图象测量及分析系统的研究与实现
|
摘要
不断降价和普及的硬件设备以及在商业、医学、科研等领域稳定涌现出的新的应用，使得图象处理领域一直保持持续发展的势头并将在未来发挥更为重要的作用。图象的在线和高精度的测量分析也越来越广泛的应用于工业的许多领域，而已经商业化的应用系统对特定应用场合有许多的不便和不足，因此，该课题的提出有其必要性和现实的意义。
本论文在一般光学镜头的基础上实现了一种基于面阵CCD摄像机的图象测量及分析系统。该系统能够实现自动对焦采集图象和在线对回转体工件装配间隙进行测量分析的任务。
本论文综合分析了图象处理系统的各个组成部分对整个系统测量分析的精度影响的因素，明确了在已经预先配置好的硬件设计的基础上，通过软件的方法来加以补偿，在硬件设备以及用于对数字图象处理进行处理和定量分析的算法之间建立一个平衡，使系统的整体性能满足任务的需要。
在自动聚焦过程中，论文采用了图象的灰度差分的绝对值之和的平方作为焦距评价函数，解决了是否正确聚焦的快速判断问题；采用了一种行之有效的基于阈值和曲线拟合的自动聚焦搜索方式，使聚焦速度和精度都得到了很大的提高。
论文指出可以充分利用图象卡的功能，采用在内存中多图象平均的方法来有效的消除加性噪声的影响。论文综合分析了在图象分割前各平滑滤波器对图象的影响，明确高斯滤波器可以在保留图象细节和去除噪声间寻求平衡。
论文在分析当前图象分割的现状和趋势的基础上，采用被称为最佳边缘检测算子的Marr算法，使图象分割和边缘提取的准确性得到了很大的提高；利用人机交互辅助的方法来搜索已经检出的稀疏边缘点，并用最小二乘法拟合这些边缘点，使装配间隙宽度的测量的精度和准确性得到了极大的保证。对整个系统采用了长光栅来标定，有效的减少了系统误差。
实验表明该系统具有对焦速度快、范围大、对焦精确，系统精度高、体积小、可扩充性好的特点。聚焦范围可达几十毫米，聚焦时间为5秒钟，重复精度为7。装配间隙宽度测量的精度为几微米。
The perpetually declining cost and increasing availability of hardware required and a steady flow of new applications in commercial, medical field and in scientific research indicate continued growth for the digital image-processing system field and play an important role in the future. Image measurement and analysis on-line and at high accuracy has been applied in many industrial filed widely. However, the general-purpose and some special-purpose image-processing system yield a lot of discommodity and insufficiency in particular applications. So the development of this task has its essential and practical significance.
This paper presents a type of image measurement and analysis system based on CCD plane array camera and general lens that can accomplish auto-focusing and collecting image and measuring and analyzing on-line assembly-gap of columned work-piece.
This paper summarizes and analyzes the factor influencing upon the precision of image measurement and analysis system for all components. This paper brings forward that one can control over the computer program to compensate for these effects and establish a balance among the optics, camera, other harder portion and the algorithms used for processing and quantitative analysis of the digital images so that the overall performance is adequate for the task when the system components is preset.
In course of auto-focusing, the paper takes the sum of the square of gray difference as focal distance evaluation function to decided that image is in focus or not quickly and put forward an effective auto-focus searching way based on threshold value and curve fitting. These improve speed and precision of auto-focusing.
This paper points out that one can take advantage of the function of image board to average them together with multiple images of the same scene to reduce the effects of additive random noise. The paper also analyzes synthetically the effects of some smooth filters prior to image segmentation and indicates that Gaussian lowpass filter, a alterable scale filter, is a good choice to establish a balance between the reservation of image details and removal of noise.
On the basis of current investigation of actuality and trend of image segmentation at home and abroad, the paper adopts Marr operator, called by optimal edge detection to improve greatly efficiency of image segmentation and edge abstract. The sparse edge
points are taken out by the means of PC-person interaction to fit at least squares principle. So the precision and validity of assemble gap is ensured. Calibration is taken by length grating at overall system and system error is reduced effectively.
