基于纹理特征的图像检索技术研究
摘要
近年来,国内外在图像检索和管理技术的研究中出现了一个新热点—基于内容的图像检索CBIR(Content Based Image Retrieval)。纹理是图像的重要信息和特征,在病理诊断、文物复原等多个领域有着广泛的应用。本文主要针对基于纹理特征的图像检索技术展开研究。
本文主要工作包括纹理特征提取的算法研究和相似性度量方法的研究。在纹理特征提取算法中,研究了灰度共生矩阵法,并通过实验验证了其有效性。然后研究了小波变换法,通过树形小波分解得到图像的能量值作为纹理特征。最后研究了Tamura纹理分析法,实验验证了此方法符合人眼的视觉感受。在相似性度量方法中,研究了欧氏距离、街区距离、组合距离等,并探讨了相关反馈算法。最后给出了本文选用的距离函数。
最后,通过实验验证了灰度共生矩阵法采用的四个特征参数能够有效的反映图像在方向、间隔、变化幅度及快慢的综合信息,既保证了纹理图像特征的准确性,又提高了提取速度。相似度量中的加权街区距离由于考虑了各特征对于图像的重要程度,提高了图像检索的精度。
In recent years, a new point appeared in the research of the technology of image retrieval and management. Texture is an important feature in image, used in pathological diagnose, heritage recovery, and so on. This paper mainly researched the technology of image retrieval based on texture feature.
Recently, a few image retrieval systems have been realized. But most of them stay at experiment stage, and are used only in special field. It does not receive a wide range of promotion. In this paper, the texture image retrieval algorithms were researched in-depth. It provided valuable support of the theory and frame of reference for content-based image retrieval technology.
The main work of this paper included the researches of texture feature extraction algorithms and similarity measurement methods. In the texture feature extraction algorithms, the gray level co-occurrence matrix method was studied, and through experiments, its validity was verified. And then the wavelet transform was studied, through the wavelet decomposition tree to receive the energy value as the image texture feature. Finally, the Tamura texture was analyzed. The experiments verified that this method suited with the human eye's visual experience. In the similarity measurement methods, it studyed Euclidean distance, block distance and the combination distance, and so on, and discussed the relevance feedback algorithm. Finally, this paper gave the distance function.
Finally, the four parameters of gray co-occurrence matrix could effectively reflect the image in the direction, interval, changes in the rate and speed information. It not only ensured the accuracy of the feature of the image texture, but also increased the speed of extraction. In the similarity measurement methods, because the weighted block distance considered all features' importance to the image, it improved the accuracy of the image retrieval.
本文主要工作包括纹理特征提取的算法研究和相似性度量方法的研究。在纹理特征提取算法中,研究了灰度共生矩阵法,并通过实验验证了其有效性。然后研究了小波变换法,通过树形小波分解得到图像的能量值作为纹理特征。最后研究了Tamura纹理分析法,实验验证了此方法符合人眼的视觉感受。在相似性度量方法中,研究了欧氏距离、街区距离、组合距离等,并探讨了相关反馈算法。最后给出了本文选用的距离函数。
最后,通过实验验证了灰度共生矩阵法采用的四个特征参数能够有效的反映图像在方向、间隔、变化幅度及快慢的综合信息,既保证了纹理图像特征的准确性,又提高了提取速度。相似度量中的加权街区距离由于考虑了各特征对于图像的重要程度,提高了图像检索的精度。
In recent years, a new point appeared in the research of the technology of image retrieval and management. Texture is an important feature in image, used in pathological diagnose, heritage recovery, and so on. This paper mainly researched the technology of image retrieval based on texture feature.
Recently, a few image retrieval systems have been realized. But most of them stay at experiment stage, and are used only in special field. It does not receive a wide range of promotion. In this paper, the texture image retrieval algorithms were researched in-depth. It provided valuable support of the theory and frame of reference for content-based image retrieval technology.
