小波变换在套件组装视觉检测中的应用
|
摘要
随着国民经济和计算机技术的高速发展，计算机视觉已经成为一门内容丰富的综合性学科，它在现代工业、国防、医学、空间技术等领域有着广泛的应用前景。计算机视觉在实际应用中很大程度上依赖于数字图像处理及模式识别技术。本文所叙述的正是基于计算机视觉的课题——小波变换在套件组装视觉检测中的应用，而且，这一课题来源于生产实践。在生产套件产品的流水线上，套件产品的组装可能会出现漏装和错位的情况。为了及时有效地检测出产品组装的错误，在课题中利用小波变换对套件产品图像进行处理，并识别出上述两种组装错误。本文详细叙述了从套件产品的图像输入到最后识别的全部过程。其中，套件组装视觉检测的基本框架如下：第一步，用CCD摄像机将套件产品的彩色图像输入计算机中，以便于进行数字图像处理；第二步，为了使图像处理过程更加简便，将套件产品彩色图像变成灰度图像。另外，为了增强处理的实时性，将灰度图像缩小；第三步，为了减轻光线不均匀对灰度图像造成的影响，对产品图像进行灰度直方图均衡；第四步，构造小波滤波器组，并利用小波滤波器组对产品图像实现二维可分离小波变换；第五步，在小波变换的基础上得到套件产品的梯度图像，然后根据图像梯度模值的极大值提取图像边沿；最后，用一维小波变换提取图像边沿上的犄角点，并利用犄角点之间的联系实现对套件产品的识别。
With the high speed development of the social economy and computer technology, computer vision has become a synthetical subject and has a bright future for wide appliances in such fields as modern industry, national defense, medicine and space technology. Its appliances to great extent depends on digital image processing and
pattern recognition. This thesis is about the project based on computer vision ------
applications of wavelet transform to visual inspection of assembling of the manufacture kit. The subject comes from practice. On the assembly line of producing the manufacture kit there is the possibility that failure of assembly and malposition of manufacture kit would happen. In order to examine the errors in assembling the manufacture timely and effectively, in this subject some treatments are given to the image of the manufacture kit through the wavelet transform and the manufacture kit through the wavelet transform and the errors are recognized. The thesis details the whole process from the input of the image of the manufacture kit to the final recognition. The procedure of the visual inspection is as follows: First, the color image of the manufacture kit is put into computer by the CCD pickup camera for the purpose of the digital image processing. Second, the color image of the manufacture kit is converted to grayscale image so as to simplify the image processing. In addition, the size of the grayscale
image is reduced for the improvement of actual effect of the image processing. Third, the gray-level histogram of image is equalized in order to reduce the influence of uneven light to the gray-level image. Fourth, the wavelet filter banks is constructed and the two-dimensional separable wavelet transform of the image is realized by the wavelet filter banks. Fifth the grads image of manufacture kit is gained on the base of wavelet transform and then image edge can be detected with the use of grads module of image. In the end , the inflection points of image edge can be detected with the use of one-dimensional wavelet transform and the recognition of the manufacture kit can be realized by the relation of the inflection points.
With the high speed development of the social economy and computer technology, computer vision has become a synthetical subject and has a bright future for wide appliances in such fields as modern industry, national defense, medicine and space technology. Its appliances to great extent depends on digital image processing and
pattern recognition. This thesis is about the project based on computer vision ------
applications of wavelet transform to visual inspection of assembling of the manufacture kit. The subject comes from practice. On the assembly line of producing the manufacture kit there is the possibility that failure of assembly and malposition of manufacture kit would happen. In order to examine the errors in assembling the manufacture timely and effectively, in this subject some treatments are given to the image of the manufacture kit through the wavelet transform and the manufacture kit through the wavelet transform and the errors are recognized. The thesis details the whole process from the input of the image of the manufacture kit to the final recognition. The procedure of the visual inspection is as follows: First, the color image of the manufacture kit is put into computer by the CCD pickup camera for the purpose of the digital image processing. Second, the color image of the manufacture kit is converted to grayscale image so as to simplify the image processing. In addition, the size of the grayscale
image is reduced for the improvement of actual effect of the image processing. Third, the gray-level histogram of image is equalized in order to reduce the influence of uneven light to the gray-level image. Fourth, the wavelet filter banks is constructed and the two-dimensional separable wavelet transform of the image is realized by the wavelet filter banks. Fifth the grads image of manufacture kit is gained on the base of wavelet transform and then image edge can be detected with the use of grads module of image. In the end , the inflection points of image edge can be detected with the use of one-dimensional wavelet transform and the recognition of the manufacture kit can be realized by the relation of the inflection points.
