彩色印品显微网点图像分色算法的研究
|
摘要
针对当前印刷领域中印刷图像色彩处理方式的稳定性及存在的不足，本课题提出了一种新的基于数字图像处理和模式识别技术的色彩处理方式，即印刷色彩空间转换及分色原理，同时设计出了一个彩色印品显微网点图像分色系统。分色系统借助CCD数码相机和光学显微镜等设备，获取彩色印品显微网点图像的RGB的色彩数据信息；而后采用数字图像处理和模式识别技术对网点图像中各像素进行颜色模式识别，最终得到各单色网点图像的分色结果以及各色网点面积率。分色系统中算法模型的设计充分体现了当今人工智能方面的前沿技术，如神经网络，信息融合等。具体构建了为便于识别的两个分类知识库，即初分类知识库和细分类知识库，这些分类器的设计综合了模糊逻辑，聚类算法、神经网络、以及信息融合等方面的技术应用，使分色系统具有识别精度比较高及速度较快的优点。课题对各种分类器的构建过程及工作原理进行了详尽的阐述，对分色结果进行了总结归纳和误差分析，同时指出了后续研究方向。
According to the drawbacks and stability of algorithms in color processing of printing images in current printing industry, a new method for transforming color spaces and color separation based on digital image processing and pattern recognition is presented, and a system of color printing micro-dot image has been designed . With CCD digital camera and optical microscope, color printing micro-dot image RGB data information can be got, then adopt the digital image processing and pattern recognition techniques to recognize the color patterns of pixels in the micro-dot images, at the same time, the color separation and area coverage can be retrieved.
The design of algorithms model fully embodies the advanced techniques in the field of AI, such as neural network, information fusion. In the color separation system, two classifying knowledge libraries have been established, namely, one is the rough knowledge library, and the other is the fine knowledge library. The two classifying libraries have comprehensively applied the fuzzy logical, clustering algorithms, neural network and information fusion, made the classifiers have the merits of high degree recognition and fast speed. The task has introduced the construction procedure and principals of the classifiers in details, analyzed the errors and summed up the color separation results. In the end, it also indicated the direction of study in the future.
According to the drawbacks and stability of algorithms in color processing of printing images in current printing industry, a new method for transforming color spaces and color separation based on digital image processing and pattern recognition is presented, and a system of color printing micro-dot image has been designed . With CCD digital camera and optical microscope, color printing micro-dot image RGB data information can be got, then adopt the digital image processing and pattern recognition techniques to recognize the color patterns of pixels in the micro-dot images, at the same time, the color separation and area coverage can be retrieved.
The design of algorithms model fully embodies the advanced techniques in the field of AI, such as neural network, information fusion. In the color separation system, two classifying knowledge libraries have been established, namely, one is the rough knowledge library, and the other is the fine knowledge library. The two classifying libraries have comprehensively applied the fuzzy logical, clustering algorithms, neural network and information fusion, made the classifiers have the merits of high degree recognition and fast speed. The task has introduced the construction procedure and principals of the classifiers in details, analyzed the errors and summed up the color separation results. In the end, it also indicated the direction of study in the future.
引文
[1] 阮秋琦编著．数字图像处理学．北京：电子工业出版社，200l，1
[2] 孙即祥等著．模式识别中的特征提取与计算机视觉变量．北京：国防工业出版社，2001
[3] 沈清，汤霖编著．模式识别导论．长沙：国防科技大学出版社，1991，5
