基于数字图像处理方法的三维测量
|
摘要
计算机视觉的研究目标是使计算机具有通过一幅或多幅图像认知周围环境信息的能力。基于图像的测量技术是从计算机视觉领域中发展起来的新型非接触测量技术，它主要是以图像为载体来检测和传递信息的测量方法，通过提取图像的特征信息，从图像中萃取被测对象的深度信息。基于数字图像的测量方法具有精度高、稳定性好、非接触性测量等特点，在精度、速度、智能化等方面具有较强的适应性。
采用数码相机拍摄驾驶室内物体的图像，对常用数字图像处理方法进行分析比较，得到适合于本文的图像处理方法，使得处理完后的边缘图像能够容易提取空间点在图像上的投影坐标。
比较了不同数量和分布图形的控制点情况下的简单DLT(直接线性变换)定标精度，选取定标精度好的控制点分布图形，采用DLT定标算法直接求取定标矩阵的11个线性定标参数，然后以11个线性定标参数为初值，用迭代的方法来解算含有镜头畸变参数(k_1，k_2，k_3)的非线性模型摄像机定标参数，并利用一些额外未曾参与平差的控制点残差参与误差拟合来提高定标矩阵参数的精度，使这些额外控制点的测量值误差达到最小值，减少被测物体的三维测量误差。
根据立体视觉的基本原理，以汽车驾驶室内手伸及界面部分参数为实验对象，进行三维重建，确定待测物体的三维信息及物体的相对位置，实验验证了基于图像处理的三维测量方法可行性。本文优化了定标算法，提高了测量精度。
The research purpose of computer vision is to make computer perceive surroundings information. The image-based measurement technique is one kind of novel noncontact measurement technique, which develops from the computer vision realm and primarily regards images as the carrier and means to detect or transfer the information. The image-based measurement technique can bring us the depth information of the examined object after dealing with the characters detected from the images. Moreover, this measurement technique has very strong adaptability in the aspects of accuracy, speed, intelligence and so on.
Use the digital camera to take the drier's cab images, analyze and compare the common digital image transformation technique, find the digital image transformation which are fit for textual, then can easily to take the point coordinate from the edge image.
Compare the calibration precision of simple Direct Linear Transformation in the case of different quantity and distributing figure reference point, choose good reference point of calibration precision distributing figure. Adopt DLT calibration to work out eleven linearity calibration parameters. Take eleven linearity parameters as initial value, and resolving nonlinearity model camera calibration parameter which contain distorted lens parameter (k_1, k_2, k_3) with the iteration method. Using some extra not participation DLT reference point's residual to participation error fitting, and then improves calibration precision. Make these measured value error of the additional reference point to achieve the minimum value with the optimization algorithm, reduces survey error of object three dimensional.
According to the basic principle of stereoscopic vision, with parts of parameters in the car drive for test object, carry on a three dimension reconstruction, get the three dimension information of the object and the opposite position of the object. The experiment verified a method which based on digital image transformation of three dimension measurement possibility. This text optimized the calibration method and raising survey precision.
采用数码相机拍摄驾驶室内物体的图像，对常用数字图像处理方法进行分析比较，得到适合于本文的图像处理方法，使得处理完后的边缘图像能够容易提取空间点在图像上的投影坐标。
比较了不同数量和分布图形的控制点情况下的简单DLT(直接线性变换)定标精度，选取定标精度好的控制点分布图形，采用DLT定标算法直接求取定标矩阵的11个线性定标参数，然后以11个线性定标参数为初值，用迭代的方法来解算含有镜头畸变参数(k_1，k_2，k_3)的非线性模型摄像机定标参数，并利用一些额外未曾参与平差的控制点残差参与误差拟合来提高定标矩阵参数的精度，使这些额外控制点的测量值误差达到最小值，减少被测物体的三维测量误差。
根据立体视觉的基本原理，以汽车驾驶室内手伸及界面部分参数为实验对象，进行三维重建，确定待测物体的三维信息及物体的相对位置，实验验证了基于图像处理的三维测量方法可行性。本文优化了定标算法，提高了测量精度。
The research purpose of computer vision is to make computer perceive surroundings information. The image-based measurement technique is one kind of novel noncontact measurement technique, which develops from the computer vision realm and primarily regards images as the carrier and means to detect or transfer the information. The image-based measurement technique can bring us the depth information of the examined object after dealing with the characters detected from the images. Moreover, this measurement technique has very strong adaptability in the aspects of accuracy, speed, intelligence and so on.
