基于内容图象检索中关键技术的研究
|
摘要
本文主要用整数小波变换研究基于内容的图象检索技术，该算法的优点在于：整数小波变换的输入输出都是整数，适合数字图象处理；使用小波变换可以用多分辨率分析提取图象特征，可以降低特征量的维数。
本文首先介绍了小波变换的基本知识，上升型方案和整数小波变换。其次用整数小波变换对彩色图象进行多分辨率分析；由于小波变换的低频部分保持了图象的概貌，因此用小波变换低频部分的局部区域能量和F-范数作为彩色图象的特征向量，对图象进行检索，可以降低特征量的维数，提高检索速度。此外，由于纹理图象的主要特征表现在细节部分，而高频部分的小波系数体现了图象的细节，所以从这些小波系数中提取的特征，能够表征纹理图象的主要特性；实验结果表明，用该方法检索纹理图象，能够达到较好的检索效果，并且对亮度不敏感，这一特点是传统的纹理分析方法难以达到的。
图象匹配算法中使用了比值相似度定义，这种相似度计算简单，易于实现，而且能够获得较好的检索结果。
还实现了基于颜色矩的彩色图象检索算法，基于区域不变矩的二值目标检索算法，提出了一种CBIR系统模型，介绍了常用的CBIR系统评价标准。最后概述了CBIR有待于继续研究的相关领域。
The content-based image retrieval algorithm using integer-to-integer WT is studied in this paper. The integer-to-integer WT has the following advantages: it fits for image processing because its input and output values are all integers; it extracts image features by using multi- resolution analysis; and it decreases the size of dimensions of the image features.
In this paper, the basic knowledge on WT, lifting-scheme and integer-to-integer WT is briefly introduced first, and then the color image is analyzed by integer-to-integer WT multi-resolution in order to reduce the dimensions of the image. Since the low pass part of WT preserves the sketch of the image, better results of the index of color images can be obtained by taking the local region energy and F-norm of WT as the color image features Furthermore, WT coefficients are used to analyze the texture image. As the main features of the texture image are present in the details and the high pass of the WT coefficients denotes the details of the image, the character deduced from WT coefficients can be used to retrieve the texture image.
The degree of similarity with ratio in image matching algorithms is defined, which is single to calculate, easy to realize and effective to get better retrieval results.
This paper also realizes color image retrieval and binary-target indexing algorithms based on moments; proposes a CBER framework model; introduces general criterions for CBIR system; and finally summarizes the future research direction of CBIR.
本文首先介绍了小波变换的基本知识，上升型方案和整数小波变换。其次用整数小波变换对彩色图象进行多分辨率分析；由于小波变换的低频部分保持了图象的概貌，因此用小波变换低频部分的局部区域能量和F-范数作为彩色图象的特征向量，对图象进行检索，可以降低特征量的维数，提高检索速度。此外，由于纹理图象的主要特征表现在细节部分，而高频部分的小波系数体现了图象的细节，所以从这些小波系数中提取的特征，能够表征纹理图象的主要特性；实验结果表明，用该方法检索纹理图象，能够达到较好的检索效果，并且对亮度不敏感，这一特点是传统的纹理分析方法难以达到的。
图象匹配算法中使用了比值相似度定义，这种相似度计算简单，易于实现，而且能够获得较好的检索结果。
还实现了基于颜色矩的彩色图象检索算法，基于区域不变矩的二值目标检索算法，提出了一种CBIR系统模型，介绍了常用的CBIR系统评价标准。最后概述了CBIR有待于继续研究的相关领域。
The content-based image retrieval algorithm using integer-to-integer WT is studied in this paper. The integer-to-integer WT has the following advantages: it fits for image processing because its input and output values are all integers; it extracts image features by using multi- resolution analysis; and it decreases the size of dimensions of the image features.
In this paper, the basic knowledge on WT, lifting-scheme and integer-to-integer WT is briefly introduced first, and then the color image is analyzed by integer-to-integer WT multi-resolution in order to reduce the dimensions of the image. Since the low pass part of WT preserves the sketch of the image, better results of the index of color images can be obtained by taking the local region energy and F-norm of WT as the color image features Furthermore, WT coefficients are used to analyze the texture image. As the main features of the texture image are present in the details and the high pass of the WT coefficients denotes the details of the image, the character deduced from WT coefficients can be used to retrieve the texture image.
The degree of similarity with ratio in image matching algorithms is defined, which is single to calculate, easy to realize and effective to get better retrieval results.
This paper also realizes color image retrieval and binary-target indexing algorithms based on moments; proposes a CBER framework model; introduces general criterions for CBIR system; and finally summarizes the future research direction of CBIR.