[4] Henry R.Kang. Color Technology for Electronic Imaging Devices. SPIT Optical Engineering Press, 1997
[5] 郭茂祖，王亚东，苏小红，王兴起。基于BP网络的色彩匹配方法研究。计算机学报，2000，23(8)：819-823
[6] 王永刚，彩色印刷图像网点检测与识别算法的研究．西安理工大学硕士学位论文，2002，3
[7] Meir Deutsch, Javier Garcia, and David Mendlovic. Multichannel single-output color pattern recognition by use of a joint-transformation correlator. Applied Optics, 1996, 35(35):6976-6982
[8] 史瑞芝．用查表插值法实现CIELab与CMYK颜笆空间的转换．测绘学院学报．2002，19(3)：224-226
[9] 杨旭等．彩色地图图像分色的研究及算法实现．测绘学院学报，2001，18(2)：127-129
[10] 余扬，黄唯一．光电混合处理系统识别高相似度工业零件图像的研究初探．中国图像图形学报，2002，7(A)(7)：633-637
[11] 潘晨，顾峰．基于3D直方图的彩色图像分割方法．中国图像图形学报，2002，7(A)(8)：800-805
[12] Kurngollu F., Sanku B, Harmanci A. E. Color image segmentation using histogram multithresholding and fusing. Image and Vision Computing. 2001,19(4):915-928
[13] J.S.Arney, P.G.Engeldrum and H.Zeng. An Expanded Murray-Davies Model of Tone Reproduction in Halftone Imaging. Journal of Imaging Science and Technology, 1995, 39:502～508
[14] Pobborovsky I, Pearson M (TAGA》, 1972. P65
[15] 崔元日，杨力等．网点印刷颜色再现物理模型．郑州测绘学院学报，V01．11．No．1，1994．
[16] David R.Wyble, Roy S.Berns. A Critical Review of Spectral Models Applied to Binary Color Printing. Color Research and Application, 2000, 25：4～19
[17] Henry R.Kang. Color Technology for Electronic Imaging Devices. SPIT Optical Engineering Press, 1997
[18] 史瑞芝．用查表插值法实现ClELab与CMYK颜色空间的转换．测绘学院学报，2002，19(3)：224-226
[19] M.L. Pearson. The use of the color gamut of an ink set in the selection of process inks. TAGA Proceedings 1964
[20] J.A.C. Yule. Principles of Color Reproduction. John Wiley&Sons.Inc. 1967
[21] 刘真，崔虎平．地图电子出版系统颜色空间转换模型的研究．测绘学报，2000，29(4)：342-347
[22] David R.Wyble, Roy S.Berns. A Critical Review of Spectral Models Applied
to Binary Color Printing. Color Research and Application, 2000, 25: 4～19
[23] 刘世吕，于开龙．纽介堡方程组及尤耳-尼尔森公式中修正系数(n)值的检测与计算．中国印刷，1991(32)：39～45
[24] 刘世吕，董明达．近距优先匹配理论与组分油墨网点覆盖率唯一最优解集的确定．中国印刷，1994(48)：20～27
[25] 阎平凡，张长水．人工神经网络与模拟进化计算．北京：清华大学出版社，2000
[26] 色彩学编写组。色彩学。北京：科学技术出版社，2001．9
[27] 章毓晋著．图像分割．北京：科学出版社，2001
[28] 李介谷，蔡国廉编著．计算机模式识别技术．上海：上海交大出版社，1986，2
[29] 孙圣和，陆哲明．矢量量化技术及应用．北京：科学出版社，2002
[30] 叶秀清等．快速模糊分割算法．模式识别与人工智能，1996，9(3)
[31] 孙圣和，陆哲明，矢量量化技术及应用，科学出版社，2002
[32] Miller W T, Glanz H F, CMAC:An associative neural network alternative to back-propation[J].ProcIEEE 1990,78:1561-1567
[33] 阎平凡，张长水．人工神经网络与模拟进化计算．北京：清华大学出版社，2000
[34] 王俊松等．基于NIF-CMAC的系统参数和智能辨识．天津职业技术学院学报，2003，13(2)：12-14
[35] 王俊松．基于联想记忆的机器人智能控制系统[D]．北京：北京工业大学，1998
[36] Huang-Ching Lu and Jui-Chi Chang, Enhance The Performance Of CMAC Neural Network Via Theory And Credit Apportionment. IEEE Transactions on Neural Networks, Vol.7, 196-202,1995
[37] XU Ning-Shou, A new content-addressing mechanism of CMAC associative memory system[R]. Proc of The 13th IFAC World Congress, USA: 1996,435-439
[38] Selahattin Sayil, Kang Y.Lee. A Hybrid Maximum Error Algorithm with Neighborhood Training for CMAC, Proc. International Joint Conference on Neural Networks, 2583-7, 2002
[39] 信息融合 神经网络-模糊推理理论与应用，权太范。北京：国防工业出版社，2002，8
[40] 机器学习，(美)Tom M.Mitchell著；曾华军，张银奎等译。北京：机械工业出版社，2003，1
[41] Liggins M E et al. Distrubited fusion architectures and algorithms for target tracking. Proceedings of IEEE, 1997,85(1): 95-107
[42] B.D.Ripley. Pattern Recognition and Neural Networks. Cambridge University Press, 1996
[43] 智能控制理论和方法，李人厚，西安：西安电子科技大学出版社，1999
[44] 刘治等．一种高阶模糊CMAC自适应控制及其应用。自动化学报，2001，27(2)
[2] 孙即祥等著．模式识别中的特征提取与计算机视觉变量．北京：国防工业出版社，2001