Experiment indicates that the system has fast-speed, high precision, small bulk and dynamic range and well-extensible characteristics. The system possesses dynamic range of scores of millimeters and focusing time of five seconds and repeatability of seven microns and measurement precision of microns level.
本论文在一般光学镜头的基础上实现了一种基于面阵CCD摄像机的图象测量及分析系统。该系统能够实现自动对焦采集图象和在线对回转体工件装配间隙进行测量分析的任务。
本论文综合分析了图象处理系统的各个组成部分对整个系统测量分析的精度影响的因素，明确了在已经预先配置好的硬件设计的基础上，通过软件的方法来加以补偿，在硬件设备以及用于对数字图象处理进行处理和定量分析的算法之间建立一个平衡，使系统的整体性能满足任务的需要。
在自动聚焦过程中，论文采用了图象的灰度差分的绝对值之和的平方作为焦距评价函数，解决了是否正确聚焦的快速判断问题；采用了一种行之有效的基于阈值和曲线拟合的自动聚焦搜索方式，使聚焦速度和精度都得到了很大的提高。
论文指出可以充分利用图象卡的功能，采用在内存中多图象平均的方法来有效的消除加性噪声的影响。论文综合分析了在图象分割前各平滑滤波器对图象的影响，明确高斯滤波器可以在保留图象细节和去除噪声间寻求平衡。
论文在分析当前图象分割的现状和趋势的基础上，采用被称为最佳边缘检测算子的Marr算法，使图象分割和边缘提取的准确性得到了很大的提高；利用人机交互辅助的方法来搜索已经检出的稀疏边缘点，并用最小二乘法拟合这些边缘点，使装配间隙宽度的测量的精度和准确性得到了极大的保证。对整个系统采用了长光栅来标定，有效的减少了系统误差。
实验表明该系统具有对焦速度快、范围大、对焦精确，系统精度高、体积小、可扩充性好的特点。聚焦范围可达几十毫米，聚焦时间为5秒钟，重复精度为7。装配间隙宽度测量的精度为几微米。
The perpetually declining cost and increasing availability of hardware required and a steady flow of new applications in commercial, medical field and in scientific research indicate continued growth for the digital image-processing system field and play an important role in the future. Image measurement and analysis on-line and at high accuracy has been applied in many industrial filed widely. However, the general-purpose and some special-purpose image-processing system yield a lot of discommodity and insufficiency in particular applications. So the development of this task has its essential and practical significance.
This paper presents a type of image measurement and analysis system based on CCD plane array camera and general lens that can accomplish auto-focusing and collecting image and measuring and analyzing on-line assembly-gap of columned work-piece.
This paper summarizes and analyzes the factor influencing upon the precision of image measurement and analysis system for all components. This paper brings forward that one can control over the computer program to compensate for these effects and establish a balance among the optics, camera, other harder portion and the algorithms used for processing and quantitative analysis of the digital images so that the overall performance is adequate for the task when the system components is preset.
In course of auto-focusing, the paper takes the sum of the square of gray difference as focal distance evaluation function to decided that image is in focus or not quickly and put forward an effective auto-focus searching way based on threshold value and curve fitting. These improve speed and precision of auto-focusing.
This paper points out that one can take advantage of the function of image board to average them together with multiple images of the same scene to reduce the effects of additive random noise. The paper also analyzes synthetically the effects of some smooth filters prior to image segmentation and indicates that Gaussian lowpass filter, a alterable scale filter, is a good choice to establish a balance between the reservation of image details and removal of noise.
On the basis of current investigation of actuality and trend of image segmentation at home and abroad, the paper adopts Marr operator, called by optimal edge detection to improve greatly efficiency of image segmentation and edge abstract. The sparse edge
points are taken out by the means of PC-person interaction to fit at least squares principle. So the precision and validity of assemble gap is ensured. Calibration is taken by length grating at overall system and system error is reduced effectively.
Experiment indicates that the system has fast-speed, high precision, small bulk and dynamic range and well-extensible characteristics. The system possesses dynamic range of scores of millimeters and focusing time of five seconds and repeatability of seven microns and measurement precision of microns level.