引文
[1]Yong Rui and Thomas S.Huang, Shih-FuChang .Image Retrieval: Past, Present, and Future. In Jour of Visual Communication and Image Representation, Volume 10,p1-23,1999.
[2]John P Eakins and Margaret E Graham .Content-based Image Retrieval :A report to the JISC Technology Applications Programme. Institute for Image Data Research, University of orthumbria at Newcastle .January 1999. In http://www.unn.ac.uk/iidr/report.html.
[3]卢汉清，孔维新等．基于内容的视频信号与图像库检索中的图像技术．自动化学报，2001，Vo．27，No．1，p56-69．
[4]郑南宁，计算机视觉与模式识别．北京：国防工业出版社，1998，第1版．
[5]Michael Swain and Dana Ballard.Color indexing. International Journal of Computer Vision,7(1),1991.
[6]Markus Stricker and Markus Orengo. Similarity of color image. In Proc. SPIE Storage and Retrieval for Image and Video Database, 1995.
[7]John R. Smith and Shih-Fu Chang. Single color extraction and image query. In Proc. IEEE Int. Conf. on Image Proc.,1995.
[8]John R. Smith and Shih-Fu Chang. Tools and Techniques for color image retrieval. In IS T/SPIE proceedings Vol.2670,Storage retrieval for Image and Video Databases Ⅳ, 1995.
[9]Remco C. Veltkamp, Mirela Tanase. Content-based Image Retrieval System:A Survey. In http://citeseer.nj.nec.com/327932.html.
[10]Jefferey R. Bach, etc., The Virage image search engine: An open framework for image management. In Proc. SPIE Storage and Retrieval for Image and Video Database. Ⅳ, pp.76-87, Feb. 1996.
[11]Song-Chun Zhu, etc. Exploring Texture Ensembles by Efficient Markov Chain Monte Carlo, in IEEE Trans. On PAMI, 1999.
[12]崔屹，数字图象处理技术与应用。电子工业出版社，1997年第1版．P93-201。
[13]章毓晋，图象工程(上册)。清华大学出版社，1999年第1版，P179-252。
[14]杨福生．小波变换的工程分析与应用[M]．北京：科学出版社，1999．112-119．
[15]姚玉荣，章毓晋．利用小波和矩进行基于形状的图象检索[J]．中国图象图形学报，2000，5(3)：206～210．
[16]解梅，马争，B-样条小波边缘检测算子的应用研究。电子学报，1999．27(3)：106—108。
[17] Esther M.Allan, etc., An efficient computable metric for comparing polygonal shapes. IEEE Trans. Patt. Recog. And Mach. In tell., 13(3) , March 1991．
[18] Hideyuki Tamura etc. Texture features corresponding to visual perception.IEEE Trans, on Sys, Man, and Cyb, SMCS-8(6) ,1978．
[19] Will Equitz and Wayne Niblack. Retrieving images from a database using texture-algorithms from the QBIC system. Computer science, IBM Research Report, May 1994．
[20] Thomas S. Huang, etc. Multimedia analysis and retrieval system(MARS) project. In Proc of 33rd Annual Clinic on Library Application of Data processing-Digital Image Access and Retrieval, 1996．
[21] Michael Ortega, etc., Supporting similarity queries in MARS. In Proc. of ACM Conf. On Multimedia, 1997．
[22] John R. Smith, etc., Transform features for texture classification and discrimination in large image database. In Proc. IEEE Int Conf. on Image Proc., 1994．
[23] Tianhorng Chang etc. Texture analysis and classification with tree-structured wavelet transform. IEEE Trans. Image Proc., 2(4) :429-441,October 1993．
[24] Shih-Fu Chang and John Smith. Extracting multi-dimensional signal features for content-based visual query. In Proc SPIE Symposium on Visual Communications and Signal Processing, 1995．
[25] Gene C.-H. Chuang etc., Wavelet descriptor of Planar curves: Theory and applications. IEEE Trans. Image Proc., 5(1) : 56-70, Jan. 1996．
[26] G. C. Cross and A. K.. Jain. Markov random field texture models. IEEE Trans. Patt. Recog. And Mach. Intell., 5:25-39, 1983．
[27] Andrew Laine and Jian Fan. Texture Classification by wavelet packet signatures. IEEE Trans. Patt. Recog. and Mach. Intell., 15(11) : 1186-1191, 1993．
[28] M. H. Gross, etc., Multiscale image texture analysis in wavelet spaces. In Proc. IEEE int. Conf. on Image Proc., 1994．
