基于实拍照片的纹理映射研究
|
摘要
虚拟现实技术是计算机图形学和计算机视觉领域研究的一个重要课题,在文物保护、电影特效、广告制作、电子商务等领域有着广泛的应用。对于真实场景的三维重建,通常采用基于三维几何模型为其表面添加纹理的方法来增强真实感,这个过程一般需要两步,构建三维物体的几何模型和为几何模型添加纹理。目前,采用三维扫描采集数据的方法已经可以对真实对象的形状进行三维重建,重建对象纹理就成为了三维重建的研究重点。基于实拍照片的纹理映射主要是指利用数码照相机拍摄获得真实对象的表面纹理,然后利用透视投影矩阵建立三维模型和纹理图像之间的映射关系,将真实纹理映射到三维模型表面,使模型具有真实感。本文针对实拍照片纹理映射中的关键技术进行了研究,主要包括以下内容：
(1)分析了纹理映射技术在国内外的发展概况,介绍了与纹理映射技术相关的计算机图形学、计算机视觉、数码相机参数标定理论、图像配准方法等知识；
(2)提出了一种利用四组对应点求解相机外部参数的几何标定方法,利用人机交互的方法选取了4组对应点,分别求取了旋转矩阵和平移矩阵,得到了相机的全部外部参数,然后根据相机的内外参数信息求解了透视投影矩阵,建立了三维模型和数码照片之间的映射关系,实现了3D-2DD特征点的配准；
(3)基于3D-2D特征点的配准方法关键是确定三维模型和二维图片的3D-2D匹配点,采用了人机交互的方式选取对应特征点对的方法,并提出了一种基于透视投影的快速三维拾取方法,可以较为精确的拾取到模型的三维空间坐标；
(4)针对纹理图像存在视角变化的特点,在研究SIFT特征配准算法基础上,采用了马氏距离对SIFT算法误匹配点进行剔除,提高了图像配准的准确率。
Virtual reality is an important subject in the field of computer graphics and computer vision, and it has a wide range of applications in cultural relics preservation, special effects on movies, advertising, e-commerce and other fields. Usually,to the 3D reconstruction of real scenes, it will enhance realism through the method that adding texture for the 3D surface model. The process usually takes two steps constructing three-dimensional geometrical model and adding texture for the geometric model. With the developing of photoelectric range finder technology, it has become reality that reconstructing the shape of an object by the 3D scanner. Using digital photos rebuilding objects texture is the key points of the texture mapping research.
The texture mapping based on digital photos is that using digital cameras gain actual object surface texture image, and then establish the mapping relationship between 3D model and texture image through perspective projection matrix, mapping the true texture to the surface of 3D model to enhance realism. This paper has a research on the key technologies of the texture mapping based on digital photos, including the following:
(1) Introduced the development situation of texture mapping technology at home and abroad, Recountted the theory knowledge related on texture mapping technology, including computer graphics, computer vision, camera calibration theory, image registration method and some other relevant knowledge:
(2) In order to solve the calibration of camera extrinsic parameters problem in texture mapping,this paper presents a geometric calibration method based on four point pairs.The first one selects three points in the world coordinate system and calculate their coordinates in the camera coordinate system(world points, light heart, and image point are collinear). Transforms the coordinates to a reference coordinate system and the euclidean distance is invariant. According to the homothetic triangle theory, calculates the rotation matrix and translation vectoer respectively. Finally, gets the complete extrinsic parameters..Applied in the texture mapping, this method can establish a mapping between 3D model and digital photo:
(3) The key points of the registration method based on 3D-2D feature points is finding the matching points between 3D model and 2D image. This paper adopts man-machine interactive way of selecting corresponding feature point, and puts forward a fast 3D pick-up method based on the perspective projection. It can pick up the three-dimensional space coordinates;
(4) Aiming at the perspective change of the texture images, after research the SIFT features registration algorithm, using Mahalanobis distance eliminate the mismatch points of the SIFT algorithm to improve the accuracy of image registration.
