基于三维激光扫描仪的点云配准
|
摘要
随着现代信息技术的飞速发展及图形图像应用领域的扩大,如何快速、准确的对现实世界中的三维实体进行数字化已成为急待解决的问题。而三维扫描技术的出现为该问题的解决提供了新的技术手段。该技术能快速、精确、无接触地获取复杂物体表面的三维信息(点云),进而完成实体三维重建。由于该技术的自身特点,要获得实体完整的点云数据,必须进行多视点数据采集。因而,多视点点云的配准成为基于三维扫描三维重建的一项关键技术,对模型重建的精度有重要影响。
本文针对点云配准问题进行了研究,主要工作包括以下几方面:
(1)从实验室现有设备及其功能出发,提出了基于反射体配准技术路线的数据采集流程,利用反射体进行了点云的配准,并结合实际工作中遇到的问题分析了基于反射体配准的缺点及局限性,明确了点云配准算法研究的重要性。
(2)通过对三维激光扫描仪的配套软件Riscan Pro的数据组织结构及所存储参数的分析,阐明了Riscan Pro参数解析工作对于点云配准及三维重建的重要性。鉴于RiscanPro的工程文件project.rsp是符合W3C标准的XML文档,本文概述了XML文档的语法基础,并利用SAX解析技术对Riscan Pro文档进行解析及参数获取。
(3)分析了ICP算法的优缺点及其改进算法的实现思路,提出了基于近邻搜索的多分辨率ICP改进算法,并结合Riscan Pro软件提供的手动配准功能,提出了一种改进的点云配准流程。
上述算法均在VC++平台下进行了实现,并以本实验室的三维激光扫描设备采集实验数据,对算法进行了实验,验证了上述算法的有效性。
With the rapid development of modern information technology and the expanding applications of computer graphics, how to quickly and accurately digitalize 3D entities in real world has become an urgent issue. The 3D laser scanner provides a new technology to deal with this problem, whicn can acquire pointclouds of complex object surface fast and precisely without contact and then completes the 3D reconstruction of entities. Because of its own feature, pointclouds must be acquired from several positions to complete point cloud model. So the registration of multi-view point clouds has become a key technology of 3D reconstrution based on 3D scanning,which has great effect to the precision of model reconstruction.
The above-mentioned issue is studied in this paper,and the main contributions are as follows:
(1) With the equipment of the laboratory, a flow of data acquiration base on registration via tiepoints is proposed and the pointclouds are registered. The advantages and disadvantages of registration via tiepoints are analysed.The importance of the research of pointcloud registration algorithm is emphasized.
(2) The data organization structure and saved parameters of Riscan Pro are analyzed. The importance of Riscan Pro parameter analysis to the point cloud registration and 3D reconstruction is clarified. As the project file of Riscan Pro is a XML file conforming to W3C the standard, this paper summarized XML document grammar foundations.And analyzed Riscan Pro file and acquired parameters with the technology of SAX.
(3) The advantages and disadvantages of ICP algorithm are analyzed. The improved ICP algorithm of neighbor searching multi-resolution is proposed and a registration flow based on Riscan Pro are also proposed.
The algorithms are all implemented under the platform of VC++ and carried on the experiment to the algorithm with datas acquired by 3D laser scanner. The validity is also confirmed.
本文针对点云配准问题进行了研究,主要工作包括以下几方面:
(1)从实验室现有设备及其功能出发,提出了基于反射体配准技术路线的数据采集流程,利用反射体进行了点云的配准,并结合实际工作中遇到的问题分析了基于反射体配准的缺点及局限性,明确了点云配准算法研究的重要性。
(2)通过对三维激光扫描仪的配套软件Riscan Pro的数据组织结构及所存储参数的分析,阐明了Riscan Pro参数解析工作对于点云配准及三维重建的重要性。鉴于RiscanPro的工程文件project.rsp是符合W3C标准的XML文档,本文概述了XML文档的语法基础,并利用SAX解析技术对Riscan Pro文档进行解析及参数获取。
(3)分析了ICP算法的优缺点及其改进算法的实现思路,提出了基于近邻搜索的多分辨率ICP改进算法,并结合Riscan Pro软件提供的手动配准功能,提出了一种改进的点云配准流程。
上述算法均在VC++平台下进行了实现,并以本实验室的三维激光扫描设备采集实验数据,对算法进行了实验,验证了上述算法的有效性。
With the rapid development of modern information technology and the expanding applications of computer graphics, how to quickly and accurately digitalize 3D entities in real world has become an urgent issue. The 3D laser scanner provides a new technology to deal with this problem, whicn can acquire pointclouds of complex object surface fast and precisely without contact and then completes the 3D reconstruction of entities. Because of its own feature, pointclouds must be acquired from several positions to complete point cloud model. So the registration of multi-view point clouds has become a key technology of 3D reconstrution based on 3D scanning,which has great effect to the precision of model reconstruction.
