基于机器视觉的小型规则零件二维尺寸测量研究
|
摘要
基于机器视觉的检测技术具有非接触、在线实时、速度快、抗干扰能力强等优点,为了适应了现代制造业的进步和发展要求,本文应用机器视觉技术、图像处理和图像分析技术,在现有的实验条件下以轴承和端子为具体研究对象,对小型规则零件的尺寸测量技术进行研究和分析。
本文主要内容包括:
1.首先阐述了基于机器视觉的小型规则零件二维尺寸测量技术的研究背景和意义,然后介绍了机器视觉技术的发展和在尺寸测量方面的应用,概述了本文的研究目的和主要工作。
2.分析了基于机器视觉的小型规则零件二维尺寸测量系统的组成和工作原理。讨论了图像测量中涉及的直方图修正、图像滤波和锐化,阈值分割等预处理技术以及经典边缘检测、Canny边缘检测、轮廓提取和轮廓跟踪等边缘检测技术。并分析了锐化给图像测量带来的弊端,提出使用了二次滤波。为减小后续图像处理数据量,提出了感兴趣区域提取技术。达到提高准确度、降低难度的目的。
3.在保证获得较好图像质量的前提下,从经济实用的角度出发,选择合适的相机、镜头、光源和图像采集卡搭建出了一套有效的轴承尺寸测量系统和端子尺寸测量系统。
4.对工业中常出现的几何特征:直线、单一圆、多圆、角点、切点、拐点、三角形和多边形等进行了特征提取和检测,并分析比较了各种方法的适用范围。对角点检测时的特殊角点情况进行了详细的分析,对改进的Hough变换方法的参数选择进行了实验验证,得出了参数选择的一般规律。在基本的检测基础上提出采用了最小二乘拟合技术进行亚像素定位,使得圆的检测精度得到提高。
5.通过Visual C++6.0环境下的MFC编程,对各种检测算法编写程序集,分别对轴承和端子进行了尺寸测量。对端子检测提出了分区域检测,以降低难度,提高检测速度。对轴承采用标准零件进行标定,以避免由分辨率带来的影响,提高了检测精度。
本文的研究内容对推动我国视觉测试技术的发展具有一定的实际意义。
With the character of non-contact, on-line and real-time, rapid, suitable precision and strong scene anti-jamming, the machine vision measurement technology has suited for the need of modern manufacturing progress and development. Combining machine vision with image processing technology, under the experimental condition on hand, this paper adopts basic dimension parameter of some normal parts (e g, bearing, serial terminal etc.) as studied object in manufacturing and develops the 2-D dimension research on machine vision system.
Each chapter central contents are as follows:
Firstly, this paper expounds the research background and significance of machine vision inspecting. Based on the development and applied status of machine vision measurement technology in manufacturing, the research aim and central task are introduced.
Secondly, the basic makeup and work principle for machine vision measurement system has been introduced. The digital image processing technology as histogram Process, image filtering, sharpening, threshold segmentation, classical edge-detection, canny edge-detection and so on, has been discussed. Re-filtering for sharpening image is put forward to improve image outline result and the interesting regions technology can reduce the measurement difficulty.
Thirdly, bearing and serial terminal dimension measurement systems have been designed by appropriate camera, lens, lamp-house, image collection card and other hardware instrument. Inspection board for serial terminal has been designed for fixing the position of measurement.
Forthly, .this paper presents the method to extract and measurement beeline, singularity circle, double circles, corner and tangent point, triangle and polygon and so on. The application area of these methods and special corner points have been analyst and discussed. Circle preferences rules based on the improvement Hough transformation have been validated by experiment. The technique of sub-pixel edge orientation is an efficient method which uses software arithmetic to advance measure precision, so it plays an important roll in visual inspection.
Finally, the paper develops corresponding software by Visual C++6.0 on the Window XP plate. Measurement result indicates the software can meet industrial measurement requirement. Calibration is a critical part of visual inspection. The paper uses standard specimen as reference objects to calibrate by relative calibrating method to make the calibration process fast and accurate.