The main work of this paper included the researches of texture feature extraction algorithms and similarity measurement methods. In the texture feature extraction algorithms, the gray level co-occurrence matrix method was studied, and through experiments, its validity was verified. And then the wavelet transform was studied, through the wavelet decomposition tree to receive the energy value as the image texture feature. Finally, the Tamura texture was analyzed. The experiments verified that this method suited with the human eye's visual experience. In the similarity measurement methods, it studyed Euclidean distance, block distance and the combination distance, and so on, and discussed the relevance feedback algorithm. Finally, this paper gave the distance function.
Finally, the four parameters of gray co-occurrence matrix could effectively reflect the image in the direction, interval, changes in the rate and speed information. It not only ensured the accuracy of the feature of the image texture, but also increased the speed of extraction. In the similarity measurement methods, because the weighted block distance considered all features' importance to the image, it improved the accuracy of the image retrieval.
引文
[1]施智平,胡宏,李清永等.基于纹理普描述了的图像检索.软件学报.2005,16(6):1039-1045
[2]董卫军,周明全,耿国华.基于纹理-形状特征的图像检索技术.计算机工程与应用.2004,24:9-11
[3]B.S.Manjunath,W.Y.Ma.Texture features for browsing and retrieval of image data.IEEE Transactions on Pattern Analysis and Machine Intelligence.1996,18(8):837-842
[4]Stricker M,Orengo M.Similarity of color images.Pro.of SPIE:Storage and Retrieval for Image and Video Database.2002,2185:381-392
[5]刘忠伟.利用局部累加直方图进行彩色图像检索.中国图像图形学报.2003,3(7):533-537
[6]王炜,武德峰等.基于模糊分类的图像颜色直方图研究.模糊系统与数学.2003,17(4):94-98
[7]王宇生,陈纯.一种新的基于色彩的图像检索算法.计算机研究与发展.2002,39(1):105-109
[8]梁艳梅,翟宏坤,毋国光.基于模糊相关的彩色图像检索.中国科学(E辑).2003,33(10):934-938
[9]H.Tamura,S.Mori,T.Yamawaki.Texture features corresponding to visual perception.IEEE Transactions on Systems Man and Cybernetics.1978,8(6):460-473
[10]Robert M.Haralick,K.Shanmugam.Texture features for image classification.IEEE Transactions on Systems Man and Cybernetics.1979,3(6):610-621
[11]M.K.Hu.Visual pattern recognition by moment invariants.IEEE Transactions on Information Theory.1962,8(2):179-187
[12]L.Yang,F.Algregtsen.Fast computation of invariant geometric moments.Proc.IEEE Int.Conf.on Image Processing.1994
[13]Deepak Kapur,Y.N.Lakshman,Tushar Saxena.Computing invariants using elimination methods.Proc.IEEE Int.Conf.on Image Processing.1995
[14]孙君顶.基于内容的图像检索研究:[博士学位论文].西安:西安电子科技大学,2005
[15]乔静,袁海东,王国营等.一种结合颜色和纹理特征的图像检索系统的实现.计算机应用与软件.2007,24(2):106-136
[16]鲍健鹏,陈昭炯.聚类技术在图像检索中的应用.福建电脑.2007
[17]李迎新,张明,陆鹏.基于索引和相关反馈的图像检索研究.图形图像.2007:94-97
[18]孙兴华.基于内容的图像检索研究:[博士学位论文].南京:南京理工大学,2001
[19]蒋仁波.基于内容的图像检索技术研究与系统实现:[硕士学位论文].武汉:武汉大学,2004
[20]John R.Smith,Shih-FuChang.Tools and techniques for color image retrieval.Proc.of SPIE:Storage and Retrieval for Image and Video Database.1995,vol 2670