[29] Amlan Kundu, etc. Texture classification using qmf bank-based subband decomposition. CVGIP: Graphical Models and Image Processing, 54(5) : 369-384, Sept. 1992．
[30] K. S. Thyagarajan, etc., A maximum likehood approach to texture classification using wavelet transform. In Proc. IEEE Int. Conf. on Image Proc
Proc., 1994．
[31] John R. Smith, etc., Transformation features for texture classification and discrimination in large image database. In Proc. IEEE Int. Conf. on Image Proc., 1994．
[32] John R. Smith, etc., Automated binary texture feature sets for image retrieval. In Proc ICASSP-96, Atlanta, GA, 1996．
[33] Philippe P. Ohanian, etc., Performance evaluation for four classes of texture features. Pattern Recognition, 25(8) :819-833, 1992．
[34] Yong Rui, etc., Modified Fourier descriptors for shape represenatation-a practical approach. In Proc of First International Workshop on Image Databases and Multi Media Search, 1996．
[35] Yong Rui, etc., Content-based image retrieval with relevance feedback in MARS. In Proc. IEEE Int. Conf. on Image Proc., 1997．
[36] M. K. Hu. Visual pattern recognition by moment invariants, computer methods in image analysis. ERE Trans, on Information Theory,8,1962．
[37] A. Pentland, etc., Photobook: Content-based manipulation of image database.International Journal of Computer Vision, 1996．
[38] Babu M. Mehtre, etc., Shape measures for content based image retrieval: A comparison. Information Processing &Management, 33(3) , 1997．
[39] J.R.Smith, etc., Local color and texture extraction and spatial query. In Proc . IEEE Int. Conf. on Image Proc., 1996．
[40] John R. Smith, etc., Single color extraction and image query. In Proc. IEEE Int. Conf. on Image Proc., 1995．
[41] Bernice E. Rogowitz, etc., Perceptual image similarity experiments. In Conf. on Human Vision and Electronic Imaging III, 1998．
[42] Michael D. Adams. Reversible Integer-to-integer Wavelet Transforms for Image Compression: Performance Evaluation and Analysis . in IEEE Trans on Image processing,. VOL.. 9． NO.6 2000,1010-1024．
[43] .Huilin Xiong, etc. A Translation-and Scale-Invariant Adaptive Wavelet Transform. In IEEE Trans, on Image Processing, VOL.9, NO. 12, DEC. 2000．
[44] .Dinggang Shen, Horace H.S. Ip. Discriminative wavelet shape descriptor for recognition of 2-D patterns. In Pattern Recognition 32(1999) 151-166．
[45] 乔世杰，小波图象编码中的对称边沿延拓发，中国图象图形学报，2000 5(9) , 725-729．
[46] 马社祥等，基于多尺度均值和小波变换的Internet图象可分级压缩编码传
输技术．中国图象图形学报，2000，5(11)，942～947．[47]James Ze Wang etc. Wavelet Based Image Indexing Techniques with Partial Sketch Retrieval Capability. Stanford University, Stanford, CA 94305.
[48]Calderbank AR, Daubechies I, W. Swekdens. Wavelet Transfors that map Integers to Integers.
[49]W. Swekdens. The lifting scheme: A custom-design constuction of Biorthogonal Wavelets. Appl. Comput. Harmon. Anal, 3(2): 186-200,1996.
[50]W. Swekdens. The lifting scheme: A constuction of second generation wavelets. SIAM J. Math. Anal., to appear, 1997.
[51]Ingrid Daubechies, W. Sweldens. Factoring Wavelet Transforms Into Lifting Steps. Technical report, Bell Lab., Lucent Technologies, 1996.
[52]章毓晋．图象分割．北京：科学出版社，2001．
[53]Jan Flussen. On the independence of rotation moment invariants. In Pattern Recognition 33(2000)1405-1410.
[54]张妙兰，付新文。一种纹理图象分类方法研究。中国图象图形学报，1999，4(8)，1580～683．
[55]何振亚，鲍锴等。纹理图象分割的分形方法研究。数据采集与处理。1996．11(3)，163～167．
[56]程义民，王以孝等。一种基于主矢量集的纹理图象分析方法。中国图象图形学报，1999，4(2)，129～134．
[57]张建宝，陈晓锋等。用于遥感图象分类的神经网络的构造。中国图象图形学报，1999，4(10)，831～834．
[58]熊桢，童庆禧等。用于高光谱遥感图象分类的一种高阶神经网络算法
[59]Haria BartoLini, Paolo Ciaccia etc. Feedback By Pass: A new Approach to Interactive Similarity Query Processing. In proceedings of the 27th VLDB Conference Roma, Italy, 2001.
