图像特征在印鉴识别中的应用
|
摘要
印鉴作为一种个人、企事业单位、社会团体、政府部门乃至整个国家的具有法律意义的标志和论据，在现实生活中占有重要的地位。利用原始的手工核对逐渐暴露出一些不足之处，其正确率受人为因素和各种客观条件的制约，且核对效率低。因此开展印鉴自动识别技术的研究，具有重要的理论意义和广泛的实用价值。
本文提出了一套印鉴预处理方案。首先将输入所得到的图像RGB值转换成对应的HSI值，其中根据印鉴图像的颜色分布特点，提出一个简化的H计算公式，以快速提取出红色印鉴图像。再采取迭代自动阈值法实现印鉴图像的灰度二值化，以得到较好的处理速度和效果。然后依据对图像缺失原因的分析，提出了一种图像修复方法。最后采取开运算和部分中值滤波对印鉴图像进行去除书写线和噪声。实验结果表明，这套方案能够获得较好的提取效果和处理速度。
本文重点对图像的常用特征进行分析，并在矩理论的基础上提出了一种基于伪Zernike矩不变量的鉴别方法。首先通过对印鉴图像进行规格化来获得平移和缩放的不变性，然后提取印鉴图像的伪Zernike矩不变量作为特征。该方法不需要对印鉴进行配准，对印鉴形状也无要求，与基于Hu矩不变量的鉴别方法相比，它利用正交矩，具有更小的冗余度信息，提高识别效率；与基于Zernike矩不变量的鉴别方法相比，它具有更多的高阶信息以对图像细节进行描述，提高识别准确度。
最后，对所提出的方法采用matlab7.0进行仿真，实现印鉴标准样本库建立、印鉴预处理、印鉴特征提取、印鉴分类判决的整套基于伪Zernike矩不变量的印鉴识别方案，并对大量的印鉴图像进行实验。证明了与Hu矩不变量、Zernike矩不变量的鉴别方法相比，该方案提高了印鉴识别的准确率。
Seal plays an important role in our life as the symbol of the individual enterprise social party government and even the whole country in the law. The old manual method of seal verification is discovered with many defects. Its correction is restricted by the subject factors and many kinds of subject conditions. Also its efficiency is low. Thus, the study of the automatic seal verification has the important theory meaning and wide application.
In this paper, a complete seal extraction scheme is proposed. Firstly, the RGB information of the input image is transformed to the corresponding HSI information by the simplified formula which is put forward according to the characters of the color distribution of image to extract the red seal image quickly. Secondly, the alternate threshold method is adopted to get the binary image of the seal having the good procession speed and result. Thirdly by the analysis of the reason of the loss in image, a method to renovate image is proposed. At last, the writing line and noises are removed by the open and smooth operation. The experiment result indicates a better extraction effect and procession speed.
By the analysis of the common image features, a new verification method is proposed on the basis of the Pseudo-Zernike moment invariants. The Pseudo-Zernike moment invariants are extracted as features after the normalization of the seals translation and scaling. This method can be implemented without rotation compensation and is not restrained by the shape of the seal. Compare with the Hu moment invariants-based seal verification method, it makes use of the orthogonal moment invariants having less redundancies and improved recognition rate. And compared with the Zernike moment invariants-based seal verification method, it has more high order moments to describe the detail of the image and to improve the correction of the recognition.
Finally, the arithmetic supposed above is simulated by matlab7.0, and the whole scheme which includes the establishment of the standard seal characteristics the preprocessing of the seal the extraction of the seal characteristics and the classification of the seal is realized based on the Pseudo-Zernike moment invariants. Through the experiment on a mass of seal images, the improved correction rate of the recognition is testified in compare with the method based on Hu moment invariants or Zernike moment invariants.