[21]G.Pass,R.Zabih.Histogram refinement for content-based image retrieval.IEEE Workshop on Applications of Computer Vision.1996,96-102
[22]J.Mao,A.K.Jain.Texture classification and segmentation using multiresolution simultaneous autoregressive models.Pattern Recognition.1992,25(2):173-188
[23]章毓晋.基于内容的视觉信息检索.北京:科学出版社,2003:84-102
[24]Wu P,Manjunath B S,Newsam et al.Texture descriptor for browsing and similarity retrieval.Signal Processing.2000,16(1-2):33-43
[25]Moses Amadasum,Robert King.Texture features corresponding to texture properties.IEEE Transactions on System Man and Cybernetics.1989,19(5):1264-1274
[26]王辉.基于灰度共生矩阵木材表面纹理模式识别方法的研究:[硕士学位论文].哈尔滨:东北林业大学,2007
[27]T.Chang,C.C.Jay Kuo.Texture analysis and classification with tree-structured wavelet transform.IEEE Transactions on Image Processing.1993,2(4):429-441
[28]A.Laine,J.Fan.Texture classification by wavelet packet signatures.IEEE Transactions on Pattern Analysis and Machine Intelligence.1993,15(11):1186-1191
[29]冉启文.小波变换与分数傅里叶变换理论及应用.哈尔滨:哈尔滨工业大学出版社.2001,1-14
[30]K.C.Hayes,Jr.A.N.Shah,A.Rosenfeld.Texture coarseness:further experiments.IEEE Transactions on SMC-4.1974:467-472
[31]Swain M.J.,Ballard D.H.Color indexing.Inti.J.on Computer Vision.1991,7(1):11-32
[32]庄越挺,潘云鹤,吴飞.网上多媒体信息分析与检索.北京:清华大学出版社.2002
[33]Rui Y,Huang S,Ortega M et al.Relevance Feedback:a power tool in interactive content-based image retrieval.IEEE Transactions on Circuits Systems for Video Technology.1998,8(5):644-655
[34]Rui Y,Huang T S.A novel relevance feedback technique in image retrieval.In:Proceedings of the 7~(th) ACM International Conference on Multimedia.1999:67-70
[35]刘倩.基于内容的图像检索中的相关反馈技术.华东交通大学学报.2003,20(4):71-74
[36]Cox I J,Miller M L,Minka T P et al.The Bayesian image retrieval system,picHunter:theory,implementation and psychophysical experiments.Institute of Electrical and Electronics Engineers Transaction on Image Procession.1996,9:20-37
[37]Meilhac C,Nastar C.Relevance feedback and category search in image databases.IEEE International Conference on Multimedia Computing and Systems.1999,1:97-105
[38]董立岩,苑森淼,刘光远等.基于贝叶斯分类器的图像分类.吉林大学学报(理学版).2007,45(2):249-253
[39]章舜仲,王树梅,黄河燕等.基于属性相关性分析的贝叶斯分类模型.情报学报.2007,26(2):271-274
[2]董卫军,周明全,耿国华.基于纹理-形状特征的图像检索技术.计算机工程与应用.2004,24:9-11
[3]B.S.Manjunath,W.Y.Ma.Texture features for browsing and retrieval of image data.IEEE Transactions on Pattern Analysis and Machine Intelligence.1996,18(8):837-842
[4]Stricker M,Orengo M.Similarity of color images.Pro.of SPIE:Storage and Retrieval for Image and Video Database.2002,2185:381-392
[5]刘忠伟.利用局部累加直方图进行彩色图像检索.中国图像图形学报.2003,3(7):533-537
[6]王炜,武德峰等.基于模糊分类的图像颜色直方图研究.模糊系统与数学.2003,17(4):94-98
[7]王宇生,陈纯.一种新的基于色彩的图像检索算法.计算机研究与发展.2002,39(1):105-109
[8]梁艳梅,翟宏坤,毋国光.基于模糊相关的彩色图像检索.中国科学(E辑).2003,33(10):934-938
[9]H.Tamura,S.Mori,T.Yamawaki.Texture features corresponding to visual perception.IEEE Transactions on Systems Man and Cybernetics.1978,8(6):460-473
[10]Robert M.Haralick,K.Shanmugam.Texture features for image classification.IEEE Transactions on Systems Man and Cybernetics.1979,3(6):610-621