Use the digital camera to take the drier's cab images, analyze and compare the common digital image transformation technique, find the digital image transformation which are fit for textual, then can easily to take the point coordinate from the edge image.
Compare the calibration precision of simple Direct Linear Transformation in the case of different quantity and distributing figure reference point, choose good reference point of calibration precision distributing figure. Adopt DLT calibration to work out eleven linearity calibration parameters. Take eleven linearity parameters as initial value, and resolving nonlinearity model camera calibration parameter which contain distorted lens parameter (k_1, k_2, k_3) with the iteration method. Using some extra not participation DLT reference point's residual to participation error fitting, and then improves calibration precision. Make these measured value error of the additional reference point to achieve the minimum value with the optimization algorithm, reduces survey error of object three dimensional.
According to the basic principle of stereoscopic vision, with parts of parameters in the car drive for test object, carry on a three dimension reconstruction, get the three dimension information of the object and the opposite position of the object. The experiment verified a method which based on digital image transformation of three dimension measurement possibility. This text optimized the calibration method and raising survey precision.
引文
[1] 马颂德，张正友．计算机视觉-计算机理论与算法基础[M]．北京：科学出版社，1997．
[2] D. Marr, Vision, San Francisco: W. H. Freeman and Company, 1982.
[3] 韦穗等．计算机视觉中的多视图几何[M]．合肥：安徽大学出版社，2002．
[4] 蔡祥鹏，林建，张翔．光学三维信息测量技术应用的现状与发展[J]．机电技术，2006(增刊)：12-17．
[5] E. Davies Machine Vision: Theory, Algorithms and Practicalities Academic Press, 1990.
[6] Shah Y C, et al. A new technique to extract range information from stereo images [J]. IEEE Trans on Pattern Analysis and Machine Intelligence. 1989, Vol. 11, No. 7: 768-773.
[7] Ensj. Et al. An investigation of methods for determining depth from focus[J]. IEEE Trans on Pattern Analysis and Machine Intelligence 1993, Vol. 15, No. 2: 97-108.
[8] Meadows DM, et al. Generation of surface contours by mo ire pat terns. Applied Optics, 1990, 9(9): 1467.
[9] Macy WW Jr. Two-dimensional fringe pattern analysis. Applied Optics, 1983, 22(23): 3898-3901.
[10] Srinivasan V, Liu H C, et al. Automated phase measuring profilometry of 3-D diffuse object. Appl. Opt, 1984, 23(18): 3105～3108.
[11] PiroddaL. Shadow and project ion moire techniques for absolute or relative mapping of surface shapes. Optical Engineering, 1982, 21(4): 640-649.
[12] Idesaw aM, Yatagai T, Soma T. Scanning moire method and automatic measurement of 3-D shapes. Applied Optics, 1977, 16(8): 2152-2161.
[13] Cline HE, Lo rensen WE, Holik AS. Automatic moire contouring, Applied Optics, 1984, 23(10): 1454-1459.
[14] Indebetouw G, Profile measurement using project ion of running fringes. Applied Optics, 1978, 17(18): 2930-2933.
[15] Blatt JH, Hooker JA. 3-D inspect ion of large objects by Moire profilometry. 1989, L. I. A 70 ICAL EO: 12-16.
[16] 吴祖斌，王颖，曹士英等．用改进的光谱相位相干直接电场重构法装置测量飞秒激光脉冲的相位[J]．中国激光，2006，331)：21-25．
[17] 陈锋．动态三维面形测量中相位展开问题的研究[J]．四川大学硕士论文，2005．
[18] 翁嘉文，钟金钢．小波变换轮廓术在双频光栅条纹中的应用[J]．光学学报，2005，25(5)：603-608．
[19] Srinivasan V, Liu H C, etal.. Automated phase measuring profilometry of 3-D diffuse object. Appl. Opt, 1984, 23(18): 3105-3108.