The above-mentioned issue is studied in this paper,and the main contributions are as follows:
(1) With the equipment of the laboratory, a flow of data acquiration base on registration via tiepoints is proposed and the pointclouds are registered. The advantages and disadvantages of registration via tiepoints are analysed.The importance of the research of pointcloud registration algorithm is emphasized.
(2) The data organization structure and saved parameters of Riscan Pro are analyzed. The importance of Riscan Pro parameter analysis to the point cloud registration and 3D reconstruction is clarified. As the project file of Riscan Pro is a XML file conforming to W3C the standard, this paper summarized XML document grammar foundations.And analyzed Riscan Pro file and acquired parameters with the technology of SAX.
(3) The advantages and disadvantages of ICP algorithm are analyzed. The improved ICP algorithm of neighbor searching multi-resolution is proposed and a registration flow based on Riscan Pro are also proposed.
The algorithms are all implemented under the platform of VC++ and carried on the experiment to the algorithm with datas acquired by 3D laser scanner. The validity is also confirmed.
引文
[1]Besl P J,McKay N D.A method for registration of 3D shapes[J].IEEE Transactions on Pattern Analysis and Maohine Intelligence,1992,14(2):239-256
[2]Y.Chen,GMedioni.Object modeling by registration of multiple range images.Image and Vision Computing,1992,10(3):145-155
[3]G.Turk,M.Levoy.Zippered Polygon Meshes from Range Images.In Proceedings of SIG -GRAPH,1994,311-318
[4]G..Barequet,M.Sharir.Partial Surface matching by using directed footprints.Computati -onal Geometry:Theory and Applications,1999,12(2):45-62
[5]K S Arun,T S Huang,S D Blostein.Least square fitting of two 3-D point sets.IEEE Transactions on Pattern Analysis and Machine Intelligence,1987,9(5):698-700
[6]Chen Y,Medioni G.Object modeling by registration of multiple range images[J].Image and Vision Computing,1992,10(3):145-155
[7]Potmesil M.Generating models of solid objects by matching 3D surface segments.Proc -eedingsof the 8th International Joint Conference on Artificial Intelligence,1983:1089-1093
[8]Masuda T,Yokoya N.A robust method for registration and segmentation of multiple ran -ge images[J].Computer Vision and Image Understanding,1995,61(3):295-307
[9]Johnson A,Hebert M.Surface registration by matching oriented points[C].Proceedings of International Conference on Recent Advances in 3-D Digital Imaging and Modeling,1997:121-128.
[10]Norman Walsh.A Technical Introduction to XML.http://www.xml.com/pub/a/98/10/guide0,html.1998
[11]Ann Navarro,Chuck White,Linda Burman.Mastering XML.北京电子工业出版社.2000
[12]李力鸿,邵敏,郑震坤.XML文档信息的几种转换方法分析与应用实例.计算机科学,2003,30(2):55-58.