引文
[1] 章毓晋.图象工程(上册)--图象处理和分析.北京.清华大学出版社.1999:1-12.
[2] 章毓晋.图像工程及在中国的研究状况和文献分布.中国工程科学. 2000.2(8):91-94.
[3] 阮秋琦.数字图像处理学. 北京. 电子工业出版社.2001:1-7.
[4] 郭力,H.S.Cheng.21世纪机械科学的展望(下).精密制造与自动化.2001.2:11-14.
[5] 叶声华,王仲,曲兴华.精密测试技术展望.中国机械工程.2000.11(3):262-263.
[6] 周铭,骆瑞伦,单越康.零件几何参数的计算机测量系统.计算机自动测量与控制.2000.8(4):38-48.
[7] 曲梅.测量数据的自动处理.自动化与仪表.1999.14(5):56-58.
[8] 张国雄.测量与信息技术.中国工程科学.2000.2(3):62-66.
[9] 沈元如,章安仁.球壳组合缝隙位置自动检测.计量技术.1999.2:6-8.
[10] K.R.Castleman著.朱志刚，林学訚，石定机等译.数字图像处理. 北京. 电子工业出版社.1998: 294-320.
[11] 吴晓波,杨永琴. 图像测量技术的新应用. 光学精密工程. 1998.6(3):10-16.
[12] 冷何英,戴俊钊. 实用型亚像元定位方法的研究. 红外与激光工程. 2000.29(2):15-18.
[13] 吴小波,钟先信,刘厚权,张启明.应用多项式插值函数提高面阵CCD尺寸测量的分辨力.仪器仪表学报.1996.17(2):154-159.
[14] 容观澳.计算机图象处理.北京.清华大学出版社.2000: 6-10.
[15] 吴晓波,安文斗,杨钢. 图像测量系统中的误差分析及提高测量精度的途径. 光学精密工程. 1997.5(1): 133-141.
[16] 王冬梅,方如华,秦玉文.数字图像相关测量的误差分析及其改进措施.实验力学.1998.13(3):416-422.
[17] 邓湘金,杨民,路宏年,宋文爱.CCD像元间暗电流和响应不一致性的矫正方法.华北工学院学报.2000.21(1):20-22.
[18] 杨云志.CCD精密测量系统及其数字信号处理.系统工程与电子技术.1999.21(6):74-76.
[19] 肖沙里,钟先信,陈愚,徐道连. 线阵CCD在异形回转体轮廓尺寸测量中的应用. 重庆大学学报(自然科学版).1997.20(6):85-90.
[20] 庞长富，刘榴娣.CCD摄像机用于测量中存在的问题及解决方法.光学技术.1996(2):5-8.
[21] 袁祥辉.固体图象传感器及其应用.重庆.重庆大学出版社.1996:39-54.
[22] 陈后金,吴湘淇.计算机图像形态测量中的校准与归一化.铁道学报.1996.18(1):70-73.
[23] 马品仲.近代几何量精测技术进展.计量技术.1995.6:2-4.
[24] 程开富.CCD图像传感器的市场与发展. 国外电子元器件.2000.7:2-7.
[25] 应骏,何国锋,叶秀清.基于PCI总线的立体视觉系统实现非接触式实时测量.测控技术.2000.19(2):52-54.
[26] 徐定杰,王晓溪,王武义,袁哲俊.数字图像处理在焦距测量中的应用.仪表技术与传感器.1998(7):37-39.
[27] 黄剑琪,冯华君,徐之海,戴顺林等.边缘特征增强算法和小波分析在精确聚焦中的应用.光子学报.2000.29(10):932-935.
[28] 王建民,浦昭邦,尹继学,晏磊.图像式显微自动调焦瞄准系统的研究.光学技术.2000.26(5):410-417.
[29] Kang-sun chio, Jun-suk Lee, Sung-Jae Ko. New autofocusing technique using the frequency selective weighted median filter for video cameras.IEEE-CE.1999.45(3):820-827.