本文主要内容包括:
1.首先阐述了基于机器视觉的小型规则零件二维尺寸测量技术的研究背景和意义,然后介绍了机器视觉技术的发展和在尺寸测量方面的应用,概述了本文的研究目的和主要工作。
2.分析了基于机器视觉的小型规则零件二维尺寸测量系统的组成和工作原理。讨论了图像测量中涉及的直方图修正、图像滤波和锐化,阈值分割等预处理技术以及经典边缘检测、Canny边缘检测、轮廓提取和轮廓跟踪等边缘检测技术。并分析了锐化给图像测量带来的弊端,提出使用了二次滤波。为减小后续图像处理数据量,提出了感兴趣区域提取技术。达到提高准确度、降低难度的目的。
3.在保证获得较好图像质量的前提下,从经济实用的角度出发,选择合适的相机、镜头、光源和图像采集卡搭建出了一套有效的轴承尺寸测量系统和端子尺寸测量系统。
4.对工业中常出现的几何特征:直线、单一圆、多圆、角点、切点、拐点、三角形和多边形等进行了特征提取和检测,并分析比较了各种方法的适用范围。对角点检测时的特殊角点情况进行了详细的分析,对改进的Hough变换方法的参数选择进行了实验验证,得出了参数选择的一般规律。在基本的检测基础上提出采用了最小二乘拟合技术进行亚像素定位,使得圆的检测精度得到提高。
5.通过Visual C++6.0环境下的MFC编程,对各种检测算法编写程序集,分别对轴承和端子进行了尺寸测量。对端子检测提出了分区域检测,以降低难度,提高检测速度。对轴承采用标准零件进行标定,以避免由分辨率带来的影响,提高了检测精度。
本文的研究内容对推动我国视觉测试技术的发展具有一定的实际意义。
With the character of non-contact, on-line and real-time, rapid, suitable precision and strong scene anti-jamming, the machine vision measurement technology has suited for the need of modern manufacturing progress and development. Combining machine vision with image processing technology, under the experimental condition on hand, this paper adopts basic dimension parameter of some normal parts (e g, bearing, serial terminal etc.) as studied object in manufacturing and develops the 2-D dimension research on machine vision system.
Each chapter central contents are as follows:
Firstly, this paper expounds the research background and significance of machine vision inspecting. Based on the development and applied status of machine vision measurement technology in manufacturing, the research aim and central task are introduced.
Secondly, the basic makeup and work principle for machine vision measurement system has been introduced. The digital image processing technology as histogram Process, image filtering, sharpening, threshold segmentation, classical edge-detection, canny edge-detection and so on, has been discussed. Re-filtering for sharpening image is put forward to improve image outline result and the interesting regions technology can reduce the measurement difficulty.
Thirdly, bearing and serial terminal dimension measurement systems have been designed by appropriate camera, lens, lamp-house, image collection card and other hardware instrument. Inspection board for serial terminal has been designed for fixing the position of measurement.
Forthly, .this paper presents the method to extract and measurement beeline, singularity circle, double circles, corner and tangent point, triangle and polygon and so on. The application area of these methods and special corner points have been analyst and discussed. Circle preferences rules based on the improvement Hough transformation have been validated by experiment. The technique of sub-pixel edge orientation is an efficient method which uses software arithmetic to advance measure precision, so it plays an important roll in visual inspection.
Finally, the paper develops corresponding software by Visual C++6.0 on the Window XP plate. Measurement result indicates the software can meet industrial measurement requirement. Calibration is a critical part of visual inspection. The paper uses standard specimen as reference objects to calibrate by relative calibrating method to make the calibration process fast and accurate.
引文
[1] B G Batchelor, D W Braggins. Commercial vision systems in computer vision [J]. Theory and Industrial Applications, 1992:405-452.
[2]章毓晋.图象理解与计算机工程[M].北京:清华大学出版社,2000.
[3]朱森良.计算机视觉[M].浙江大学出版社,1997.
[4] D.H. Ballard, C.M. Brown. Computer vision[M].Prentice-Hall, Englewood Cliffs,New Jersey,1982.
[5]贾云得.机器视觉[M].北京:科学出版社,2000,4.
[6] F. Keceki, H. H. Nagel. Machine vision based estimation of pose and size parameters from a genetic work piece description [J]. IEEE Tran On robotics an Automation, 2001, 3:2159-2164.
[7] Eric P. Krotkov. Visual Hyperacuity: Representation and Computation of High Precision Position Information [J]. COMPUTER VISION, GRAPHICS AND IMAGE PROCESSING, 1986, 33:99-115.
[8] S.Sirkis, T. J. Lim, Displacement and Strain Measurement with automated Grid Methods [J]. Experimented Mechanics, December 1991:382-388.