[20] 严明，高志山．移相法测量波片的相位延迟量[J]．光电子．激光，2005，16(2)：183-187．
[21] Creath K, Schm it J. N-point spatial phase-measurement techniques for non-destructive testing. Opt. Lasers Eng, 1996, 24: 365-379.
[22] FBlsis. Review of 20 years of range sensor development [A]. Proceedings of SP IEEE IS&T Electronic Imaging[C]. SPIE 2003, Vol. 5013.
[23] 陶涛，郭红卫，何海涛．镜面反射面形光学三维测量技术综述[J]．测量仪器，2005，27(2)：90-95．
[24] H O Saldner, J M Huntley. Profilometry by temporal phase unwrapping and spatial light modulator based fringe projector [J]. Opt. Eng., 1997, 36 (2): 610-615.
[25] SeulinR, Merienne F, Gorria P. Dynamic lighting system for seculars surface inspection [J]. SPIE, 2001, 4301: 199-2061
[26] 吴立德．计算机视觉[M]．复旦大学出版社，上海，1993．
[27] Murali Subbaro, Tae Choi. Accurate recovery of three dimensional shape from image focus [J]IEEE rrans on PAMI, 1995, 17(3): 266-274.
[28] 祝世平，强锡富．工件特征点三维坐标视觉测量方法综述[J]．光学精密工程，2000，8(2)：192-197．
[29] Indebetouw G, Profile measurement using project ion of running fringes. Applied Optics, 1978, 17(18): 2930-2933.
[30] 陈棣湘，罗飞路，潘孟春．立体视觉测量中的图像匹配策略研究[J]．光学技术，2002，28(5)：9．
[31] 任重，邵军力．双目立体视觉的匹配方法研究[J]．军事通信技术，2001，22(2)：45-48，72．
[32] S.M. Smith. SUSAN-A New Approach to Low Level Image Processing. Technical Report TR95SMSIc, 1995.
[33] S. M. Smith and J. M. Brady, SUSAN-a new approach to low level image processing [J] Int. Journal of Computer Vision, May 1997, Vol. 23, No. 1:45-78.
[34] 龙兴明．三维测量中的图像处理技术研究．重庆师范大学硕士论文，2005．
[35] 康文炜．基于图像确定空间坐标方法的研究．吉林大学硕士论文，2004．
[36] 姚敏等．数字图像处理[M]．北京：机械工业出版社，2006．
[37] Rosenfeld A, Kak A C. Digital Picture Processing. 2nd Edition. New York: Academic Press, 1982.
[38] 陈书海，傅录祥著．实用数字图像处理[M]．北京：科学出版社，2005．1：139-160．
[39] William K．Pratt著，邓鲁华，张延恒等译．数字图像处理[M]．北京：机械工业出版社，2005．
[40] G. P. Dineen, "Programming Pattern Recognition," Proc. Western Joint Computer Conference, March 1955, 94-100.
[41] R. E. Graham, "Snow Removal: A Noise Stripping Process for Picture Signals," IRE Trans. Information Theory, IT8, 1, February 1962, 129-144.
[42] A. Rosenfeld, C. M. Park, and J. P. Strong, "Noise Cleaning in Digital Pictures," Proc. EASCON Convention Record, October 1969, 264-273.
[43] R. Nathan, "Spatial Frequency Filtering," in Picture Processing and Psychohistories, B. S. Lipkin and A. Rosenfeld, Eds., Academic Press, New York, 1970, 151-164.
[44] J. W. Tukey, Exploratory Data Analysis, Addison-Wesley, Reading, MA, 1971.