[60]Henning Müller, Wolfgang Müller etc. Strategies for positive and negative relevance feedback in image retrieval. In http://viper.unige.ch/.
[2]John P Eakins and Margaret E Graham .Content-based Image Retrieval :A report to the JISC Technology Applications Programme. Institute for Image Data Research, University of orthumbria at Newcastle .January 1999. In http://www.unn.ac.uk/iidr/report.html.
[3]卢汉清，孔维新等．基于内容的视频信号与图像库检索中的图像技术．自动化学报，2001，Vo．27，No．1，p56-69．
[4]郑南宁，计算机视觉与模式识别．北京：国防工业出版社，1998，第1版．
[5]Michael Swain and Dana Ballard.Color indexing. International Journal of Computer Vision,7(1),1991.
[6]Markus Stricker and Markus Orengo. Similarity of color image. In Proc. SPIE Storage and Retrieval for Image and Video Database, 1995.
[7]John R. Smith and Shih-Fu Chang. Single color extraction and image query. In Proc. IEEE Int. Conf. on Image Proc.,1995.
[8]John R. Smith and Shih-Fu Chang. Tools and Techniques for color image retrieval. In IS T/SPIE proceedings Vol.2670,Storage retrieval for Image and Video Databases Ⅳ, 1995.
[9]Remco C. Veltkamp, Mirela Tanase. Content-based Image Retrieval System:A Survey. In http://citeseer.nj.nec.com/327932.html.
[10]Jefferey R. Bach, etc., The Virage image search engine: An open framework for image management. In Proc. SPIE Storage and Retrieval for Image and Video Database. Ⅳ, pp.76-87, Feb. 1996.
[11]Song-Chun Zhu, etc. Exploring Texture Ensembles by Efficient Markov Chain Monte Carlo, in IEEE Trans. On PAMI, 1999.
[12]崔屹，数字图象处理技术与应用。电子工业出版社，1997年第1版．P93-201。
[13]章毓晋，图象工程(上册)。清华大学出版社，1999年第1版，P179-252。
[14]杨福生．小波变换的工程分析与应用[M]．北京：科学出版社，1999．112-119．
[15]姚玉荣，章毓晋．利用小波和矩进行基于形状的图象检索[J]．中国图象图形学报，2000，5(3)：206～210．
[16]解梅，马争，B-样条小波边缘检测算子的应用研究。电子学报，1999．27(3)：106—108。
[17] Esther M.Allan, etc., An efficient computable metric for comparing polygonal shapes. IEEE Trans. Patt. Recog. And Mach. In tell., 13(3) , March 1991．
[18] Hideyuki Tamura etc. Texture features corresponding to visual perception.IEEE Trans, on Sys, Man, and Cyb, SMCS-8(6) ,1978．
[19] Will Equitz and Wayne Niblack. Retrieving images from a database using texture-algorithms from the QBIC system. Computer science, IBM Research Report, May 1994．
[20] Thomas S. Huang, etc. Multimedia analysis and retrieval system(MARS) project. In Proc of 33rd Annual Clinic on Library Application of Data processing-Digital Image Access and Retrieval, 1996．
[21] Michael Ortega, etc., Supporting similarity queries in MARS. In Proc. of ACM Conf. On Multimedia, 1997．
[22] John R. Smith, etc., Transform features for texture classification and discrimination in large image database. In Proc. IEEE Int Conf. on Image Proc., 1994．
[23] Tianhorng Chang etc. Texture analysis and classification with tree-structured wavelet transform. IEEE Trans. Image Proc., 2(4) :429-441,October 1993．
[24] Shih-Fu Chang and John Smith. Extracting multi-dimensional signal features for content-based visual query. In Proc SPIE Symposium on Visual Communications and Signal Processing, 1995．
[25] Gene C.-H. Chuang etc., Wavelet descriptor of Planar curves: Theory and applications. IEEE Trans. Image Proc., 5(1) : 56-70, Jan. 1996．
[26] G. C. Cross and A. K.. Jain. Markov random field texture models. IEEE Trans. Patt. Recog. And Mach. Intell., 5:25-39, 1983．
[27] Andrew Laine and Jian Fan. Texture Classification by wavelet packet signatures. IEEE Trans. Patt. Recog. and Mach. Intell., 15(11) : 1186-1191, 1993．
[28] M. H. Gross, etc., Multiscale image texture analysis in wavelet spaces. In Proc. IEEE int. Conf. on Image Proc., 1994．
[29] Amlan Kundu, etc. Texture classification using qmf bank-based subband decomposition. CVGIP: Graphical Models and Image Processing, 54(5) : 369-384, Sept. 1992．