(1)分析了纹理映射技术在国内外的发展概况,介绍了与纹理映射技术相关的计算机图形学、计算机视觉、数码相机参数标定理论、图像配准方法等知识；
(2)提出了一种利用四组对应点求解相机外部参数的几何标定方法,利用人机交互的方法选取了4组对应点,分别求取了旋转矩阵和平移矩阵,得到了相机的全部外部参数,然后根据相机的内外参数信息求解了透视投影矩阵,建立了三维模型和数码照片之间的映射关系,实现了3D-2DD特征点的配准；
(3)基于3D-2D特征点的配准方法关键是确定三维模型和二维图片的3D-2D匹配点,采用了人机交互的方式选取对应特征点对的方法,并提出了一种基于透视投影的快速三维拾取方法,可以较为精确的拾取到模型的三维空间坐标；
(4)针对纹理图像存在视角变化的特点,在研究SIFT特征配准算法基础上,采用了马氏距离对SIFT算法误匹配点进行剔除,提高了图像配准的准确率。
Virtual reality is an important subject in the field of computer graphics and computer vision, and it has a wide range of applications in cultural relics preservation, special effects on movies, advertising, e-commerce and other fields. Usually,to the 3D reconstruction of real scenes, it will enhance realism through the method that adding texture for the 3D surface model. The process usually takes two steps constructing three-dimensional geometrical model and adding texture for the geometric model. With the developing of photoelectric range finder technology, it has become reality that reconstructing the shape of an object by the 3D scanner. Using digital photos rebuilding objects texture is the key points of the texture mapping research.
The texture mapping based on digital photos is that using digital cameras gain actual object surface texture image, and then establish the mapping relationship between 3D model and texture image through perspective projection matrix, mapping the true texture to the surface of 3D model to enhance realism. This paper has a research on the key technologies of the texture mapping based on digital photos, including the following:
(1) Introduced the development situation of texture mapping technology at home and abroad, Recountted the theory knowledge related on texture mapping technology, including computer graphics, computer vision, camera calibration theory, image registration method and some other relevant knowledge:
(2) In order to solve the calibration of camera extrinsic parameters problem in texture mapping,this paper presents a geometric calibration method based on four point pairs.The first one selects three points in the world coordinate system and calculate their coordinates in the camera coordinate system(world points, light heart, and image point are collinear). Transforms the coordinates to a reference coordinate system and the euclidean distance is invariant. According to the homothetic triangle theory, calculates the rotation matrix and translation vectoer respectively. Finally, gets the complete extrinsic parameters..Applied in the texture mapping, this method can establish a mapping between 3D model and digital photo:
(3) The key points of the registration method based on 3D-2D feature points is finding the matching points between 3D model and 2D image. This paper adopts man-machine interactive way of selecting corresponding feature point, and puts forward a fast 3D pick-up method based on the perspective projection. It can pick up the three-dimensional space coordinates;
(4) Aiming at the perspective change of the texture images, after research the SIFT features registration algorithm, using Mahalanobis distance eliminate the mismatch points of the SIFT algorithm to improve the accuracy of image registration.
引文
[1]张茂军,虚拟现实技术.(第一版)北京：科学出版社,2001