[45] Linda G．Shapiro，George C．Stockman著，赵清杰等译．计算机视觉[M]．北京：机械工业出版社，2004．
[46] 史长琼，易昂．基于Canny准则的颗粒图像边缘检测算法[J]．计算机工程与科学．2004，26(3)：52-54．
[47] 张冬生．基于计算机视觉的三维测量技术．华北电力大学硕士论文，2003．
[48] 苗红杰，赵文吉，刘先林．数码相机检校和摄像测量的部分问题探讨[J]．首都师范大学学报，2005，26(1)：117-120．
[49] 杨国强．数字近景摄像测量系统研究．西安科技大学硕士论文，2005．
[50] 刘曙光，刘明远等．机器视觉机器应用[J]．机械制造，2000，Vol．38(431)：20-23．
[51] 钟玉琢，乔炳新等．机器人视觉技术[M]．国防工业出版社，北京，1994．
[52] 刘成君，戴汝为．计算机视觉的进展[J]．模式识别与人工智能，1995，Vol．8增刊．
[53] 王天珍．计算机视觉研究进展[J]．武汉汽车工业大学学报，1998，20(1)，23-27．
[54] 贾云德．机器视觉[M]．科学出版社，北京，2000．
[55] 于洪川，吴福朝等．基于主动视觉的摄像机自标定技术[J]．机器人，1999，1．
[56] 张毓晋．图像处理和分析[M]．北京：清华大学出版社，1999．3：43-49．
[57] Heath, M. 1997. Sciecntific Computing: An Introductory Survey. McGraw-Hill, Inc., New York.
[58] J. Weng, P. Cohen and M. Herniou, Calibration of Stereo Cameras Using a Non-Linear Distortion Model, Proc, International Conference on Pattern Recognition, pp. 246-253, 1990.
[59] Tsai R Y. An efficient and accurate camera calibration technique for3-D machine vision [A]. In Proc IEEC Conf. On Computer Vision and Pattern Recognition[C], Miami Beach: 1986, 364-374.
[60] O. D. Faugeras and G. Toscani, Camera Calibration for 3D Computer Vision, Proc. of International Workshop on Industrial Application of Machine Vision an Machine Intelligence, pp. 240-247, Japan, 1987.
[61] 张靖瑜，董明得，吕乃光，邓文怡．用广义逆法解决摄影测量系统的非线性问题[J]．北京机械工业学院学报．2001，16(1)：36-39．
[62] 伍雪冬．计算机视觉中摄像机定标及位姿和运动估计方法的研究．湖南大学博士论文，2004．
[63] 范洪．3D计算机视觉系统的高精度定标．北京交通大学博士论文，1993
[64] 周婷婷．基于图像的三维测量．安徽大学硕士论文，2005．8
[65] 丁丽娟．数值计算方法[M]．北京理工大学出版社，2004．
[66] 张翔．优化设计方法及编程[M]．中国农业大学出版社，2000．
[67] Francesco Balena著，李珂，张文波，赵明等译，Visual Studio．NET产品组审校．Visual Basic．NET技术内幕，2003．
[68] 管业鹏，童林夙，尹涵春．基于数码相机的三维物体空间几何位置的摄像测量[J]．电子学报．2002，30(6)：849-852．
[69] 林成森．数值分析[M]．科学出版社，2002．
[2] D. Marr, Vision, San Francisco: W. H. Freeman and Company, 1982.
[3] 韦穗等．计算机视觉中的多视图几何[M]．合肥：安徽大学出版社，2002．
[4] 蔡祥鹏，林建，张翔．光学三维信息测量技术应用的现状与发展[J]．机电技术，2006(增刊)：12-17．
[5] E. Davies Machine Vision: Theory, Algorithms and Practicalities Academic Press, 1990.
[6] Shah Y C, et al. A new technique to extract range information from stereo images [J]. IEEE Trans on Pattern Analysis and Machine Intelligence. 1989, Vol. 11, No. 7: 768-773.
[7] Ensj. Et al. An investigation of methods for determining depth from focus[J]. IEEE Trans on Pattern Analysis and Machine Intelligence 1993, Vol. 15, No. 2: 97-108.
[8] Meadows DM, et al. Generation of surface contours by mo ire pat terns. Applied Optics, 1990, 9(9): 1467.
[9] Macy WW Jr. Two-dimensional fringe pattern analysis. Applied Optics, 1983, 22(23): 3898-3901.
[10] Srinivasan V, Liu H C, et al. Automated phase measuring profilometry of 3-D diffuse object. Appl. Opt, 1984, 23(18): 3105～3108.