[30] K. S. Thyagarajan, etc., A maximum likehood approach to texture classification using wavelet transform. In Proc. IEEE Int. Conf. on Image Proc
Proc., 1994．
[31] John R. Smith, etc., Transformation features for texture classification and discrimination in large image database. In Proc. IEEE Int. Conf. on Image Proc., 1994．
[32] John R. Smith, etc., Automated binary texture feature sets for image retrieval. In Proc ICASSP-96, Atlanta, GA, 1996．
[33] Philippe P. Ohanian, etc., Performance evaluation for four classes of texture features. Pattern Recognition, 25(8) :819-833, 1992．
[34] Yong Rui, etc., Modified Fourier descriptors for shape represenatation-a practical approach. In Proc of First International Workshop on Image Databases and Multi Media Search, 1996．
[35] Yong Rui, etc., Content-based image retrieval with relevance feedback in MARS. In Proc. IEEE Int. Conf. on Image Proc., 1997．
[36] M. K. Hu. Visual pattern recognition by moment invariants, computer methods in image analysis. ERE Trans, on Information Theory,8,1962．
[37] A. Pentland, etc., Photobook: Content-based manipulation of image database.International Journal of Computer Vision, 1996．
[38] Babu M. Mehtre, etc., Shape measures for content based image retrieval: A comparison. Information Processing &Management, 33(3) , 1997．
[39] J.R.Smith, etc., Local color and texture extraction and spatial query. In Proc . IEEE Int. Conf. on Image Proc., 1996．
[40] John R. Smith, etc., Single color extraction and image query. In Proc. IEEE Int. Conf. on Image Proc., 1995．
[41] Bernice E. Rogowitz, etc., Perceptual image similarity experiments. In Conf. on Human Vision and Electronic Imaging III, 1998．
[42] Michael D. Adams. Reversible Integer-to-integer Wavelet Transforms for Image Compression: Performance Evaluation and Analysis . in IEEE Trans on Image processing,. VOL.. 9． NO.6 2000,1010-1024．
[43] .Huilin Xiong, etc. A Translation-and Scale-Invariant Adaptive Wavelet Transform. In IEEE Trans, on Image Processing, VOL.9, NO. 12, DEC. 2000．
[44] .Dinggang Shen, Horace H.S. Ip. Discriminative wavelet shape descriptor for recognition of 2-D patterns. In Pattern Recognition 32(1999) 151-166．
[45] 乔世杰，小波图象编码中的对称边沿延拓发，中国图象图形学报，2000 5(9) , 725-729．
[46] 马社祥等，基于多尺度均值和小波变换的Internet图象可分级压缩编码传
输技术．中国图象图形学报，2000，5(11)，942～947．[47]James Ze Wang etc. Wavelet Based Image Indexing Techniques with Partial Sketch Retrieval Capability. Stanford University, Stanford, CA 94305.
[48]Calderbank AR, Daubechies I, W. Swekdens. Wavelet Transfors that map Integers to Integers.
[49]W. Swekdens. The lifting scheme: A custom-design constuction of Biorthogonal Wavelets. Appl. Comput. Harmon. Anal, 3(2): 186-200,1996.
[50]W. Swekdens. The lifting scheme: A constuction of second generation wavelets. SIAM J. Math. Anal., to appear, 1997.
[51]Ingrid Daubechies, W. Sweldens. Factoring Wavelet Transforms Into Lifting Steps. Technical report, Bell Lab., Lucent Technologies, 1996.
[52]章毓晋．图象分割．北京：科学出版社，2001．
[53]Jan Flussen. On the independence of rotation moment invariants. In Pattern Recognition 33(2000)1405-1410.
[54]张妙兰，付新文。一种纹理图象分类方法研究。中国图象图形学报，1999，4(8)，1580～683．
[55]何振亚，鲍锴等。纹理图象分割的分形方法研究。数据采集与处理。1996．11(3)，163～167．
[56]程义民，王以孝等。一种基于主矢量集的纹理图象分析方法。中国图象图形学报，1999，4(2)，129～134．
[57]张建宝，陈晓锋等。用于遥感图象分类的神经网络的构造。中国图象图形学报，1999，4(10)，831～834．
[58]熊桢，童庆禧等。用于高光谱遥感图象分类的一种高阶神经网络算法
[59]Haria BartoLini, Paolo Ciaccia etc. Feedback By Pass: A new Approach to Interactive Similarity Query Processing. In proceedings of the 27th VLDB Conference Roma, Italy, 2001.
[60]Henning Müller, Wolfgang Müller etc. Strategies for positive and negative relevance feedback in image retrieval. In http://viper.unige.ch/.