[3] 沈清，汤霖编著．模式识别导论．长沙：国防科技大学出版社，1991，5
[4] Henry R.Kang. Color Technology for Electronic Imaging Devices. SPIT Optical Engineering Press, 1997
[5] 郭茂祖，王亚东，苏小红，王兴起。基于BP网络的色彩匹配方法研究。计算机学报，2000，23(8)：819-823
[6] 王永刚，彩色印刷图像网点检测与识别算法的研究．西安理工大学硕士学位论文，2002，3
[7] Meir Deutsch, Javier Garcia, and David Mendlovic. Multichannel single-output color pattern recognition by use of a joint-transformation correlator. Applied Optics, 1996, 35(35):6976-6982
[8] 史瑞芝．用查表插值法实现CIELab与CMYK颜笆空间的转换．测绘学院学报．2002，19(3)：224-226
[9] 杨旭等．彩色地图图像分色的研究及算法实现．测绘学院学报，2001，18(2)：127-129
[10] 余扬，黄唯一．光电混合处理系统识别高相似度工业零件图像的研究初探．中国图像图形学报，2002，7(A)(7)：633-637
[11] 潘晨，顾峰．基于3D直方图的彩色图像分割方法．中国图像图形学报，2002，7(A)(8)：800-805
[12] Kurngollu F., Sanku B, Harmanci A. E. Color image segmentation using histogram multithresholding and fusing. Image and Vision Computing. 2001,19(4):915-928
[13] J.S.Arney, P.G.Engeldrum and H.Zeng. An Expanded Murray-Davies Model of Tone Reproduction in Halftone Imaging. Journal of Imaging Science and Technology, 1995, 39:502～508
[14] Pobborovsky I, Pearson M (TAGA》, 1972. P65
[15] 崔元日，杨力等．网点印刷颜色再现物理模型．郑州测绘学院学报，V01．11．No．1，1994．
[16] David R.Wyble, Roy S.Berns. A Critical Review of Spectral Models Applied to Binary Color Printing. Color Research and Application, 2000, 25：4～19
[17] Henry R.Kang. Color Technology for Electronic Imaging Devices. SPIT Optical Engineering Press, 1997
[18] 史瑞芝．用查表插值法实现ClELab与CMYK颜色空间的转换．测绘学院学报，2002，19(3)：224-226
[19] M.L. Pearson. The use of the color gamut of an ink set in the selection of process inks. TAGA Proceedings 1964
[20] J.A.C. Yule. Principles of Color Reproduction. John Wiley&Sons.Inc. 1967
[21] 刘真，崔虎平．地图电子出版系统颜色空间转换模型的研究．测绘学报，2000，29(4)：342-347
[22] David R.Wyble, Roy S.Berns. A Critical Review of Spectral Models Applied
to Binary Color Printing. Color Research and Application, 2000, 25: 4～19
[23] 刘世吕，于开龙．纽介堡方程组及尤耳-尼尔森公式中修正系数(n)值的检测与计算．中国印刷，1991(32)：39～45
[24] 刘世吕，董明达．近距优先匹配理论与组分油墨网点覆盖率唯一最优解集的确定．中国印刷，1994(48)：20～27
[25] 阎平凡，张长水．人工神经网络与模拟进化计算．北京：清华大学出版社，2000
[26] 色彩学编写组。色彩学。北京：科学技术出版社，2001．9
[27] 章毓晋著．图像分割．北京：科学出版社，2001
[28] 李介谷，蔡国廉编著．计算机模式识别技术．上海：上海交大出版社，1986，2
[29] 孙圣和，陆哲明．矢量量化技术及应用．北京：科学出版社，2002
[30] 叶秀清等．快速模糊分割算法．模式识别与人工智能，1996，9(3)
[31] 孙圣和，陆哲明，矢量量化技术及应用，科学出版社，2002
[32] Miller W T, Glanz H F, CMAC:An associative neural network alternative to back-propation[J].ProcIEEE 1990,78:1561-1567
[33] 阎平凡，张长水．人工神经网络与模拟进化计算．北京：清华大学出版社，2000
[34] 王俊松等．基于NIF-CMAC的系统参数和智能辨识．天津职业技术学院学报，2003，13(2)：12-14
[35] 王俊松．基于联想记忆的机器人智能控制系统[D]．北京：北京工业大学，1998
[36] Huang-Ching Lu and Jui-Chi Chang, Enhance The Performance Of CMAC Neural Network Via Theory And Credit Apportionment. IEEE Transactions on Neural Networks, Vol.7, 196-202,1995
[37] XU Ning-Shou, A new content-addressing mechanism of CMAC associative memory system[R]. Proc of The 13th IFAC World Congress, USA: 1996,435-439
[38] Selahattin Sayil, Kang Y.Lee. A Hybrid Maximum Error Algorithm with Neighborhood Training for CMAC, Proc. International Joint Conference on Neural Networks, 2583-7, 2002
[39] 信息融合 神经网络-模糊推理理论与应用，权太范。北京：国防工业出版社，2002，8
[40] 机器学习，(美)Tom M.Mitchell著；曾华军，张银奎等译。北京：机械工业出版社，2003，1
[41] Liggins M E et al. Distrubited fusion architectures and algorithms for target tracking. Proceedings of IEEE, 1997,85(1): 95-107
[42] B.D.Ripley. Pattern Recognition and Neural Networks. Cambridge University Press, 1996
[43] 智能控制理论和方法，李人厚，西安：西安电子科技大学出版社，1999
[44] 刘治等．一种高阶模糊CMAC自适应控制及其应用。自动化学报，2001，27(2)