[2]虚拟现实系统简介.http://wenku.baidu.com/view/0ae1032de2bd960590c67708html
[3]陈任,鲁东明,潘云鹤.基于几何模型与照片序列的不规则物体纹理的获取.中国图象图形学报,2003,8(8)：903-906
[4]E Catmull. Subdivision algorithm for computer display of curved surfaces. PH. D Thesis, De. pt of Computer Science, Univ. of Utah, Salt Lake City, Utah,1974
[5]E. Catmul. A hidden-surface algorithm with anti-abasing. Computer Graphics,1978,12(3):6-11
[6]E. Catmull, A.R. Smith.3-D transformations of image in scanline order. Computer Graphics,1980,14(3):279-285
[7]Blinn. J. F. A scan line algorithm for the computer display of parametrically defined surfaces. Computer Graphics,1978,12(3)
[8]James F, Blinn. Simulation of Wrinkled Surface. Computer Graphics,1978, 12(2):286-292
[9]J. T.Kajiya. Anisotropic Reflection Models. Computer Graphics,1985, 19(3):15-21
[10]E. A. Bier, K. R. Sloan. Two-part Texture Mappings. IEEE Computer Graphics Application.1986,6:40-53
[11]Wolfgang Niew. Automatic reconstruction of 3D objects using a mobile camera. Image and Vision Computing,1999,17:125-134
[12]H. Busch. Automatic modeling of rigid 3D objects using an analysis by synthesis system. SPIE Proceedings Visual Communication and Image Processing IV, Philadelphia,1989:356-364
[13]Song-Yong Park, Murali Subbarao. Automat ic 3D model reconstrution based on novel pose estimation and integration techniques.Image and Vision Computing,2004,22:623-635
[14]王君秋,查红彬.结合兴趣点与边缘的建筑物与物体识别方法.计算机辅助设计与图形学学报,2006,18(8)：1257-1263
[15]孟放,查红斌.基于LOD控制与内外存调度的大型三维点云数据绘制.计算机辅助设计与图形学学报,2006,18(1)：18
[16]刘钢,彭群生,鲍虎军基于多幅实拍照片为真实景物模型添加纹理.软件学报,2005,16(11)：2014-2020
[17]戴立玲,丁立军,卢章平.三维实体表面纹理的模型重建.江苏大学学报,2006,27(5A)：21-24
[18]戴雯.基于文物的表面纹理重建技术研究.江苏大学硕士学位论文.2009
[19]B. Levy, J. L. Mallet. Non-distorted texture mapping for sheared triangulated meshes. Proceedings of the 25th annual conference on Computer graphics and interactive techniques SIGGRAPH'98, July 1998: 343-352
[20]刁常宇.三维模型实拍纹理高精度重建研究.浙江大学博士学位论文.2007
[21]E.Bier, K. Sloan. Two-part texture mapping. IEEE Computer Graphics and Applications.1986:40-53
[22]A. W. F. Lee, W. Sweldens, P. Schroder, L. Cowsar, D. Dobkin. Maps:multires-olution adaptive parameterization of surfaces. Computer Graphics (Siygraph Conf Proc).July 1998:95-104
[23]Haralick R M, Joo H Lee C. Pose estimation from corresponding point data. IEEE Transactions on Systems. Man and Cybernetics,1989:1426-1445
[24]Rocchini C, Cignoni P, Montani C. Multiple textures stitching and blending on 3D objects.Proceedings of Eurographics Workshop on Rendering. Granada, Spain,1999:173-180
[25]Zha Hongbin, Wang Peng. Realistic face modeling by registration of a 3-D mesh model and mufti-view color images. Proceedings of 8th International Conference on CAD/Graphics. Macao, China,2003:217-222
[26]Matsushita K, Kaneko T.Efficient and handy texture mapping on 3D surfaces Computer Graphcs Forum.1999,18(3):349-358
[27]Neugebauer P J, Kiein K. Texturing 3D models of real world objects from multiple unregistered photographic views. Computer Graphics Forum.1999,18(3):245-256
[28]Lensch H, Heidrich W, Seidel H P. Automated texture registration and stitching for real world models. Proceedings of Pacific Graphics. Hong Kong, China,2000:317-326
[29]David A. Forsyth, Jean Ponce.林学訚,王宏等译.计算机视觉--一种现在方法.北京：电子工业出版社.2004