[11]M.K.Hu.Visual pattern recognition by moment invariants.IEEE Transactions on Information Theory.1962,8(2):179-187
[12]L.Yang,F.Algregtsen.Fast computation of invariant geometric moments.Proc.IEEE Int.Conf.on Image Processing.1994
[13]Deepak Kapur,Y.N.Lakshman,Tushar Saxena.Computing invariants using elimination methods.Proc.IEEE Int.Conf.on Image Processing.1995
[14]孙君顶.基于内容的图像检索研究:[博士学位论文].西安:西安电子科技大学,2005
[15]乔静,袁海东,王国营等.一种结合颜色和纹理特征的图像检索系统的实现.计算机应用与软件.2007,24(2):106-136
[16]鲍健鹏,陈昭炯.聚类技术在图像检索中的应用.福建电脑.2007
[17]李迎新,张明,陆鹏.基于索引和相关反馈的图像检索研究.图形图像.2007:94-97
[18]孙兴华.基于内容的图像检索研究:[博士学位论文].南京:南京理工大学,2001
[19]蒋仁波.基于内容的图像检索技术研究与系统实现:[硕士学位论文].武汉:武汉大学,2004
[20]John R.Smith,Shih-FuChang.Tools and techniques for color image retrieval.Proc.of SPIE:Storage and Retrieval for Image and Video Database.1995,vol 2670
[21]G.Pass,R.Zabih.Histogram refinement for content-based image retrieval.IEEE Workshop on Applications of Computer Vision.1996,96-102
[22]J.Mao,A.K.Jain.Texture classification and segmentation using multiresolution simultaneous autoregressive models.Pattern Recognition.1992,25(2):173-188
[23]章毓晋.基于内容的视觉信息检索.北京:科学出版社,2003:84-102
[24]Wu P,Manjunath B S,Newsam et al.Texture descriptor for browsing and similarity retrieval.Signal Processing.2000,16(1-2):33-43
[25]Moses Amadasum,Robert King.Texture features corresponding to texture properties.IEEE Transactions on System Man and Cybernetics.1989,19(5):1264-1274
[26]王辉.基于灰度共生矩阵木材表面纹理模式识别方法的研究:[硕士学位论文].哈尔滨:东北林业大学,2007
[27]T.Chang,C.C.Jay Kuo.Texture analysis and classification with tree-structured wavelet transform.IEEE Transactions on Image Processing.1993,2(4):429-441
[28]A.Laine,J.Fan.Texture classification by wavelet packet signatures.IEEE Transactions on Pattern Analysis and Machine Intelligence.1993,15(11):1186-1191
[29]冉启文.小波变换与分数傅里叶变换理论及应用.哈尔滨:哈尔滨工业大学出版社.2001,1-14
[30]K.C.Hayes,Jr.A.N.Shah,A.Rosenfeld.Texture coarseness:further experiments.IEEE Transactions on SMC-4.1974:467-472
[31]Swain M.J.,Ballard D.H.Color indexing.Inti.J.on Computer Vision.1991,7(1):11-32
[32]庄越挺,潘云鹤,吴飞.网上多媒体信息分析与检索.北京:清华大学出版社.2002
[33]Rui Y,Huang S,Ortega M et al.Relevance Feedback:a power tool in interactive content-based image retrieval.IEEE Transactions on Circuits Systems for Video Technology.1998,8(5):644-655
[34]Rui Y,Huang T S.A novel relevance feedback technique in image retrieval.In:Proceedings of the 7~(th) ACM International Conference on Multimedia.1999:67-70
[35]刘倩.基于内容的图像检索中的相关反馈技术.华东交通大学学报.2003,20(4):71-74
[36]Cox I J,Miller M L,Minka T P et al.The Bayesian image retrieval system,picHunter:theory,implementation and psychophysical experiments.Institute of Electrical and Electronics Engineers Transaction on Image Procession.1996,9:20-37
[37]Meilhac C,Nastar C.Relevance feedback and category search in image databases.IEEE International Conference on Multimedia Computing and Systems.1999,1:97-105
[38]董立岩,苑森淼,刘光远等.基于贝叶斯分类器的图像分类.吉林大学学报(理学版).2007,45(2):249-253
[39]章舜仲,王树梅,黄河燕等.基于属性相关性分析的贝叶斯分类模型.情报学报.2007,26(2):271-274