[30] Murali Subbarao, Jenn-Kwei Tyan. Selecting the optimal focus measure for autofocusing and depth-from-focus.IEEE-PAMI.1998.20(8):864-870.
[31] 郭彦珍,邱宗明,李信,王吉晖.图像测量技术中一种调焦的判别方法.西安理工大学学报.2001.17(1):40-42.
[32] Nikhil R.Pal and Sankar K.Pal.A review on image segmentation techniques. Pattern Recognition.1993.26(9):1277-1289.
[33] 贾云得.机器视觉.北京.科学出版社.2000:69-80.
[34] 章毓晋.图象分割.北京.科学出版社.2001:9-17.
[35] 罗希平,田捷,诸葛婴,王靖等.图像分割方法综述.模式识别与人工智能.1999.12(3):300-309.
[36] 周心明,兰赛,徐燕.图像处理中几种边缘检测算法的比较.现代电力.2000.17(3):65-69.
[37] 张凯丽,刘辉.边缘检测技术的发展研究.昆明理工大学学报.2000.25(5):36-39.
[38] 周凌翔,顾伟康.最佳边缘检测的准则与算子.模式识别与人工智能.1998.11(1):54-61.
[39] 郑南宁.计算机视觉与模式识别.北京.国防工业出版社.1998:56-61.
[40] Lawrence O'Gorman. Subpixel precision of straight-edged shapes for registration and measurement. IEEE-PAMI.1996.18(7):746-751.
[41] Wenwu Zhu,Yao Wang,Qin-Fan Zhu.Second-order derivative-based smoothness measure for error concealment in DCT-based codecs.IEEE-CSTV.1998.8(6):713-718.
[42] Federico Pedersini，Augusto Sarti etc.．Estimation and Compensation of Subpixel Edge Location Error．IEEE-PAMI,Vol.19,No.10,Nov.1997：1278~1284.
[43] Clark J J. Authenticating edges produced by zero-crossing algorithm.IEEE-PAMI.1998.11(1):43-57.
[44] Shen J,Castan S. An Optimal linear operator for step edge detection.CVGIP-GMIP,1992.54:112-133.
Tabb M, Ahuja N. Multiscale image segmentation by integrated edge and region detection,
[45] IEEE-IP.1997.6(5):642-655.
[46] 刘文军,张小锋,陈轩.一种图像分割方法的实现.南昌航空工业学院学报.2000.14(1):25-27.
[47] He Y, Zhang Y J. An analysis study of the effect of noise on the performance of differential edge operator. Proceedings of the 2nd ICSP, 1993.2:1314-1317.
[48] Marr D, Hildreth E. Theory of edge detection. Proceeding of R. Soc. London, 1980, B207:187-217.
[49] Yuille A L, Poggio T. Scaling theorems for zero crossing. IEEE-PAMI. 1986.8:15-25.
[50] Shen J, Castan S. An optimal edge detection in two-dimensional images. IEEE-IP. 1996.5(7):1215-1220.
[51] 鲍宗泛,李红华.关于图像边缘检测的Laplace算子的改进.中国计量学院学报.2000.11(2):169-172.
[52] 李小文.利用拉普拉斯-高斯模板进行边缘检测.华南师范大学学报(自然科学版).1997.(2):53-55.
[53] 董汉莉.Marr-Hildreth算子边缘精确定位的研究.郑州工业大学学报.1999.20(2):36-38.
[54] 谢昭辉, 吴立得.关于算子的一些研究.自动化学报.1990.16(3):193-200.
[55] 张娇, 胡福乔,李文舜.边缘检测中的LOG滤波器的改进方法的研究.微电子学与计算机.1998.6:36-40.
[56] Kozo Ohtani．A Fast Edge Location Measurement with Subpixel Accuracy Using CCD Image．IEEE Instrumentation and measurement Technology Conference ,Budapest, Hungary, May 21-23-2001.
[57] 王春森.程序设计.北京.清华大学出版社.1999:207-211.
[58] 费业泰.误差理论与数据处理.北京.机械工业出版社.1994:82-84.