[13]陆新年,陆新宇等.XML揭秘——入门.应用.精通.北京:清华大学出版社.2001.6
[14]Heather Williamson.XML技术大全.北京机械工业出版社.2002
[15]Elliotte Rusty Harold.XML使用大全.中国水利出版社.2002
[16]Natanya Pitts XML Black Book(2~(nd)Edition)机械工业出版社.2002.1
[17]李拮.XML高级编程.机械工业出版社.2001
[18]朱前飞,高芒.XML解析技术研究.电脑开发与应用,2004,17(11):138-142
[19]刘芳,削铁军.XML应用的基石:XML解析技术.计算机工程与设计,2005,26(10):78-82
[20]Elliotte Rusty Harold,W.Scott Means.XML in a Nutsbell 中国电力出版社,2001.11
[21]O.D.Faugeras,M.Hebert.The representation,recognition,and locating of 3d objects.International Journal of Robotic Research,.1986,5(3):27-52
[22]EStein,G.Medioni.Structural indexing:efficient 3-D object recognition.IEEE Transac -tions on Pattern Analysis and Machine Intelligence,1992,14(2):125-145
[23]Z.Zhang.Iterative point matching for registration of freefonrm curves and surfaces.Int -emational Journal of Computer Vision,1994,13(2):119-152
[24]C.Dorai,J.Weng,A.K.Jain.Optimal registration of object views using range data.IEEE Transactions on Pattern Analysis and Machine Intelligence,1997,19(10):1131-1138
[25]C.Dorai,G.Wang,A.K.Jain,C.Mercer.Registration and integration of multiple obj -ect views for 3D model constmction.IEEE Transactions on Pattern Analysis and Machine Intelligence,1998,20(1):83-89
[26]D.W.Eggert,A.W.Fitzgibbon,R.B.Fisher.Simultaneous registration of multiple range views for use in reverse engineering of CAD models.Computer Vision and Image Understan -ding,1998,69(3):253-272
[27]I.Stamos,P.K.Allen.Geometry and Texture Recovery of Scenes of Large Scale.Comp -uter Vision and Image Understanding(CVIU),2002,88(2):94-118
[28]DA.Simon,M.Hebert,T.Kanade,"Techniques for Fast and Accurate Intra Surgical Reg -istration".Joumal of Image Guided Surgery,1995,1(1):17-29
[29]S.Rusinkiewicz,M.Levoy,"Efficient Variants of the ICP Algorithm".Proceedings of 3DIM01,2001,17(3):145-152
[30]G.Turk,M.Levoy,"Zippered Polygon Meshes from Range Images",Proceedings of ACM Siggraph,1994,3(1):311-318
[31]Gregory C.Sharp,"ICP Registration Using Invariant Features",IEEE Transactions on Pattern Analysis and Machine Intelligence,2002,24(1):90-102
[32]S.Rusinkiewicz,M.Levoy,"Efficient Variants of the ICP Algorithm",Proceedings of 3DIM01,Quebec,2001,24(5):145-152.
[33]G.Godin,D.Laurendeau,R.Bergevin,"A method for the registration of attributed range images",Proceedings of 3DIM01,2001,28(6):179-186.
[34]M.Greenspan,G.Godin,"A nearest neighbor method for efficient ICP",Proceedings of 3DIM01,2001,31(3):161-168.
[35]G.Blais and M.Levine,"Registering multiview range data to create 3D computer obje -cts",IEEE Transactions on Pattern Snslysis and Machine Intelligence(PAMI),1995,17(8):820-824.
[36]R.Benjemaa,F.Sehmitt,"Fast Global Registration of 3D Sampled Surfaces Using a Multi-Z-Buffer Technique".Proceedings of the International Conference on Recent Advanc -es in 3D Imaging and Modeling(3DIM97),1997,41(10):113-119
[37]T.Jost,Fast Geometric Matching for Surface Registration,PhD thesis,University of Neuchatel,Switzerland,2002
[38]DA.Simon,M.Hebert,T.Kanade,"Techniques for Fast and Accurate Intra Surgical Registration",Journal of Image Guided Suigery,1995,1(1):17-29
[39]H.H(u|¨)gli,C.Seh(u|¨)tz,"Geometric Matching of 3D Objects:Assessing the Range of Suec -essful Initial Configurations",Proceedings of 3DIM97,1997,101-106
[40]T.Jost,H.H(u|¨)gli,"Fast ICP algorithms for shape registration",Proceedings of DAGM -02,Z(u|¨)rich,2002,91-99
[41]T.Jost,H.H(u|¨)gli,"A multi-resolution scheme ICP algorithm for fast shape registration",Proceedings of the 1 st Int.Syrup.on 3D Data Processing,Visualization and Transmission(3DP -VT02)Padova,2002,540-543
[42]带静兰,陈志杨,叶修梓.ICP算法在点云配准中的应用.中国图形图像学报,2007,12(3):517-521
[43]程俊廷,赵灿,王从军,莫健华.基于参考点和ICP算法的点云数据重定位研究.计算计测量与控制,2006,14(9):1222-1238
[44]朱根松,周天瑞,潘海鹏,胡世飞.激光扫描点云数据的快速读取新方法.计算机系统应用,2007,1:102-104
[45]梁永波.基于拾取点与ICP算法的三维图形重定位于实现.煤矿机械,2006.27(8):59-61