[9]张文景,张文渊,苏键峰等.计算机视觉检测技术及其在机械零件检测中的应用.上海交通大学学报[J],1999, 33(5):635-638.
[10]邹定海,叶声华,王春和.用于在线测量的视觉检测系统.仪器仪表学报[J],1995, 16(4):337-340.
[11]阮秋琦.数字图像处理学[M].电子工业出版社,2001.
[12]杨淑莹,边奠英.VC++图像处理程序设计[M].清华大学出版社,2005.
[13]夏良正.数字图像处理[M].南京:东南大学出版社,1999.
[14]于起峰,陆宏伟,刘肖琳.基于图像的精密测量与运动测量[M].北京:科学出版社,2002.
[15]于殿泓.图像检测与处理技术[M].西安电子科技大学出版社,2006.
[16]Shi Hongli, Cai Yuanli, Qiu Zulian. Adaptive wavelet domain thresholding denoising. In: Proceedings of SPIE-The International Society for Optical Engineering, Fifth International Symposium on Instrumentation and Control Technology. 2003, 5253:245-250.
[17]Byung-Jun Yoan, Vaidyanathan P P. Wavelet-based denoising by customizedthresholding. In: IEEE International Conference on Acoustics, Speech and SignalProcessing.2004:925-928
[18]Coyle Edward J, Lin Jean-Hsang. Stack filters and the mean absolute error criterion.IEEE Transactions on Acoustics, Speech and Signal Processing. 1988, 36(8):1244-1254
[19]Haritopoulos Michel,Yin Hujun, Allinson Nigel M. Image denoising using self-organizing map-based nonlinear independent component analysis, Neural Networks.2002, 15(8-9):1085-1098.
[20]杨淑莹,李兰友.图像模式识别——VC++技术实现[M].清华大学出版社,2005.
[21]杨枝灵,王开.Visual C++数字图像获取、处理及实践应用[M].人民邮电出版社,2003.
[22]Philippe Salembier. Morphological multiscale segmentation for image coding[J].Signal Processing,1994,38:359-386.
[23]刘立波.图像分割方法探讨[J].宁夏农学院学报,2001,22(4):51-54.
[24](美)冈萨雷斯著,阮秋琦,阮宇智译.数字图像处理[M].电子工业出版社,2007.
[25]张凯丽,刘辉.边缘检测技术的发展研究[J].昆明理工大学学报,2000,25(5): 36-39.
[26]李庆利.尺寸测量中的边缘检测算法研究[J].计算机测量与控制,2004,12(4):334-337.
[27]何斌.Visual C++数字图象处理[M].北京:人民邮电出版社,2001.
[28]Canny J. A computational approach to edge detection [J]. IEEE-PAMI, 1986, 8:679-698.
[29]余松煜,周源华,张瑞.数字图像处理[M].上海交通大学出版社,2007.
[30]周长发.精通Visual C++图像处理编程[M].北京:电子工业出版社,2006,6.
[31]求是科技. Visual C++图像处理典型算法及实现[M].人民邮电出版社,2006,6.
[32]Stan Z Li, Lie Gu. Kernel machine based learning for mufti-view face detection and pose estimation [J]. Technical Report MSR TR-2001-07.2001:674-679.
[33]林大钊.计算机工程图形算法与应用[M].华东理工大学出版社,2006.
[34]王锁明,侯彬,陈泽珍.最小二乘法直线拟合数据处理的VB程序设计[J].电脑知识与技术,2007,8:782-783.
[35]俞巧云,邢晓正.直线拟合方法在一维图像边缘检测中的应用[J].光学工程,2001,28(6):56-58,65.
[36]郝宏文.圆[M].大象出版社,2004,2.
[37]Illingworth, J. and J. Kittler. A survey of the Hough Transform [J]. Computer Vision Graphics, and Image Processing, 1988(44): 87-116.
[38]Tsuji, S. and F. Matsumoto. Detection of ellipses by modified Hough transformation [J].IEEE transaction on Computers, 1978(27): 777-781.
[39]Wechsler, H. and J. Sklansky. Automatic detection of ribs in chest radiographs [J]. Pattern Recognition, 1977(9): 21-30.
[40]陈传波,金先级.数字图像处理[M].机械工业出版社,2004,7.
[41]F.Mokhtarian, R.Suomela. Robust image corner detection through curvature scale space [J].IEEE Trans. on Pattern Analysis and Machine Intelligence. 1998, 20(12):1376-1381.