引文
[1]A. D. brink, Thresholding of digital images using two-dimensional entropice, Pattem Recognition, 1992-25: 8033-808
[2]J-C. olivo, Automatic threshold selection using the wavelet transform, CVGIP, 1994-56: 205～218
[3]M.Shensa. Affine wavelets, Wedding the a'trous and the Mallat algorithm·IEEE Trans·SP-40(10)·2464～2484·1992
[4]Spacek L A·Edge detection and motion detection·Image and Vision Computing, 1986,4(1): 43～56
[5]Wilkinson M H F·Optimizing edge detectors for robust automatic threshold selection: coping with edge curvature and noise GMIP, 1998, 60(4): 385～401
[6]Hueckel M·An operator which locates edges in digital pictures·J·ACM, 1987, 18(1): 113～125
[7]Mallat S·A theory for multiresolution signal decomposition: the wavelet representation·IEEE-PAMI, 1989, 11(7): 647～693
[8]Mallat S,Zhong S·Characterozation of signals fron multiscale edges·IEEE-PAMI, 1992, 14(7): 710～732
[9]Wilson S·Applications of Matrix Morphology·SPIE, vol.1350, pp.44～55, July, 1990
[10]Zhaol D and Daut D·Morphological Hit-Or-Miss Transformation For Shape Recognition·Visual Comunication and Image Representation·vol·2, No.3, September, 1991
[11]郑南宁·计算机视觉与模式识别·国防工业出版社·1998
[12]章毓晋·图像理解与计算机视觉·清华大学出版社·2000
[13]何斌等·VC++数字图像处理·人民邮电出版社·2001
[14]周长发·精通Visual C++图像编程·电子工业出版社·2000
[15]远鹏·计算机图像处理技术基础·北京大学出版社·1996
[16]阮秋琦·数字图像处理基础·中国铁道出版社·1998
[17]杨福生·小波变换的工程应用与分析·科学出版社·1999
[18]杨福生·带通信号的采样定理·信号处理·1986，2(1)，58-61
[19]刘曙光·基于Canny准则的基数B样条小波边缘检测·信号处理·2001·5·vol·17
[20]章毓晋·图像分割·科学出版社·2001
[21]皮明红等·一种多尺度边缘检测的方法·信号处理·2000，vol·16·5～8
[22]辛炜等·车牌自动识别系统中车牌分割研究·信号处理·2000，vol·16·41～44
[23]刘宏兵·利用完全重构滤波器组进行图像的多尺度边缘提取·号处理·2000，vol·16·45～47
[24]王敏等·一种新的自动化多阈值图像分割方法·信号处理·2000，vol·16·90～94
[25]范劲松等·基于粗集理论和SVM算法的模式分类方法·模式识别与人工智能·2000，vol·13·419～412
[26]王信义等·生产系统中的监控检测技术·北京理工大学出版社·1998
[27]虞和济等·基于神经网络的智能诊断·治金工业出版社·2000
[28]董士海等·图象格式编程指南·清华大学出版社·1994
[29]沈清等·模式识别导论·国防科技大学出版社·1991
[30]包约翰[美]·自适应识别与神经网络·科学出版社·1992
[31]边肇棋等·模式识别·清华大学出版社·1999
[32]堰野充[日]·BASIC图像处理程序1.50例·中国科学技术大学出版社·1992
[33]王润生·图像理解·国防科技大学出版社·1995
[34]崔锦泰[美]·小波分析导论·西安交通大学出版社·1995
[35]彭玉华·小波变换与工程应用·科学出版社·1999
[36]赵松年·子波变换与子波分析·电子工业出版社·1996
[37]程正兴·小波分析算法与应用·西安交通大学出版社·1998
[38]徐佩霞等·小波分析与应用实例·中国科学技术大学出版社·1996
[39]秦前清·实用小波分析·西安电子科技大学出版社·1982
[40]英格尔等[美]·数学信号处理MATLAB及其实现·电子工业出版社·1998
[41]卡斯尔曼[美]·数字图像处理·清华大学出版社·1998
[42]冈萨雷斯[美]·数字图像处理·科学出版社·1981
[43]吴健康·数字图像分析·人民邮电出版社·1989
[44]姚天任等·现代数字信号处理·华中理工大学出版社·1999
[45]王振武等·具有二阶消失矩的多进制 Daubechies 型小波及应用·信号处理·2000，vol·16·31～35
[46]张静远·基于小波变换的特征提取方法分析·信号处理·2000，vol·16·156～162
[47]周蓉蓉等·一种基于小波的SAR图象边缘检测方法·计算技术与自动化·2000，vol·19·36～41
[48]陈自东·数学形态和小波分析理论及其在数字图像处理与分析中的应用[D]·国防科技大学博士论文·1998
[49]杜啸晓等·一种新的基于区域和边界的图像分割方法·中国图象图形学报·2001，vol·6·755～758
[50]王波涛等·相对矩及在几何形状识别中的应用·中国图象图形学报·2001，vol·6·296～300
[51]蒋晓瑜等·基于小波变换的多分辨模板匹配·中国图象图形学报·2000，vol·5·304～308
