摘要
三维重构是计算机视觉技术研究的主要内容之一,在工程技术和其它很多领域,一般要对物体进行三维分析,从而得到对研究有用的信息。本文研究的是基于二维图像的三维重构技术,是近年来计算机视觉技术与计算机图形学技术相结合而产生的一门新技术。
基于二维图像的三维重构技术的研究具有重要的现实意义,而且具有广阔的应用范围,包括考古学、建筑学、以及材料加工、工业检测以及医疗成像设备等领域。目前,国内外对于大型铸件表面缺陷的监测尚无特别好的办法,几乎都是通过肉眼观察的方式来对缺陷进行判断;这种判断方式不仅费时、费力,而且很难对表面缺陷的所在的具体位置进行三维定位。针对这些问题,本文创造性地提出了一种表面缺陷检测方法:采用计算机视觉的最新技术来对缺陷进行自动检测。本方法的原理是:采用CMOS摄像机对大型铸件表面进行多视角成像,然后对成像得到的照片进行处理,得到需要的信息,最后将三维重构得到的几何实体与大型铸件自身的三维几何实体进行自动对比。这样就可以对大型铸件的表面缺陷准确地进行几何定位以及定量判断。
本文在分析了摄像机成像模型及其工作原理的基础上,提出了空间点重构的方案,得到了世界坐标系与像素坐标系之间的转换关系式,为摄像机标定提供了准确的数学模型。
特征点的提取是三维重构的前提和基础,在摄像机标定之前,本文采用加权平均值法对图像进行了灰度化处理,并利用Harris角点检测法和高斯滤波函数得到了图像中的角点。
在对摄像机进行标定的时候,没有考虑二维图片中的非线性畸变,从图片中选取10个角点,运用OPenCV中对矩阵进行处理的函数,结合VC++编程技术,采取了线性标定的方法得到了摄像机的内部参数和外部参数的综合矩阵——摄像机的投影矩阵,并在文中给出了详细的理论推导过程。
最后,本文在VC++平台上初步开发了一个基于两幅图像的空间点的重构系统,实现了图像读取、灰度化处理、角点提取、摄像机的标定和空间点重构的功能。同时用实验验证了该系统的准确性。
3D Reconstruction is one of the main research content of the computer vision technology. In engineering and many other areas, in order to get useful information for research, we usually analyze the 3D structure of the objects. In this paper, 3D reconstruction is based on 2D images, it is a new technology which comes from the combination of the technology of computer vision and computer graphics.
The research on 3D reconstruction has important practical significance and has a broad range of applications, such as archeology、architecture、materials processing、industrial inspection、medical imaging equipment industry.
At present, there is no effective method to monitor the surface defects of the large casting, researchers always determine defects by naked eyes; this method wastes time and energy, and it is very hard to locate the location of surface defects. To solve these problems, the paper proposes a creative method to defect the surface detection: using the latest technologies of computer vision to detect defects automatically. The principle of this method is: use CMOS camera to obtain the multi-angle images of the surface of large casting; then, process images to get the useful information; finally, the geometric entities of 3D reconstruction and the original geometric entities casting will be contrasted automatically. In this way, we can get the exact location of the surface defects of large-scale casting.
Based on the analysis of the camera imaging model and its working principle, the paper gives a plan for space point reconstruction, the paper also studies the conversion relationship between the world coordinate system and pixel coordinate system. Extraction of feature points is the prerequisite conditions and basis of the camera calibration. Before calibration, this paper uses the weighted average to process the image, uses the Harris corner detection method and the Gaussian filter function to get the corner from the image.
The paper does not consider the non-linear distortion of the 2D image during the camera calibration. It selects 10 points from the corner points, and obtains the camera projection matrix by means of OpenCV and VC++ programming techniques. Theory analysis and process design are made detailed in paper.
Finally, the paper develops a software system on the VC++ platform. It includes reading images, gray-scale processing, corner extraction, camera calibration and reconstruction of the space point. The validity and practicability of studies has been approved by the experiment in the dissertation.