本文提出了一套印鉴预处理方案。首先将输入所得到的图像RGB值转换成对应的HSI值，其中根据印鉴图像的颜色分布特点，提出一个简化的H计算公式，以快速提取出红色印鉴图像。再采取迭代自动阈值法实现印鉴图像的灰度二值化，以得到较好的处理速度和效果。然后依据对图像缺失原因的分析，提出了一种图像修复方法。最后采取开运算和部分中值滤波对印鉴图像进行去除书写线和噪声。实验结果表明，这套方案能够获得较好的提取效果和处理速度。
本文重点对图像的常用特征进行分析，并在矩理论的基础上提出了一种基于伪Zernike矩不变量的鉴别方法。首先通过对印鉴图像进行规格化来获得平移和缩放的不变性，然后提取印鉴图像的伪Zernike矩不变量作为特征。该方法不需要对印鉴进行配准，对印鉴形状也无要求，与基于Hu矩不变量的鉴别方法相比，它利用正交矩，具有更小的冗余度信息，提高识别效率；与基于Zernike矩不变量的鉴别方法相比，它具有更多的高阶信息以对图像细节进行描述，提高识别准确度。
最后，对所提出的方法采用matlab7.0进行仿真，实现印鉴标准样本库建立、印鉴预处理、印鉴特征提取、印鉴分类判决的整套基于伪Zernike矩不变量的印鉴识别方案，并对大量的印鉴图像进行实验。证明了与Hu矩不变量、Zernike矩不变量的鉴别方法相比，该方案提高了印鉴识别的准确率。
Seal plays an important role in our life as the symbol of the individual enterprise social party government and even the whole country in the law. The old manual method of seal verification is discovered with many defects. Its correction is restricted by the subject factors and many kinds of subject conditions. Also its efficiency is low. Thus, the study of the automatic seal verification has the important theory meaning and wide application.
In this paper, a complete seal extraction scheme is proposed. Firstly, the RGB information of the input image is transformed to the corresponding HSI information by the simplified formula which is put forward according to the characters of the color distribution of image to extract the red seal image quickly. Secondly, the alternate threshold method is adopted to get the binary image of the seal having the good procession speed and result. Thirdly by the analysis of the reason of the loss in image, a method to renovate image is proposed. At last, the writing line and noises are removed by the open and smooth operation. The experiment result indicates a better extraction effect and procession speed.
By the analysis of the common image features, a new verification method is proposed on the basis of the Pseudo-Zernike moment invariants. The Pseudo-Zernike moment invariants are extracted as features after the normalization of the seals translation and scaling. This method can be implemented without rotation compensation and is not restrained by the shape of the seal. Compare with the Hu moment invariants-based seal verification method, it makes use of the orthogonal moment invariants having less redundancies and improved recognition rate. And compared with the Zernike moment invariants-based seal verification method, it has more high order moments to describe the detail of the image and to improve the correction of the recognition.
Finally, the arithmetic supposed above is simulated by matlab7.0, and the whole scheme which includes the establishment of the standard seal characteristics the preprocessing of the seal the extraction of the seal characteristics and the classification of the seal is realized based on the Pseudo-Zernike moment invariants. Through the experiment on a mass of seal images, the improved correction rate of the recognition is testified in compare with the method based on Hu moment invariants or Zernike moment invariants.