[59] 宗殿瑞,宋文臣,刘朋振.最小二乘法应用探讨.青岛化工学院学报.1998.19(3):296-301.
[2] 章毓晋.图像工程及在中国的研究状况和文献分布.中国工程科学. 2000.2(8):91-94.
[3] 阮秋琦.数字图像处理学. 北京. 电子工业出版社.2001:1-7.
[4] 郭力,H.S.Cheng.21世纪机械科学的展望(下).精密制造与自动化.2001.2:11-14.
[5] 叶声华,王仲,曲兴华.精密测试技术展望.中国机械工程.2000.11(3):262-263.
[6] 周铭,骆瑞伦,单越康.零件几何参数的计算机测量系统.计算机自动测量与控制.2000.8(4):38-48.
[7] 曲梅.测量数据的自动处理.自动化与仪表.1999.14(5):56-58.
[8] 张国雄.测量与信息技术.中国工程科学.2000.2(3):62-66.
[9] 沈元如,章安仁.球壳组合缝隙位置自动检测.计量技术.1999.2:6-8.
[10] K.R.Castleman著.朱志刚，林学訚，石定机等译.数字图像处理. 北京. 电子工业出版社.1998: 294-320.
[11] 吴晓波,杨永琴. 图像测量技术的新应用. 光学精密工程. 1998.6(3):10-16.
[12] 冷何英,戴俊钊. 实用型亚像元定位方法的研究. 红外与激光工程. 2000.29(2):15-18.
[13] 吴小波,钟先信,刘厚权,张启明.应用多项式插值函数提高面阵CCD尺寸测量的分辨力.仪器仪表学报.1996.17(2):154-159.
[14] 容观澳.计算机图象处理.北京.清华大学出版社.2000: 6-10.
[15] 吴晓波,安文斗,杨钢. 图像测量系统中的误差分析及提高测量精度的途径. 光学精密工程. 1997.5(1): 133-141.
[16] 王冬梅,方如华,秦玉文.数字图像相关测量的误差分析及其改进措施.实验力学.1998.13(3):416-422.
[17] 邓湘金,杨民,路宏年,宋文爱.CCD像元间暗电流和响应不一致性的矫正方法.华北工学院学报.2000.21(1):20-22.
[18] 杨云志.CCD精密测量系统及其数字信号处理.系统工程与电子技术.1999.21(6):74-76.
[19] 肖沙里,钟先信,陈愚,徐道连. 线阵CCD在异形回转体轮廓尺寸测量中的应用. 重庆大学学报(自然科学版).1997.20(6):85-90.
[20] 庞长富，刘榴娣.CCD摄像机用于测量中存在的问题及解决方法.光学技术.1996(2):5-8.
[21] 袁祥辉.固体图象传感器及其应用.重庆.重庆大学出版社.1996:39-54.
[22] 陈后金,吴湘淇.计算机图像形态测量中的校准与归一化.铁道学报.1996.18(1):70-73.
[23] 马品仲.近代几何量精测技术进展.计量技术.1995.6:2-4.
[24] 程开富.CCD图像传感器的市场与发展. 国外电子元器件.2000.7:2-7.
[25] 应骏,何国锋,叶秀清.基于PCI总线的立体视觉系统实现非接触式实时测量.测控技术.2000.19(2):52-54.
[26] 徐定杰,王晓溪,王武义,袁哲俊.数字图像处理在焦距测量中的应用.仪表技术与传感器.1998(7):37-39.
[27] 黄剑琪,冯华君,徐之海,戴顺林等.边缘特征增强算法和小波分析在精确聚焦中的应用.光子学报.2000.29(10):932-935.
[28] 王建民,浦昭邦,尹继学,晏磊.图像式显微自动调焦瞄准系统的研究.光学技术.2000.26(5):410-417.
[29] Kang-sun chio, Jun-suk Lee, Sung-Jae Ko. New autofocusing technique using the frequency selective weighted median filter for video cameras.IEEE-CE.1999.45(3):820-827.
[30] Murali Subbarao, Jenn-Kwei Tyan. Selecting the optimal focus measure for autofocusing and depth-from-focus.IEEE-PAMI.1998.20(8):864-870.