[11] PiroddaL. Shadow and project ion moire techniques for absolute or relative mapping of surface shapes. Optical Engineering, 1982, 21(4): 640-649.
[12] Idesaw aM, Yatagai T, Soma T. Scanning moire method and automatic measurement of 3-D shapes. Applied Optics, 1977, 16(8): 2152-2161.
[13] Cline HE, Lo rensen WE, Holik AS. Automatic moire contouring, Applied Optics, 1984, 23(10): 1454-1459.
[14] Indebetouw G, Profile measurement using project ion of running fringes. Applied Optics, 1978, 17(18): 2930-2933.
[15] Blatt JH, Hooker JA. 3-D inspect ion of large objects by Moire profilometry. 1989, L. I. A 70 ICAL EO: 12-16.
[16] 吴祖斌，王颖，曹士英等．用改进的光谱相位相干直接电场重构法装置测量飞秒激光脉冲的相位[J]．中国激光，2006，331)：21-25．
[17] 陈锋．动态三维面形测量中相位展开问题的研究[J]．四川大学硕士论文，2005．
[18] 翁嘉文，钟金钢．小波变换轮廓术在双频光栅条纹中的应用[J]．光学学报，2005，25(5)：603-608．
[19] Srinivasan V, Liu H C, etal.. Automated phase measuring profilometry of 3-D diffuse object. Appl. Opt, 1984, 23(18): 3105-3108.
[20] 严明，高志山．移相法测量波片的相位延迟量[J]．光电子．激光，2005，16(2)：183-187．
[21] Creath K, Schm it J. N-point spatial phase-measurement techniques for non-destructive testing. Opt. Lasers Eng, 1996, 24: 365-379.
[22] FBlsis. Review of 20 years of range sensor development [A]. Proceedings of SP IEEE IS&T Electronic Imaging[C]. SPIE 2003, Vol. 5013.
[23] 陶涛，郭红卫，何海涛．镜面反射面形光学三维测量技术综述[J]．测量仪器，2005，27(2)：90-95．
[24] H O Saldner, J M Huntley. Profilometry by temporal phase unwrapping and spatial light modulator based fringe projector [J]. Opt. Eng., 1997, 36 (2): 610-615.
[25] SeulinR, Merienne F, Gorria P. Dynamic lighting system for seculars surface inspection [J]. SPIE, 2001, 4301: 199-2061
[26] 吴立德．计算机视觉[M]．复旦大学出版社，上海，1993．
[27] Murali Subbaro, Tae Choi. Accurate recovery of three dimensional shape from image focus [J]IEEE rrans on PAMI, 1995, 17(3): 266-274.
[28] 祝世平，强锡富．工件特征点三维坐标视觉测量方法综述[J]．光学精密工程，2000，8(2)：192-197．
[29] Indebetouw G, Profile measurement using project ion of running fringes. Applied Optics, 1978, 17(18): 2930-2933.
[30] 陈棣湘，罗飞路，潘孟春．立体视觉测量中的图像匹配策略研究[J]．光学技术，2002，28(5)：9．
[31] 任重，邵军力．双目立体视觉的匹配方法研究[J]．军事通信技术，2001，22(2)：45-48，72．
[32] S.M. Smith. SUSAN-A New Approach to Low Level Image Processing. Technical Report TR95SMSIc, 1995.
[33] S. M. Smith and J. M. Brady, SUSAN-a new approach to low level image processing [J] Int. Journal of Computer Vision, May 1997, Vol. 23, No. 1:45-78.
[34] 龙兴明．三维测量中的图像处理技术研究．重庆师范大学硕士论文，2005．
[35] 康文炜．基于图像确定空间坐标方法的研究．吉林大学硕士论文，2004．
[36] 姚敏等．数字图像处理[M]．北京：机械工业出版社，2006．
[37] Rosenfeld A, Kak A C. Digital Picture Processing. 2nd Edition. New York: Academic Press, 1982.
[38] 陈书海，傅录祥著．实用数字图像处理[M]．北京：科学出版社，2005．1：139-160．
[39] William K．Pratt著，邓鲁华，张延恒等译．数字图像处理[M]．北京：机械工业出版社，2005．
[40] G. P. Dineen, "Programming Pattern Recognition," Proc. Western Joint Computer Conference, March 1955, 94-100.