引文
[1]尚名姝,解凯.几种三位重构方法的比较[J].哈尔滨师范大学自然科学学报. 2003, 19(5):43~46
[2]蔡钦涛.基于图片的三位重建技术研究: [硕士学位论文],浙江:浙江大学,2004
[3]王斌,王知衍.计算机视觉中的三维重构建模[J].微计算机信息, 2005, 21(12-2):118~120
[4]高文,陈熙霖.计算机视觉-算法与系统原理[M].第一版.北京:清华大学出版社, 1999
[5]徐智华,袁东斌.计算机视觉中实现图像三维重构的方案[J].福建广播电视大学学报, 2008, 3(69):68~70
[6] Linda G.Shapiro, George C.Stockman. Computer Vision.中译本:计算机视觉[M].赵清杰,钱芳,蔡利栋译,机械工业出版社. 2005
[7]赵颜利.计算机视觉三维重构若干技术研究: [博士学位论文].南京:南京理工大学, 2007
[8]高文,陈熙霖.计算机视觉[M].北京:清华大学出版社, 1999
[9]姜大志.计算机视觉中三维重构的研究与应用: [博士学位论文].南京:南京航空航天大学, 2001
[10] D.Marr, Vision. W.H.Freemen and Company, San Francisco, 1982.中译本:视觉计算理论[M],姚国正等译,科学出版社, 1988
[11] R.Hartley, A.Zisserman, Multiple View Geometry in Computer Vision[M]. Cambridge University Press, 2000
[12] Drescher, Nagal. On the selection of critical points and local curvature extrema of region boundaries for interframe matching[J]. International Conference on Pattern Recognition, 1982, 542~544
[13] S.M.Smith, M.Brady. SUSAN-a new approach to low level image processing[J].Journal of Computer Vision, 1997, 23(1): 45~78
[14] Kitchen , A.Rosenfeld. Gray-Level Corner Detection[J]. Pattern Recognition Letters, 1982, Vol.1: 95~102
[15] C.Harris, M.Stephenso. A Combined Corner and Edge etectoro Proceedings of the Alvey Vision Conference[J]. 1988, 189~192
[16] U.R.Dhond, J.K.Aggarwal. Structure from Stereo: A Review. IEEE Trans on System, Man,and Cybernetics. 1989, 19(6): 1489~1510
[17] P.Beardsley, p.Tort, A.Zisserma. 3D model acquisition from extended image sequences[J]. In Processings of 4th European conference on Computer Vision, LNCS 1065, Cambrage, 1996,683~695
[18] Pritchett, Zisserman. A matching and Reconstruction from Widely Separated Views[J]. Lecture Notes in Computer Science 150, Springer-Verlag, 1998, 219~224
[19] M.Pilu. A Direct Method for Stereo Correspondence based on Singular Value Decomposition[J]. In Processings of Computer Vision and Pattern Recognition Puerto Rico, 1997, 445~453
[20] D.Burschka, M.Z.Brown, G.D.Hager. Advances in computional stereo.IEEE Trans. On pattern Analysis and Machine Intelligence, 2003,25(8):993~1008
[21] D.Scharstein, R.Szeliski. A taxonomy and evalution of dense two-frame stereo correspondence algorithms[J]. Internation Journal of Computer Vision, 2002, 47(1):7~42
[22] C.Schmid, A.Zisserman, A.Fitzgibbon. Automatic Line Matching and 3D Reconstruction of Buildings from Mutiple Views. ISPRS Conference on Automatic Extraction of GIS Objects from Digital Imagery, 1999, 32(3): 69~80
[23] R.Hartley, A.Zisserman. Multiple View Geometry in Computer Vision[M]. Cambridge University Press. Cambridge, United Kingdom,2000
[24] O.Faugeras, Q.T.Luong. The Geometry of Multiple Images[M]. The MIT Press Cambridge, Massachusetts,2001
[25] C.Tomasi, T.Kanade. Shape and motion from image streams under orthography: A factorization method. International Journal of computer Vision, 1992, 9(2): 137~154
[26] S.Bougnoux. From Projective to Ruclidean Space under any practical situation, a criticism of self-calibration, proc. International Conference on Computer Vision,Narosa Publishing house, New Delhi/Madras/Bombay/Calcutta/London, pp. 790~796,1998
[27] S.bougnoux, L.Robert. Totalcalib: a fast and reliable system for off-line calibration of images sequences, DEMO Session,CVP’97,1997
[28] P.Debevec, C.Taylor, T.Malik. Modeling and rendering architecture from photographs: A hybrid geometry and image based approach, Siggraph, 1996
[29] H.Y.Shum, R.Szeliski, S.Baker, M.Han, P.Anandan. Interative 3D modeling from multiple images using scene regularities[M]. SMILE, 1998
[30] O.Faugeras, L.Robert, S.Laveau. 3D reconstruction of urban scenes from image sequences. CVGIP, 1998
[31] M.Pollefeys. Self-calibration and metric 3D reconstruction from uncalibrated image sequences: [ph.d thesis], Katholieke Universiteit Leuven, 1999
[32]姜大志,孙闵,刘淼等.数码相机标定方法研究[J].南京航空航天大学学报, 2001,33(1): 55~59
[33]汪伟,李郝林.基于数码图片的三维重构.上海理工大学学报, 2005, 27(5): 429~432
[34]吴凤和,张晓峰,施法中.基于单幅图片数据的三维重构方法研究[J].中国机械工程, 2007, 18(17): 2071~2075
[35]章勤,虢伟.基于多幅深度图像的三位重构技术[J].华中科技大学学报, 2008, 36(2): 88~91
[36]舒远,谈正,丁礼儒.利用空间正交约束的相机自标定和三维重建[J].西安交通大学学报, 2005, 39(2): 139~141
[37]吴海峰,朱文文.基于径向畸变的数码相机及三维重构[J].机电一体化, 2006, 5: 65~66
[38]唐泽圣,周嘉玉.计算机图形学基础[M].清华大学出版社. 1995
[39]周长发.Visual C++.net图象处理编程[M].北京:电子工业出版社,2002
[40] C.Harris, M.Stephens. A combined corer and edge detector[J]. In Alvey Vision Conference, 1988. pp147~151
[41]冯军.三维目标立体信息恢复与建模: [硕士学位论文].武汉:华中科技大学,2006
[42]王保平,朱智林,李宗领.一种基于区域信息的图像高斯噪声滤除方法[J].兰州大学学报(自然科学版),2003 (05):44~47
[43]李玉山.数字视频技术[J].西安:西安电子科技大学出版社, 2006:169~181
[44]郑南宁.计算机视觉与模式识别[M].北京:北京工业出版社, 1998
[45]马颂德,张正友.计算机视觉[M].北京:科学出版社, 1998
