposium on Geometry Processing.2006.
[24]张祖勋,张剑清.数字测量摄影学[M].武汉:武汉大学出版社,2002.
[25]王之卓.摄影测量原理[M].武汉:武汉大学出版社,2007.
[26]张永军.基于序列影像的工业钣金件三维重建与视觉检测[D].武汉:武汉大学,2002.
[27]Wong K Y K, Mendonca P R S, Cipolla R. Reconstruction of Surfaces of Revolution from single uncalibrated views [C]. British:British Machine Visition Conference,2002.
[28]Utcke S., Zisserman A.. Projective Reconstruction of Surfaces of Revolution[C]. Magdeburg, Germany:DAGM03,2003.
[29]郑顺义,徐刚,张祖勋等.旋转体三维重建方法研究[J]'武汉大学学报,2006,31,(12):1047-1050.
[30]钱苏斌,林意.基于轮廓线的旋转曲面三维重建.计算机工程与设计,2009,30(16):3875-3878.
[31]Schmid C., Mohr R., Bauckhage C.. Evaluation of Interest Point Detectors. International Journal of Computer Vision.2000,37(2):151-172.
[32]Mikolajczyk K., Schmid C.. Indexing Based On Scale Invariant Interest Points. Eighth IEEE International Conference on Computer Vision.2001,1:525-531.
[33]Tuytelaars T, Van G Wide Baseline Stereo Matching Based on Local, Affine Invariant Regions. British Machine Vision Conference.2000.
[34]Luong Q, Faugeras O. The Fundamental Matrix:Theory, Algorithms, and Stability Analysis. International Journal of Computer Vision.1996,17(1):43-75.
[35]Pollefeys M, Van G, Vergauwen M, et al. Visual Modeling with a Hand-Held Camera. International Journal of Computer Vision.2004,59(3):207-232.
[36]Sturm P, Triggs B. A Factorization Based Algorithm for Multi-Image Projective Structure and Motion. Lecture Notes in Computer Science.1996,1065:709-720.
[37]Pollefeys M.. Tutorial on 3D Modeling From Images. Tutorial at ECCV.2000.
[38]Mahamud S, Hebert M, Omori Y, et al. Provably-Convergent Iterative Methods for Projective Structure From Motion. IEEE Computer Society Conference on Computer Vision and Pattern Recognition.2001,1:1018-1025.
[39]De Agapito L, Hartley R, Hayman E. Linear Self-Calibration of a Rotating and Zooming Camera. IEEE Computer Society Conference on Computer Vision and Pattern Recognition.1999,1:23-25.
[40]Hartley R. Self-Calibration of Stationary Cameras. International Journal of Computer Vision.1997,22(1):5-23.
[41]Hey den A, Astr K. Minimal Conditions on Intrinsic Parameters for Euclidean Reconstruction. Lecture Notes in Computer Science.1997:169-176.
[42]Hartley R, Hayman E, De A L, et al. Camera Calibration and the Search for Infinity, the Seventh IEEE International Conference on Computer Vision.1999,1: 510-517.
[43]Hartley R. Cheirality. International Journal of Computer Vision.1998, 26(1):41-61.
[44]Yacoub S B, Menard Ch. Robust Axis Determination for Rotational Symmetric Objects Out of Range Data, Burger and Burge[Burge97], pages 197-201.
[45]Sablatnig R, KampelM. Model-Based Registration of Front-and Back-views of Rotational Symmetric Objects. Computer Vision and Image Understanding,2002.
[46]Halir R. Estimation of the Axis of Rotation of Fragments of Archaeological Pottery, Burger and Burge [Burge97], pages 175-184.
[47]Pottmann H, Peternell M, Ravani B. An Introduction to Line Geometry with Applications [J], Computer-Aided Design,1999,31 (1):3-16.
[48]Halir R. An Automatic Estimation of the Axis of Rotation of Fragments of Archaeological Pottery:A Multi-step Model-based Approach [A]. In:Proceedings of the 7th International Conference in Central Europe on Computer Graphics, Visualization and Interactive Digital Media'99, Plzen2Bory,1999.
[49]Cao Y, Mumford D. Geometric Structure Estimation of Axially Symmetric Pots from Small Fragments [A]. In:Proceedings of The IASTED International Conference on Signal Processing, Pattern Recognition and Applications, Crete,2002.92-97.
[50]Pottmann H, Hofer M, et al. Line Geometry for 3D Shape Understanding and Reconstruction[A]. In:Proceedings of the 8th European Conference on Computer Vision, volume 3021 of Lecture Notes in Computer Science, Prague,2004.297-309.
[51]张宗芳,张宗霞.点云表示的旋转体文物碎片的整体特征的估计.湖南工业职业技术学院学报.2008,8(2):10-14.
[52]潘荣江,孟祥旭.陶器碎片旋转轴的健壮估计.计算机辅助设计与图形学学报.2005,17(11):2508-2511.
[53]Fischler M, Bolles R. Random Sample Consensus:A Paradigm for Model Fitting with Applications to Image Analysis and Automated Cartography [J] Graphics and Image Processing,1981,26(4):381-395.
[54]L.Fang, D.C.Gossard. Multidimensional Curve fitting to Unorganized Data Points by Nonlinear Minimization. Computer Aided Geometric Design,1995,27: 48-58.
[55]GTaubin, R.Ronfard. Implicial Models for Adaptive Curve Reconstruction. IEEE Transactions on Pattern Analysis and Machine Intelligence 1996,18(3):321-325.
[56]N.Amenta, M.bern, D.Eppstein. The Crust and the β-skeleton:Combinatorial Curve Reconstruction. Graphical Models and Image Processing,1998,60(2):125-135.
[57]H.Pottmann, T.Randrup. Rotational and Helical Surface Approximation for Reverse Engineering. Computing,1998,60(4):307-322.
[58]D. Levin. The Approximation Power of Moving Least-squares. Mathematics of Computation 67 (1998),1517-1531.
[59]D.Levin. Mesh-independent surface interpolation, private communication.
[60]I. Lee. Curve Reconstruction from Unorganized Points. Computer Aided Geome-tric Design,2000.
[61]A.A.Goshtasby. Grouping and Parameterizing Irregularly Spaced Points for Curve Fitting. ACM Transactions on Graphics,2000,19(3):185-203.
[62]Wenping Wang, Helmut Pottmann, Yang Liu. Fitting B-Spline Curves to Point Clouds by Squared Distance Minimization. ACM Transactions on Graphics(TOG), 2006,25(2):214-238.
[63]F. P. Preparata and M. I. Shomos, Computational Geometry:An Introduction (Springer-Verlag,1985).