引文
[1] Ueda K, Nakamura Y. Automatic Verification of Seal-Impression Pattern. In:Proceedings of the 7th Int. Conf. PR, 1984.1019-1021
[2] Fan T. J, Tsai W. S. Automatic Chinese Seal Identification. Computer Vision, Graphics and Image Processing, 1984.25:311-330
[3] Yung-Sheng Chen. Computer processing on the identification of a Chinese seal image. Document Analysis and Recognition. 1995.1:422-425
[4] LEE S, KIM J H. Unconstrained seal imprint verification using Attributed Stroke Graph Matching. Pattern Recognition, 1989.22(6): 653-654
[5] 高文，董胜富等．基于边缘匹配的印鉴自动鉴别方法．模式识别与人工智能，1994．7(4)：338～342
[6] Y.Yang, T.Horiuchi, K.Toraichi. Automatic Seal identification using Fluency Function Approximation and Relaxation Matching Method. In: Proc 2th IAPR Conf on Document Anal and Recog, 1993.786～789
[7] 胡庆，杨静宇等．基于知识的印鉴鉴别方法．自动化学报，1991．17(6)：696～704
[8] Chen Y.S. Automatic Identification for a Chinese Seal Image. Pattern Recognition. 1996.29(1): 1807-1820
[9] 陈燕新，李卫东，戚飞虎．基于边缘和模板匹配的印鉴自动鉴别方法．上海交通大学学报，1998．32(4)：44～48
[10] 曾东武，刘重庆，陶世焰．Fourier描述符的印鉴识别方法．上海交通大学学报，1998．32(9)：43～45
[11] 张先萌，罗安玉等．用矩不变量实现印鉴自动识别．电子学报，1995．23(4)：100～102
[12] 刘重庆，曾东武，王福东．在印鉴识别中的配准方法．上海交通大学学报，1998．32(9)：48～49
[13] 章毓晋．图像处理与分析．北京：清华大学出版社，2003．19～22
[14] 彭任明．基于彩色的印鉴图像分割技术．甘肃科技，2005．21(9)：24～26
[15] Zhang Y J, Liu Z W, Color image segmentation based on HSI model, High Techniques Letters, 1998b
[16] Smith A R. Color gamut transform pairs. Computer Graphics, 1978.12(3): 12～19
[17] 阴炳皓，赵臣，韩晓军．基于改进的HSI空间模型的目标搜索方法．河北工业大学学报，2003．32(1)：6～10
[18] 赵永涛，李志敏，王洪建，陈志云，王林．印章识别中的图像预处理研究．仪器仪表学报，2004．25(4)：401～404
[19] Gonzalez R.C, Woods R. E. Digital Image Processing. New Jersey :Prentice Hall, 2002
[20] Otsu N. A Threshold Selection Method from Gray-Level Histograms. IEEE Transtractions on Systems, Man and Cybernetics, 1979.9(1): 62～67
[21] Rafael C.Gonzalez, Richard E. Woods, Steven L.Eddins. Digital Image Processing Using MATLAB. New Jersey: Prentice Hall, 2005
[22] 崔屹．图像处理与分析-数学形态学方法及应用．北京：科学出版社，2000．20～40
[23] 张玲，郭磊民，何伟，陈丽敏．鉴印系统中印章的提取与修复．重庆大学学报，2005．28卷(3)：84～87
[24] 洪子泉，杨静宇．统计模式识别中的特征提取．数据采集与处理，1991．6(2)：38-43
[25] N. Kiryati. Calculating Geometic Properties of Objects Represented by Fourier Coefficients, Proc. In: The Computer Soc. Conf. On Computer Vision and Pattern Recognition, 1988. 641-646
[26] C. T. Zach and R.T. Poskies. Fourier Descriptors for Plane Closed Curves. IEEE Trans on Computers, 1972.21: 269-281
[27] E. Persoon and K. S. Fu. Shape Discrimination Using Fourier Descriptors. IEEE Trans on Syst Man Cybern, 1977.7(3): 388-397
[28] Hu M. Visual Pattern Recognition by Moment Invariant. IRE Trans on Inf.Theory. 1962.8:179-187
[29] S.A.Dudani. Aircraft Identification by Moment Invariants. IEEE Trans on Computers, 1977.C-26(1): 39-45
[30] Pasaltis. Image Normalization by Complex Moments. IEEE Trans on PAMI. 1985.7(1): 46-55
[31] M K Hu. Pattern Recognition by Moment Invariants. IEEE Transactions on Information Theory, 1962. 17 (8):179～187