[42]A Bandera, C Urdiales, et al.. Corner detection by means of adaptively estimated curvature function [J]. Electronics Letters, 2000, 36:124-126.
[43]P Reche, C Urdiales, et al.. Corner detection by means of contour local vectors [J].Electronics Letters. 2002, 38(14):699-701.
[44]C Urdiales, C Trazegnies, et al.. Corner detection based on adaptively filtered curvature function [J]. Electronics Letters, 2003, 39(5):426-428.
[45]D J Langride. Curve encoding and the detection of discontinuities [J]. CVGIf,1982,20(1):58-71.
[46]陈燕新,戚飞虎.一种新的提取轮廓特征点的方法[J].红夕屿毫米波学报,1998, 17(3):171-176.
[47]贺双喜,张少军,夏经亮.应用弧长提取轮廓特征点的方法[J].光电工程,2004,31(11):59-62.
[48]Sheu H T, et al.. Rotationally invariant two-phase scheme for corner detection [J]. Pattern Recognition, 1996, 29(5):819-828.
[49]Tsai D M. Boundary-based corner detecting using neural networks [J]. Pattern Recognition, 1997, 30(1):85-97.
[50]屈玉福,浦昭邦,王亚爱.视觉检测系统中亚像素检测技术的对比研究[J].仪器仪表学报,2003,8:460-462.
[51]苏翼雄,贺忠海等.利用二次曲线拟合CCD图像亚像素提取算法[J].计量计数,2003(7).
[52]汪传民,叶邦彦,黄先德,黄昌正.基于计算机视觉的轴承外径检测系统的研究[J].微计算机信息,2006,22(5-1):205-207.
[2]章毓晋.图象理解与计算机工程[M].北京:清华大学出版社,2000.
[3]朱森良.计算机视觉[M].浙江大学出版社,1997.
[4] D.H. Ballard, C.M. Brown. Computer vision[M].Prentice-Hall, Englewood Cliffs,New Jersey,1982.
[5]贾云得.机器视觉[M].北京:科学出版社,2000,4.
[6] F. Keceki, H. H. Nagel. Machine vision based estimation of pose and size parameters from a genetic work piece description [J]. IEEE Tran On robotics an Automation, 2001, 3:2159-2164.
[7] Eric P. Krotkov. Visual Hyperacuity: Representation and Computation of High Precision Position Information [J]. COMPUTER VISION, GRAPHICS AND IMAGE PROCESSING, 1986, 33:99-115.
[8] S.Sirkis, T. J. Lim, Displacement and Strain Measurement with automated Grid Methods [J]. Experimented Mechanics, December 1991:382-388.
[9]张文景,张文渊,苏键峰等.计算机视觉检测技术及其在机械零件检测中的应用.上海交通大学学报[J],1999, 33(5):635-638.
[10]邹定海,叶声华,王春和.用于在线测量的视觉检测系统.仪器仪表学报[J],1995, 16(4):337-340.
[11]阮秋琦.数字图像处理学[M].电子工业出版社,2001.
[12]杨淑莹,边奠英.VC++图像处理程序设计[M].清华大学出版社,2005.
[13]夏良正.数字图像处理[M].南京:东南大学出版社,1999.
[14]于起峰,陆宏伟,刘肖琳.基于图像的精密测量与运动测量[M].北京:科学出版社,2002.
[15]于殿泓.图像检测与处理技术[M].西安电子科技大学出版社,2006.
[16]Shi Hongli, Cai Yuanli, Qiu Zulian. Adaptive wavelet domain thresholding denoising. In: Proceedings of SPIE-The International Society for Optical Engineering, Fifth International Symposium on Instrumentation and Control Technology. 2003, 5253:245-250.
[17]Byung-Jun Yoan, Vaidyanathan P P. Wavelet-based denoising by customizedthresholding. In: IEEE International Conference on Acoustics, Speech and SignalProcessing.2004:925-928
[18]Coyle Edward J, Lin Jean-Hsang. Stack filters and the mean absolute error criterion.IEEE Transactions on Acoustics, Speech and Signal Processing. 1988, 36(8):1244-1254
[19]Haritopoulos Michel,Yin Hujun, Allinson Nigel M. Image denoising using self-organizing map-based nonlinear independent component analysis, Neural Networks.2002, 15(8-9):1085-1098.