[41] R. E. Graham, "Snow Removal: A Noise Stripping Process for Picture Signals," IRE Trans. Information Theory, IT8, 1, February 1962, 129-144.
[42] A. Rosenfeld, C. M. Park, and J. P. Strong, "Noise Cleaning in Digital Pictures," Proc. EASCON Convention Record, October 1969, 264-273.
[43] R. Nathan, "Spatial Frequency Filtering," in Picture Processing and Psychohistories, B. S. Lipkin and A. Rosenfeld, Eds., Academic Press, New York, 1970, 151-164.
[44] J. W. Tukey, Exploratory Data Analysis, Addison-Wesley, Reading, MA, 1971.
[45] Linda G．Shapiro，George C．Stockman著，赵清杰等译．计算机视觉[M]．北京：机械工业出版社，2004．
[46] 史长琼，易昂．基于Canny准则的颗粒图像边缘检测算法[J]．计算机工程与科学．2004，26(3)：52-54．
[47] 张冬生．基于计算机视觉的三维测量技术．华北电力大学硕士论文，2003．
[48] 苗红杰，赵文吉，刘先林．数码相机检校和摄像测量的部分问题探讨[J]．首都师范大学学报，2005，26(1)：117-120．
[49] 杨国强．数字近景摄像测量系统研究．西安科技大学硕士论文，2005．
[50] 刘曙光，刘明远等．机器视觉机器应用[J]．机械制造，2000，Vol．38(431)：20-23．
[51] 钟玉琢，乔炳新等．机器人视觉技术[M]．国防工业出版社，北京，1994．
[52] 刘成君，戴汝为．计算机视觉的进展[J]．模式识别与人工智能，1995，Vol．8增刊．
[53] 王天珍．计算机视觉研究进展[J]．武汉汽车工业大学学报，1998，20(1)，23-27．
[54] 贾云德．机器视觉[M]．科学出版社，北京，2000．
[55] 于洪川，吴福朝等．基于主动视觉的摄像机自标定技术[J]．机器人，1999，1．
[56] 张毓晋．图像处理和分析[M]．北京：清华大学出版社，1999．3：43-49．
[57] Heath, M. 1997. Sciecntific Computing: An Introductory Survey. McGraw-Hill, Inc., New York.
[58] J. Weng, P. Cohen and M. Herniou, Calibration of Stereo Cameras Using a Non-Linear Distortion Model, Proc, International Conference on Pattern Recognition, pp. 246-253, 1990.
[59] Tsai R Y. An efficient and accurate camera calibration technique for3-D machine vision [A]. In Proc IEEC Conf. On Computer Vision and Pattern Recognition[C], Miami Beach: 1986, 364-374.
[60] O. D. Faugeras and G. Toscani, Camera Calibration for 3D Computer Vision, Proc. of International Workshop on Industrial Application of Machine Vision an Machine Intelligence, pp. 240-247, Japan, 1987.
[61] 张靖瑜，董明得，吕乃光，邓文怡．用广义逆法解决摄影测量系统的非线性问题[J]．北京机械工业学院学报．2001，16(1)：36-39．
[62] 伍雪冬．计算机视觉中摄像机定标及位姿和运动估计方法的研究．湖南大学博士论文，2004．
[63] 范洪．3D计算机视觉系统的高精度定标．北京交通大学博士论文，1993
[64] 周婷婷．基于图像的三维测量．安徽大学硕士论文，2005．8
[65] 丁丽娟．数值计算方法[M]．北京理工大学出版社，2004．
[66] 张翔．优化设计方法及编程[M]．中国农业大学出版社，2000．
[67] Francesco Balena著，李珂，张文波，赵明等译，Visual Studio．NET产品组审校．Visual Basic．NET技术内幕，2003．
[68] 管业鹏，童林夙，尹涵春．基于数码相机的三维物体空间几何位置的摄像测量[J]．电子学报．2002，30(6)：849-852．
[69] 林成森．数值分析[M]．科学出版社，2002．