[30]Abdel-Aziz Y I, Karara H M. Direct linear transformation into object space coordinates in close-range photo grammetry. Proceedings of the Symposium on Close-Range Photorgammetry. Urbana, Illinois, USA,1971: 1-18
[31]魏冲.基于数码相机图像的三维重建技术研究.中北大学硕士学位论文.2009
[32]Tsai RY. An efficient and accurate camera calibration technique for 3d machine vision. Proceedings of IEEE Conference on Computer Vision and Pattern Recognition. Miami, USA,1986:364-374
[33]Tasi R Y. A versatile camera calibration technique for high-aeeuracy 3D machine vision metrology using of-the-shelf TV cameras and lenses. IEEE Joumal of Robotics and Automation.1987,3(4):323-344
[34]郑志刚.高精度摄像机标定和鲁棒立体匹配算法研究.中国科学技术大学博士学位论文.2008
[35]Zhengyou Zhang. A Flexible New Technique for Camera Calibration. Microsoft Research, One Microsoft Way, Redmond,98052-6399, USA
[36]王玉柱.基于双目视觉的钢轨磨耗测量系统关键技术的研究与实现.北京交通大学硕士学位论文.2009
[37]王立中.基于机器视觉的奶牛体型评定中的关键技术研究.内蒙古农业大学 博士学位论文.2009
[38]姜大志,孙闵等.数码相机标定方法研究.南京航空航天大学学报2001,33(1)：55-59
[39]刘力强,周明全,耿国华.一种平行透视下的三维拾取方法.西北大学学报(自然科学版).2002,32(1)：39-42
[40]姚继权,李晓豁.计算机图形学人机交中三维拾取方法的研究.工程设计学报.2006.13(2)：116-120
[41]刘卫光.图像信息融合与识别.北京：电子工业出版社.2008
[42]周亚平.图像配准技术研究.西安电子科技大学硕士学位论文.2008
[43]程龙,郭立,史鸿声.一种纹理映射算法的FPGA实现.小型微型计算机系统.2009,30(9)：1855-1859
[44]梅向明.高等几何(第二版).北京：高等教育出版社.2000
[45]Philip J.Schneider, David H. Eberly编著.周长发译.计算机图形学几何工具算法详解.北京：电子工业出版社
[46]廖斌.基于特征点的图像配准技术研究.国防科学技术大学博士学位论文.2008
[47]陈裕.基于SIFT算法的无人机遥感图像配准.中南大学硕士学位论文.2009
[48]范金城,梅长林.数据分析.北京：科学出版社.2005.160-161
[49]OpenGl Programming Guide编著.李军,徐波等译. OpenGL编程指南(第7版).北京：机械工业出版社.2010
[50]Richard S.Wright, Michael Sweet编著.潇湘工作室译. OpenGL超级宝典.第二版.北京：人民邮电出版社.2001
[2]虚拟现实系统简介.http://wenku.baidu.com/view/0ae1032de2bd960590c67708html
[3]陈任,鲁东明,潘云鹤.基于几何模型与照片序列的不规则物体纹理的获取.中国图象图形学报,2003,8(8)：903-906
[4]E Catmull. Subdivision algorithm for computer display of curved surfaces. PH. D Thesis, De. pt of Computer Science, Univ. of Utah, Salt Lake City, Utah,1974
[5]E. Catmul. A hidden-surface algorithm with anti-abasing. Computer Graphics,1978,12(3):6-11
[6]E. Catmull, A.R. Smith.3-D transformations of image in scanline order. Computer Graphics,1980,14(3):279-285
[7]Blinn. J. F. A scan line algorithm for the computer display of parametrically defined surfaces. Computer Graphics,1978,12(3)
[8]James F, Blinn. Simulation of Wrinkled Surface. Computer Graphics,1978, 12(2):286-292
[9]J. T.Kajiya. Anisotropic Reflection Models. Computer Graphics,1985, 19(3):15-21
[10]E. A. Bier, K. R. Sloan. Two-part Texture Mappings. IEEE Computer Graphics Application.1986,6:40-53
[11]Wolfgang Niew. Automatic reconstruction of 3D objects using a mobile camera. Image and Vision Computing,1999,17:125-134
[12]H. Busch. Automatic modeling of rigid 3D objects using an analysis by synthesis system. SPIE Proceedings Visual Communication and Image Processing IV, Philadelphia,1989:356-364
[13]Song-Yong Park, Murali Subbarao. Automat ic 3D model reconstrution based on novel pose estimation and integration techniques.Image and Vision Computing,2004,22:623-635
[14]王君秋,查红彬.结合兴趣点与边缘的建筑物与物体识别方法.计算机辅助设计与图形学学报,2006,18(8)：1257-1263
[15]孟放,查红斌.基于LOD控制与内外存调度的大型三维点云数据绘制.计算机辅助设计与图形学学报,2006,18(1)：18
[16]刘钢,彭群生,鲍虎军基于多幅实拍照片为真实景物模型添加纹理.软件学报,2005,16(11)：2014-2020
[17]戴立玲,丁立军,卢章平.三维实体表面纹理的模型重建.江苏大学学报,2006,27(5A)：21-24