[52]翁文国等·小波变换应用于DPIV图象去噪·中国图象图形学报，2000，vol·5·211～215
[2]J-C. olivo, Automatic threshold selection using the wavelet transform, CVGIP, 1994-56: 205～218
[3]M.Shensa. Affine wavelets, Wedding the a'trous and the Mallat algorithm·IEEE Trans·SP-40(10)·2464～2484·1992
[4]Spacek L A·Edge detection and motion detection·Image and Vision Computing, 1986,4(1): 43～56
[5]Wilkinson M H F·Optimizing edge detectors for robust automatic threshold selection: coping with edge curvature and noise GMIP, 1998, 60(4): 385～401
[6]Hueckel M·An operator which locates edges in digital pictures·J·ACM, 1987, 18(1): 113～125
[7]Mallat S·A theory for multiresolution signal decomposition: the wavelet representation·IEEE-PAMI, 1989, 11(7): 647～693
[8]Mallat S,Zhong S·Characterozation of signals fron multiscale edges·IEEE-PAMI, 1992, 14(7): 710～732
[9]Wilson S·Applications of Matrix Morphology·SPIE, vol.1350, pp.44～55, July, 1990
[10]Zhaol D and Daut D·Morphological Hit-Or-Miss Transformation For Shape Recognition·Visual Comunication and Image Representation·vol·2, No.3, September, 1991
[11]郑南宁·计算机视觉与模式识别·国防工业出版社·1998
[12]章毓晋·图像理解与计算机视觉·清华大学出版社·2000
[13]何斌等·VC++数字图像处理·人民邮电出版社·2001
[14]周长发·精通Visual C++图像编程·电子工业出版社·2000
[15]远鹏·计算机图像处理技术基础·北京大学出版社·1996
[16]阮秋琦·数字图像处理基础·中国铁道出版社·1998
[17]杨福生·小波变换的工程应用与分析·科学出版社·1999
[18]杨福生·带通信号的采样定理·信号处理·1986，2(1)，58-61
[19]刘曙光·基于Canny准则的基数B样条小波边缘检测·信号处理·2001·5·vol·17
[20]章毓晋·图像分割·科学出版社·2001
[21]皮明红等·一种多尺度边缘检测的方法·信号处理·2000，vol·16·5～8
[22]辛炜等·车牌自动识别系统中车牌分割研究·信号处理·2000，vol·16·41～44
[23]刘宏兵·利用完全重构滤波器组进行图像的多尺度边缘提取·号处理·2000，vol·16·45～47
[24]王敏等·一种新的自动化多阈值图像分割方法·信号处理·2000，vol·16·90～94
[25]范劲松等·基于粗集理论和SVM算法的模式分类方法·模式识别与人工智能·2000，vol·13·419～412
[26]王信义等·生产系统中的监控检测技术·北京理工大学出版社·1998
[27]虞和济等·基于神经网络的智能诊断·治金工业出版社·2000
[28]董士海等·图象格式编程指南·清华大学出版社·1994
[29]沈清等·模式识别导论·国防科技大学出版社·1991
[30]包约翰[美]·自适应识别与神经网络·科学出版社·1992
[31]边肇棋等·模式识别·清华大学出版社·1999
[32]堰野充[日]·BASIC图像处理程序1.50例·中国科学技术大学出版社·1992
[33]王润生·图像理解·国防科技大学出版社·1995
[34]崔锦泰[美]·小波分析导论·西安交通大学出版社·1995
[35]彭玉华·小波变换与工程应用·科学出版社·1999
[36]赵松年·子波变换与子波分析·电子工业出版社·1996
[37]程正兴·小波分析算法与应用·西安交通大学出版社·1998
[38]徐佩霞等·小波分析与应用实例·中国科学技术大学出版社·1996
[39]秦前清·实用小波分析·西安电子科技大学出版社·1982
[40]英格尔等[美]·数学信号处理MATLAB及其实现·电子工业出版社·1998
[41]卡斯尔曼[美]·数字图像处理·清华大学出版社·1998
[42]冈萨雷斯[美]·数字图像处理·科学出版社·1981
[43]吴健康·数字图像分析·人民邮电出版社·1989
[44]姚天任等·现代数字信号处理·华中理工大学出版社·1999
[45]王振武等·具有二阶消失矩的多进制 Daubechies 型小波及应用·信号处理·2000，vol·16·31～35
[46]张静远·基于小波变换的特征提取方法分析·信号处理·2000，vol·16·156～162
[47]周蓉蓉等·一种基于小波的SAR图象边缘检测方法·计算技术与自动化·2000，vol·19·36～41
[48]陈自东·数学形态和小波分析理论及其在数字图像处理与分析中的应用[D]·国防科技大学博士论文·1998
[49]杜啸晓等·一种新的基于区域和边界的图像分割方法·中国图象图形学报·2001，vol·6·755～758
[50]王波涛等·相对矩及在几何形状识别中的应用·中国图象图形学报·2001，vol·6·296～300
[51]蒋晓瑜等·基于小波变换的多分辨模板匹配·中国图象图形学报·2000，vol·5·304～308
[52]翁文国等·小波变换应用于DPIV图象去噪·中国图象图形学报，2000，vol·5·211～215