[46]刘繁名,屈昊.ICP算法的鲁棒性改进.仪器仪表学报,2004,25(4):603-635
[47]朱延娟,周来水,张丽艳.算乱点云数据配准算法.计算机辅助设计与图形学学报,2006,18(4):475-481
[48]路银北,张蕾,普杰信,杜鹏.基于曲率的点云数据配准算法.计算机应用,2007,27(1):2766-2769
[49]洪贝,孙继银..图像配准技术.战术导弹控制技术,2006,54(3):109-113
[50]张宗华,彭翔,胡晓唐.获取ICP匹配深度图像初值的研究.工程图学学报,2002,1:78-84
[2]Y.Chen,GMedioni.Object modeling by registration of multiple range images.Image and Vision Computing,1992,10(3):145-155
[3]G.Turk,M.Levoy.Zippered Polygon Meshes from Range Images.In Proceedings of SIG -GRAPH,1994,311-318
[4]G..Barequet,M.Sharir.Partial Surface matching by using directed footprints.Computati -onal Geometry:Theory and Applications,1999,12(2):45-62
[5]K S Arun,T S Huang,S D Blostein.Least square fitting of two 3-D point sets.IEEE Transactions on Pattern Analysis and Machine Intelligence,1987,9(5):698-700
[6]Chen Y,Medioni G.Object modeling by registration of multiple range images[J].Image and Vision Computing,1992,10(3):145-155
[7]Potmesil M.Generating models of solid objects by matching 3D surface segments.Proc -eedingsof the 8th International Joint Conference on Artificial Intelligence,1983:1089-1093
[8]Masuda T,Yokoya N.A robust method for registration and segmentation of multiple ran -ge images[J].Computer Vision and Image Understanding,1995,61(3):295-307
[9]Johnson A,Hebert M.Surface registration by matching oriented points[C].Proceedings of International Conference on Recent Advances in 3-D Digital Imaging and Modeling,1997:121-128.
[10]Norman Walsh.A Technical Introduction to XML.http://www.xml.com/pub/a/98/10/guide0,html.1998
[11]Ann Navarro,Chuck White,Linda Burman.Mastering XML.北京电子工业出版社.2000
[12]李力鸿,邵敏,郑震坤.XML文档信息的几种转换方法分析与应用实例.计算机科学,2003,30(2):55-58.
[13]陆新年,陆新宇等.XML揭秘——入门.应用.精通.北京:清华大学出版社.2001.6
[14]Heather Williamson.XML技术大全.北京机械工业出版社.2002
[15]Elliotte Rusty Harold.XML使用大全.中国水利出版社.2002
[16]Natanya Pitts XML Black Book(2~(nd)Edition)机械工业出版社.2002.1
[17]李拮.XML高级编程.机械工业出版社.2001
[18]朱前飞,高芒.XML解析技术研究.电脑开发与应用,2004,17(11):138-142
[19]刘芳,削铁军.XML应用的基石:XML解析技术.计算机工程与设计,2005,26(10):78-82
[20]Elliotte Rusty Harold,W.Scott Means.XML in a Nutsbell 中国电力出版社,2001.11
[21]O.D.Faugeras,M.Hebert.The representation,recognition,and locating of 3d objects.International Journal of Robotic Research,.1986,5(3):27-52
[22]EStein,G.Medioni.Structural indexing:efficient 3-D object recognition.IEEE Transac -tions on Pattern Analysis and Machine Intelligence,1992,14(2):125-145
[23]Z.Zhang.Iterative point matching for registration of freefonrm curves and surfaces.Int -emational Journal of Computer Vision,1994,13(2):119-152
[24]C.Dorai,J.Weng,A.K.Jain.Optimal registration of object views using range data.IEEE Transactions on Pattern Analysis and Machine Intelligence,1997,19(10):1131-1138
[25]C.Dorai,G.Wang,A.K.Jain,C.Mercer.Registration and integration of multiple obj -ect views for 3D model constmction.IEEE Transactions on Pattern Analysis and Machine Intelligence,1998,20(1):83-89
[26]D.W.Eggert,A.W.Fitzgibbon,R.B.Fisher.Simultaneous registration of multiple range views for use in reverse engineering of CAD models.Computer Vision and Image Understan -ding,1998,69(3):253-272
[27]I.Stamos,P.K.Allen.Geometry and Texture Recovery of Scenes of Large Scale.Comp -uter Vision and Image Understanding(CVIU),2002,88(2):94-118
[28]DA.Simon,M.Hebert,T.Kanade,"Techniques for Fast and Accurate Intra Surgical Reg -istration".Joumal of Image Guided Surgery,1995,1(1):17-29
[29]S.Rusinkiewicz,M.Levoy,"Efficient Variants of the ICP Algorithm".Proceedings of 3DIM01,2001,17(3):145-152
[30]G.Turk,M.Levoy,"Zippered Polygon Meshes from Range Images",Proceedings of ACM Siggraph,1994,3(1):311-318
[31]Gregory C.Sharp,"ICP Registration Using Invariant Features",IEEE Transactions on Pattern Analysis and Machine Intelligence,2002,24(1):90-102
[32]S.Rusinkiewicz,M.Levoy,"Efficient Variants of the ICP Algorithm",Proceedings of 3DIM01,Quebec,2001,24(5):145-152.