[31] 郭彦珍,邱宗明,李信,王吉晖.图像测量技术中一种调焦的判别方法.西安理工大学学报.2001.17(1):40-42.
[32] Nikhil R.Pal and Sankar K.Pal.A review on image segmentation techniques. Pattern Recognition.1993.26(9):1277-1289.
[33] 贾云得.机器视觉.北京.科学出版社.2000:69-80.
[34] 章毓晋.图象分割.北京.科学出版社.2001:9-17.
[35] 罗希平,田捷,诸葛婴,王靖等.图像分割方法综述.模式识别与人工智能.1999.12(3):300-309.
[36] 周心明,兰赛,徐燕.图像处理中几种边缘检测算法的比较.现代电力.2000.17(3):65-69.
[37] 张凯丽,刘辉.边缘检测技术的发展研究.昆明理工大学学报.2000.25(5):36-39.
[38] 周凌翔,顾伟康.最佳边缘检测的准则与算子.模式识别与人工智能.1998.11(1):54-61.
[39] 郑南宁.计算机视觉与模式识别.北京.国防工业出版社.1998:56-61.
[40] Lawrence O'Gorman. Subpixel precision of straight-edged shapes for registration and measurement. IEEE-PAMI.1996.18(7):746-751.
[41] Wenwu Zhu,Yao Wang,Qin-Fan Zhu.Second-order derivative-based smoothness measure for error concealment in DCT-based codecs.IEEE-CSTV.1998.8(6):713-718.
[42] Federico Pedersini，Augusto Sarti etc.．Estimation and Compensation of Subpixel Edge Location Error．IEEE-PAMI,Vol.19,No.10,Nov.1997：1278~1284.
[43] Clark J J. Authenticating edges produced by zero-crossing algorithm.IEEE-PAMI.1998.11(1):43-57.
[44] Shen J,Castan S. An Optimal linear operator for step edge detection.CVGIP-GMIP,1992.54:112-133.
Tabb M, Ahuja N. Multiscale image segmentation by integrated edge and region detection,
[45] IEEE-IP.1997.6(5):642-655.
[46] 刘文军,张小锋,陈轩.一种图像分割方法的实现.南昌航空工业学院学报.2000.14(1):25-27.
[47] He Y, Zhang Y J. An analysis study of the effect of noise on the performance of differential edge operator. Proceedings of the 2nd ICSP, 1993.2:1314-1317.
[48] Marr D, Hildreth E. Theory of edge detection. Proceeding of R. Soc. London, 1980, B207:187-217.
[49] Yuille A L, Poggio T. Scaling theorems for zero crossing. IEEE-PAMI. 1986.8:15-25.
[50] Shen J, Castan S. An optimal edge detection in two-dimensional images. IEEE-IP. 1996.5(7):1215-1220.
[51] 鲍宗泛,李红华.关于图像边缘检测的Laplace算子的改进.中国计量学院学报.2000.11(2):169-172.
[52] 李小文.利用拉普拉斯-高斯模板进行边缘检测.华南师范大学学报(自然科学版).1997.(2):53-55.
[53] 董汉莉.Marr-Hildreth算子边缘精确定位的研究.郑州工业大学学报.1999.20(2):36-38.
[54] 谢昭辉, 吴立得.关于算子的一些研究.自动化学报.1990.16(3):193-200.
[55] 张娇, 胡福乔,李文舜.边缘检测中的LOG滤波器的改进方法的研究.微电子学与计算机.1998.6:36-40.
[56] Kozo Ohtani．A Fast Edge Location Measurement with Subpixel Accuracy Using CCD Image．IEEE Instrumentation and measurement Technology Conference ,Budapest, Hungary, May 21-23-2001.
[57] 王春森.程序设计.北京.清华大学出版社.1999:207-211.
[58] 费业泰.误差理论与数据处理.北京.机械工业出版社.1994:82-84.
[59] 宗殿瑞,宋文臣,刘朋振.最小二乘法应用探讨.青岛化工学院学报.1998.19(3):296-301.