[18]戴雯.基于文物的表面纹理重建技术研究.江苏大学硕士学位论文.2009
[19]B. Levy, J. L. Mallet. Non-distorted texture mapping for sheared triangulated meshes. Proceedings of the 25th annual conference on Computer graphics and interactive techniques SIGGRAPH'98, July 1998: 343-352
[20]刁常宇.三维模型实拍纹理高精度重建研究.浙江大学博士学位论文.2007
[21]E.Bier, K. Sloan. Two-part texture mapping. IEEE Computer Graphics and Applications.1986:40-53
[22]A. W. F. Lee, W. Sweldens, P. Schroder, L. Cowsar, D. Dobkin. Maps:multires-olution adaptive parameterization of surfaces. Computer Graphics (Siygraph Conf Proc).July 1998:95-104
[23]Haralick R M, Joo H Lee C. Pose estimation from corresponding point data. IEEE Transactions on Systems. Man and Cybernetics,1989:1426-1445
[24]Rocchini C, Cignoni P, Montani C. Multiple textures stitching and blending on 3D objects.Proceedings of Eurographics Workshop on Rendering. Granada, Spain,1999:173-180
[25]Zha Hongbin, Wang Peng. Realistic face modeling by registration of a 3-D mesh model and mufti-view color images. Proceedings of 8th International Conference on CAD/Graphics. Macao, China,2003:217-222
[26]Matsushita K, Kaneko T.Efficient and handy texture mapping on 3D surfaces Computer Graphcs Forum.1999,18(3):349-358
[27]Neugebauer P J, Kiein K. Texturing 3D models of real world objects from multiple unregistered photographic views. Computer Graphics Forum.1999,18(3):245-256
[28]Lensch H, Heidrich W, Seidel H P. Automated texture registration and stitching for real world models. Proceedings of Pacific Graphics. Hong Kong, China,2000:317-326
[29]David A. Forsyth, Jean Ponce.林学訚,王宏等译.计算机视觉--一种现在方法.北京：电子工业出版社.2004
[30]Abdel-Aziz Y I, Karara H M. Direct linear transformation into object space coordinates in close-range photo grammetry. Proceedings of the Symposium on Close-Range Photorgammetry. Urbana, Illinois, USA,1971: 1-18
[31]魏冲.基于数码相机图像的三维重建技术研究.中北大学硕士学位论文.2009
[32]Tsai RY. An efficient and accurate camera calibration technique for 3d machine vision. Proceedings of IEEE Conference on Computer Vision and Pattern Recognition. Miami, USA,1986:364-374
[33]Tasi R Y. A versatile camera calibration technique for high-aeeuracy 3D machine vision metrology using of-the-shelf TV cameras and lenses. IEEE Joumal of Robotics and Automation.1987,3(4):323-344
[34]郑志刚.高精度摄像机标定和鲁棒立体匹配算法研究.中国科学技术大学博士学位论文.2008
[35]Zhengyou Zhang. A Flexible New Technique for Camera Calibration. Microsoft Research, One Microsoft Way, Redmond,98052-6399, USA
[36]王玉柱.基于双目视觉的钢轨磨耗测量系统关键技术的研究与实现.北京交通大学硕士学位论文.2009
[37]王立中.基于机器视觉的奶牛体型评定中的关键技术研究.内蒙古农业大学 博士学位论文.2009
[38]姜大志,孙闵等.数码相机标定方法研究.南京航空航天大学学报2001,33(1)：55-59
[39]刘力强,周明全,耿国华.一种平行透视下的三维拾取方法.西北大学学报(自然科学版).2002,32(1)：39-42
[40]姚继权,李晓豁.计算机图形学人机交中三维拾取方法的研究.工程设计学报.2006.13(2)：116-120
[41]刘卫光.图像信息融合与识别.北京：电子工业出版社.2008
[42]周亚平.图像配准技术研究.西安电子科技大学硕士学位论文.2008
[43]程龙,郭立,史鸿声.一种纹理映射算法的FPGA实现.小型微型计算机系统.2009,30(9)：1855-1859
[44]梅向明.高等几何(第二版).北京：高等教育出版社.2000
[45]Philip J.Schneider, David H. Eberly编著.周长发译.计算机图形学几何工具算法详解.北京：电子工业出版社
[46]廖斌.基于特征点的图像配准技术研究.国防科学技术大学博士学位论文.2008
[47]陈裕.基于SIFT算法的无人机遥感图像配准.中南大学硕士学位论文.2009
[48]范金城,梅长林.数据分析.北京：科学出版社.2005.160-161
[49]OpenGl Programming Guide编著.李军,徐波等译. OpenGL编程指南(第7版).北京：机械工业出版社.2010
[50]Richard S.Wright, Michael Sweet编著.潇湘工作室译. OpenGL超级宝典.第二版.北京：人民邮电出版社.2001