[20]杨淑莹,李兰友.图像模式识别——VC++技术实现[M].清华大学出版社,2005.
[21]杨枝灵,王开.Visual C++数字图像获取、处理及实践应用[M].人民邮电出版社,2003.
[22]Philippe Salembier. Morphological multiscale segmentation for image coding[J].Signal Processing,1994,38:359-386.
[23]刘立波.图像分割方法探讨[J].宁夏农学院学报,2001,22(4):51-54.
[24](美)冈萨雷斯著,阮秋琦,阮宇智译.数字图像处理[M].电子工业出版社,2007.
[25]张凯丽,刘辉.边缘检测技术的发展研究[J].昆明理工大学学报,2000,25(5): 36-39.
[26]李庆利.尺寸测量中的边缘检测算法研究[J].计算机测量与控制,2004,12(4):334-337.
[27]何斌.Visual C++数字图象处理[M].北京:人民邮电出版社,2001.
[28]Canny J. A computational approach to edge detection [J]. IEEE-PAMI, 1986, 8:679-698.
[29]余松煜,周源华,张瑞.数字图像处理[M].上海交通大学出版社,2007.
[30]周长发.精通Visual C++图像处理编程[M].北京:电子工业出版社,2006,6.
[31]求是科技. Visual C++图像处理典型算法及实现[M].人民邮电出版社,2006,6.
[32]Stan Z Li, Lie Gu. Kernel machine based learning for mufti-view face detection and pose estimation [J]. Technical Report MSR TR-2001-07.2001:674-679.
[33]林大钊.计算机工程图形算法与应用[M].华东理工大学出版社,2006.
[34]王锁明,侯彬,陈泽珍.最小二乘法直线拟合数据处理的VB程序设计[J].电脑知识与技术,2007,8:782-783.
[35]俞巧云,邢晓正.直线拟合方法在一维图像边缘检测中的应用[J].光学工程,2001,28(6):56-58,65.
[36]郝宏文.圆[M].大象出版社,2004,2.
[37]Illingworth, J. and J. Kittler. A survey of the Hough Transform [J]. Computer Vision Graphics, and Image Processing, 1988(44): 87-116.
[38]Tsuji, S. and F. Matsumoto. Detection of ellipses by modified Hough transformation [J].IEEE transaction on Computers, 1978(27): 777-781.
[39]Wechsler, H. and J. Sklansky. Automatic detection of ribs in chest radiographs [J]. Pattern Recognition, 1977(9): 21-30.
[40]陈传波,金先级.数字图像处理[M].机械工业出版社,2004,7.
[41]F.Mokhtarian, R.Suomela. Robust image corner detection through curvature scale space [J].IEEE Trans. on Pattern Analysis and Machine Intelligence. 1998, 20(12):1376-1381.
[42]A Bandera, C Urdiales, et al.. Corner detection by means of adaptively estimated curvature function [J]. Electronics Letters, 2000, 36:124-126.
[43]P Reche, C Urdiales, et al.. Corner detection by means of contour local vectors [J].Electronics Letters. 2002, 38(14):699-701.
[44]C Urdiales, C Trazegnies, et al.. Corner detection based on adaptively filtered curvature function [J]. Electronics Letters, 2003, 39(5):426-428.
[45]D J Langride. Curve encoding and the detection of discontinuities [J]. CVGIf,1982,20(1):58-71.
[46]陈燕新,戚飞虎.一种新的提取轮廓特征点的方法[J].红夕屿毫米波学报,1998, 17(3):171-176.
[47]贺双喜,张少军,夏经亮.应用弧长提取轮廓特征点的方法[J].光电工程,2004,31(11):59-62.
[48]Sheu H T, et al.. Rotationally invariant two-phase scheme for corner detection [J]. Pattern Recognition, 1996, 29(5):819-828.
[49]Tsai D M. Boundary-based corner detecting using neural networks [J]. Pattern Recognition, 1997, 30(1):85-97.
[50]屈玉福,浦昭邦,王亚爱.视觉检测系统中亚像素检测技术的对比研究[J].仪器仪表学报,2003,8:460-462.
[51]苏翼雄,贺忠海等.利用二次曲线拟合CCD图像亚像素提取算法[J].计量计数,2003(7).
[52]汪传民,叶邦彦,黄先德,黄昌正.基于计算机视觉的轴承外径检测系统的研究[J].微计算机信息,2006,22(5-1):205-207.