[32] Prokopand R J, Reeves A P. A Survey of Moment-Based techniques for Unconcluded Object Representation and Recognition. Graphics Models and Image Processing, 1992.54 (5): 438～460
[33] 王晓红．矩技术及其在图像处理与识别中的应用．：[硕士学位论文]．西安：西北工业大学计算机应用技术，2001
[34] 商立群，杜亚娟．Hu矩和Zernike矩在图象识别中的应用．西安科技学院学报，2000．20(1)：53～56
[35] M.Teague. Image Analysis via the General Theory Moments. Journal of Optical Society of American, 1980.70(3): 920-930
[36] 叶斌，彭嘉雄．伪Zernike矩不变性分析及其改进研究．中国图像图形学报，2003．8(3)：246～252
[37] Khotanzad A, Hong Y H. Rotation Invariant Image Recognition Using Features Selected Via a Systematic Method. Pattern Recognition, 1990.23(10): 1089-1101
[38] C H Teh and R T Chin. On Image Analysis by the Method of Moments. IEEE Trans. Pattern Analysis and Machine Intelligence, 1988.10(4): 496-513
[39] Belkasim S O, Shridhar M, Ahmadi M. Pattern Recognition with Moment Invariants: A Comparative Study and new results. Pattern Recognition, 1991.24(12): 1117-1138
[40] 孙即祥，王晓华等．模式识别中的特征提取与计算机视觉不变量．北京：国防工业出版社，2001．10～90
[41] 杨峰，周洪利．印鉴自动识别系统中模式识别方法的研究．现代计算机，2003．155：6～9
[42] 杨新军，王肇折，刘唯一，母国光．基于环投影模板匹配的印鉴鉴别法，光电子·激光，2001．12(11)：1177～1181
[43] 胥勋涛．印鉴自动识别系统算法研究：[硕士学位论文]．重庆：重庆大学，2004
[44] Wen Gao, Shengfu Dong, Xilin Chert. A System for Automatic Chinese seal Imprint Verification.Third International Conference on Document Analysis and Recognition. Canada: Montreal, 1995.660～664
[45] 廖闻剑，成瑜．银行印鉴的模糊识别方法．南京航空航天大学学报，1998．30(6)：663～667
[46] 阮秋琦．数字图像处理学．北京：电子工业出版社，2001．483～487
[47] 张淼．基于矩不变量的印鉴鉴别技术的研究：[硕士学位论文]．西安：西安电子科技大学通信与信息系统专业，2005
[2] Fan T. J, Tsai W. S. Automatic Chinese Seal Identification. Computer Vision, Graphics and Image Processing, 1984.25:311-330
[3] Yung-Sheng Chen. Computer processing on the identification of a Chinese seal image. Document Analysis and Recognition. 1995.1:422-425
[4] LEE S, KIM J H. Unconstrained seal imprint verification using Attributed Stroke Graph Matching. Pattern Recognition, 1989.22(6): 653-654
[5] 高文，董胜富等．基于边缘匹配的印鉴自动鉴别方法．模式识别与人工智能，1994．7(4)：338～342
[6] Y.Yang, T.Horiuchi, K.Toraichi. Automatic Seal identification using Fluency Function Approximation and Relaxation Matching Method. In: Proc 2th IAPR Conf on Document Anal and Recog, 1993.786～789
[7] 胡庆，杨静宇等．基于知识的印鉴鉴别方法．自动化学报，1991．17(6)：696～704
[8] Chen Y.S. Automatic Identification for a Chinese Seal Image. Pattern Recognition. 1996.29(1): 1807-1820
[9] 陈燕新，李卫东，戚飞虎．基于边缘和模板匹配的印鉴自动鉴别方法．上海交通大学学报，1998．32(4)：44～48
[10] 曾东武，刘重庆，陶世焰．Fourier描述符的印鉴识别方法．上海交通大学学报，1998．32(9)：43～45
[11] 张先萌，罗安玉等．用矩不变量实现印鉴自动识别．电子学报，1995．23(4)：100～102
[12] 刘重庆，曾东武，王福东．在印鉴识别中的配准方法．上海交通大学学报，1998．32(9)：48～49
[13] 章毓晋．图像处理与分析．北京：清华大学出版社，2003．19～22
[14] 彭任明．基于彩色的印鉴图像分割技术．甘肃科技，2005．21(9)：24～26
[15] Zhang Y J, Liu Z W, Color image segmentation based on HSI model, High Techniques Letters, 1998b
[16] Smith A R. Color gamut transform pairs. Computer Graphics, 1978.12(3): 12～19
[17] 阴炳皓，赵臣，韩晓军．基于改进的HSI空间模型的目标搜索方法．河北工业大学学报，2003．32(1)：6～10