[33]G.Godin,D.Laurendeau,R.Bergevin,"A method for the registration of attributed range images",Proceedings of 3DIM01,2001,28(6):179-186.
[34]M.Greenspan,G.Godin,"A nearest neighbor method for efficient ICP",Proceedings of 3DIM01,2001,31(3):161-168.
[35]G.Blais and M.Levine,"Registering multiview range data to create 3D computer obje -cts",IEEE Transactions on Pattern Snslysis and Machine Intelligence(PAMI),1995,17(8):820-824.
[36]R.Benjemaa,F.Sehmitt,"Fast Global Registration of 3D Sampled Surfaces Using a Multi-Z-Buffer Technique".Proceedings of the International Conference on Recent Advanc -es in 3D Imaging and Modeling(3DIM97),1997,41(10):113-119
[37]T.Jost,Fast Geometric Matching for Surface Registration,PhD thesis,University of Neuchatel,Switzerland,2002
[38]DA.Simon,M.Hebert,T.Kanade,"Techniques for Fast and Accurate Intra Surgical Registration",Journal of Image Guided Suigery,1995,1(1):17-29
[39]H.H(u|¨)gli,C.Seh(u|¨)tz,"Geometric Matching of 3D Objects:Assessing the Range of Suec -essful Initial Configurations",Proceedings of 3DIM97,1997,101-106
[40]T.Jost,H.H(u|¨)gli,"Fast ICP algorithms for shape registration",Proceedings of DAGM -02,Z(u|¨)rich,2002,91-99
[41]T.Jost,H.H(u|¨)gli,"A multi-resolution scheme ICP algorithm for fast shape registration",Proceedings of the 1 st Int.Syrup.on 3D Data Processing,Visualization and Transmission(3DP -VT02)Padova,2002,540-543
[42]带静兰,陈志杨,叶修梓.ICP算法在点云配准中的应用.中国图形图像学报,2007,12(3):517-521
[43]程俊廷,赵灿,王从军,莫健华.基于参考点和ICP算法的点云数据重定位研究.计算计测量与控制,2006,14(9):1222-1238
[44]朱根松,周天瑞,潘海鹏,胡世飞.激光扫描点云数据的快速读取新方法.计算机系统应用,2007,1:102-104
[45]梁永波.基于拾取点与ICP算法的三维图形重定位于实现.煤矿机械,2006.27(8):59-61
[46]刘繁名,屈昊.ICP算法的鲁棒性改进.仪器仪表学报,2004,25(4):603-635
[47]朱延娟,周来水,张丽艳.算乱点云数据配准算法.计算机辅助设计与图形学学报,2006,18(4):475-481
[48]路银北,张蕾,普杰信,杜鹏.基于曲率的点云数据配准算法.计算机应用,2007,27(1):2766-2769
[49]洪贝,孙继银..图像配准技术.战术导弹控制技术,2006,54(3):109-113
[50]张宗华,彭翔,胡晓唐.获取ICP匹配深度图像初值的研究.工程图学学报,2002,1:78-84