[18] 赵永涛，李志敏，王洪建，陈志云，王林．印章识别中的图像预处理研究．仪器仪表学报，2004．25(4)：401～404
[19] Gonzalez R.C, Woods R. E. Digital Image Processing. New Jersey :Prentice Hall, 2002
[20] Otsu N. A Threshold Selection Method from Gray-Level Histograms. IEEE Transtractions on Systems, Man and Cybernetics, 1979.9(1): 62～67
[21] Rafael C.Gonzalez, Richard E. Woods, Steven L.Eddins. Digital Image Processing Using MATLAB. New Jersey: Prentice Hall, 2005
[22] 崔屹．图像处理与分析-数学形态学方法及应用．北京：科学出版社，2000．20～40
[23] 张玲，郭磊民，何伟，陈丽敏．鉴印系统中印章的提取与修复．重庆大学学报，2005．28卷(3)：84～87
[24] 洪子泉，杨静宇．统计模式识别中的特征提取．数据采集与处理，1991．6(2)：38-43
[25] N. Kiryati. Calculating Geometic Properties of Objects Represented by Fourier Coefficients, Proc. In: The Computer Soc. Conf. On Computer Vision and Pattern Recognition, 1988. 641-646
[26] C. T. Zach and R.T. Poskies. Fourier Descriptors for Plane Closed Curves. IEEE Trans on Computers, 1972.21: 269-281
[27] E. Persoon and K. S. Fu. Shape Discrimination Using Fourier Descriptors. IEEE Trans on Syst Man Cybern, 1977.7(3): 388-397
[28] Hu M. Visual Pattern Recognition by Moment Invariant. IRE Trans on Inf.Theory. 1962.8:179-187
[29] S.A.Dudani. Aircraft Identification by Moment Invariants. IEEE Trans on Computers, 1977.C-26(1): 39-45
[30] Pasaltis. Image Normalization by Complex Moments. IEEE Trans on PAMI. 1985.7(1): 46-55
[31] M K Hu. Pattern Recognition by Moment Invariants. IEEE Transactions on Information Theory, 1962. 17 (8):179～187
[32] Prokopand R J, Reeves A P. A Survey of Moment-Based techniques for Unconcluded Object Representation and Recognition. Graphics Models and Image Processing, 1992.54 (5): 438～460
[33] 王晓红．矩技术及其在图像处理与识别中的应用．：[硕士学位论文]．西安：西北工业大学计算机应用技术，2001
[34] 商立群，杜亚娟．Hu矩和Zernike矩在图象识别中的应用．西安科技学院学报，2000．20(1)：53～56
[35] M.Teague. Image Analysis via the General Theory Moments. Journal of Optical Society of American, 1980.70(3): 920-930
[36] 叶斌，彭嘉雄．伪Zernike矩不变性分析及其改进研究．中国图像图形学报，2003．8(3)：246～252
[37] Khotanzad A, Hong Y H. Rotation Invariant Image Recognition Using Features Selected Via a Systematic Method. Pattern Recognition, 1990.23(10): 1089-1101
[38] C H Teh and R T Chin. On Image Analysis by the Method of Moments. IEEE Trans. Pattern Analysis and Machine Intelligence, 1988.10(4): 496-513
[39] Belkasim S O, Shridhar M, Ahmadi M. Pattern Recognition with Moment Invariants: A Comparative Study and new results. Pattern Recognition, 1991.24(12): 1117-1138
[40] 孙即祥，王晓华等．模式识别中的特征提取与计算机视觉不变量．北京：国防工业出版社，2001．10～90
[41] 杨峰，周洪利．印鉴自动识别系统中模式识别方法的研究．现代计算机，2003．155：6～9
[42] 杨新军，王肇折，刘唯一，母国光．基于环投影模板匹配的印鉴鉴别法，光电子·激光，2001．12(11)：1177～1181
[43] 胥勋涛．印鉴自动识别系统算法研究：[硕士学位论文]．重庆：重庆大学，2004
[44] Wen Gao, Shengfu Dong, Xilin Chert. A System for Automatic Chinese seal Imprint Verification.Third International Conference on Document Analysis and Recognition. Canada: Montreal, 1995.660～664
[45] 廖闻剑，成瑜．银行印鉴的模糊识别方法．南京航空航天大学学报，1998．30(6)：663～667
[46] 阮秋琦．数字图像处理学．北京：电子工业出版社，2001．483～487
[47] 张淼．基于矩不变量的印鉴鉴别技术的研究：[硕士学位论文]．西安：西安电子科技大学通信与信息系统专业，2005