基于语义网的图像检索算法的研究
摘要
目前我们所使用的互联网,计算机对文字、图像、视频等信息只起到存储、传输的作用,对其语义无法进行识别、理解和处理。语义网是当前现有计算机网络的扩展,它提供对数据本身的语义描述,实现不同计算机之间的智能化交互,从而使Web成为全球化信息共享的智能服务平台。图像搜索在Internet网、数码照片进行自动分类、医学图像分类及卫星遥感图像处理等有着非常广泛应用,对下一代Internet网,即语义网图像搜索的研究具有非常重要的意义和潜在的应用价值。
主要研究工作包括:
对颜色量化及颜色特征提取作了深入研究。为了更准确的提取图像的颜色特征及方便后面的语义分类,提出一种基于图像块颜色量化码书的码词分布直方图来描述图像的颜色特征。该描述子提取简单,计算速度快,符合人类感知。提出了一种梯度方向直方图金字塔PHOG描述子,它能很好的描述局部形状特征及形状的空间布局分布情况,具有较好的鲁棒性和稳定性。结合颜色码书和PHOG的图像检索方法在实验中取得了很好的效果。
用SVM支持向量机建立了一个决策树层次架构,用于对前面建立的本体框架进行分类,实现基于语义的图像检索。对同一语义类的图像在按与查询例图的相似度进行RANK时,提出一种新的称为互相似序号的测度用于更好的测度图像间的相互相似性,同时对传统RANK方法加以改进,使之更符合人类对图像相似性的感知。在实验中取得了较好的效果。
Currently we are using the Internet, the computer on the text, images, video and other information has only a storage, transmission function, its meaning can not be identified, understood and addressed.
Semantic Web is the current expansion of the existing computer network, which provides semantic description of the data itself, to achieve an intelligent interaction between different computers, so that a global Web services platform for information sharing intelligence. Image search in Internet networks, automatic classification of digital photos, and satellite remote sensing image classification medical image processing has a very wide application.
The main work includes:
On the color quantization and color feature extraction depth study. For more accurate extraction of color features and convenience of the back of the semantic classification, the introduction of a block-based color image quantization codebook distribution histogram of code words to describe the color characteristics of the image. A pyramid PHOG gradient direction histogram descriptor, it is a good description of local shape features and shape of the distribution of the spatial layout.Combined with the color code books and PHOG image retrieval method in the experiment achieved good results.
SVM Support Vector Machines using a decision tree-level framework for the establishment of the ontological framework of the previous classification, to achieve the semantic-based image retrieval.
Images of the same semantic class in accordance with the query's similarity RANK case diagram, the paper presents a new measure known as the serial number for each other like a better measure of similarity between images of each other, while the traditional method RANK be improved to make it more consistent with human perception of image similarity. In the experiment, achieved good results
主要研究工作包括:
对颜色量化及颜色特征提取作了深入研究。为了更准确的提取图像的颜色特征及方便后面的语义分类,提出一种基于图像块颜色量化码书的码词分布直方图来描述图像的颜色特征。该描述子提取简单,计算速度快,符合人类感知。提出了一种梯度方向直方图金字塔PHOG描述子,它能很好的描述局部形状特征及形状的空间布局分布情况,具有较好的鲁棒性和稳定性。结合颜色码书和PHOG的图像检索方法在实验中取得了很好的效果。
用SVM支持向量机建立了一个决策树层次架构,用于对前面建立的本体框架进行分类,实现基于语义的图像检索。对同一语义类的图像在按与查询例图的相似度进行RANK时,提出一种新的称为互相似序号的测度用于更好的测度图像间的相互相似性,同时对传统RANK方法加以改进,使之更符合人类对图像相似性的感知。在实验中取得了较好的效果。
Currently we are using the Internet, the computer on the text, images, video and other information has only a storage, transmission function, its meaning can not be identified, understood and addressed.
Semantic Web is the current expansion of the existing computer network, which provides semantic description of the data itself, to achieve an intelligent interaction between different computers, so that a global Web services platform for information sharing intelligence. Image search in Internet networks, automatic classification of digital photos, and satellite remote sensing image classification medical image processing has a very wide application.
The main work includes:
On the color quantization and color feature extraction depth study. For more accurate extraction of color features and convenience of the back of the semantic classification, the introduction of a block-based color image quantization codebook distribution histogram of code words to describe the color characteristics of the image. A pyramid PHOG gradient direction histogram descriptor, it is a good description of local shape features and shape of the distribution of the spatial layout.Combined with the color code books and PHOG image retrieval method in the experiment achieved good results.
SVM Support Vector Machines using a decision tree-level framework for the establishment of the ontological framework of the previous classification, to achieve the semantic-based image retrieval.
Images of the same semantic class in accordance with the query's similarity RANK case diagram, the paper presents a new measure known as the serial number for each other like a better measure of similarity between images of each other, while the traditional method RANK be improved to make it more consistent with human perception of image similarity. In the experiment, achieved good results
引文
[1]丁艳春.基于图像语义和内容的半自动标注系统.浙江大学硕士论文.2008, pp. 11～27
[2]张磊.图像分类和图像语义标注的研究.山东大学硕士学位论文.2008, pp. 22～33
[3]NiblackW.,etal.“The QBIC Projeet:Querying Images by Content Using Color, Texture and Shape”,in Proeeedihgs SPIE,San Jose,CA,1993,28(4):56-58.
[4]A.Pentland,R.W.Pieard,Selaroff,“Content-based Manipulation of image databases”, In International Journal of Computer Vision,1996,12(2):345-248.
[5]J.R.Bach,C.fuller,A.Gupta,A.Hampapur,“The Virage images search engine:An open framework for image management”,in Proc.SPIE Storage and Retrieval for Image and Video Databases,1989,37(11):122-124.
[6]RuiY,HuangTS,Meht.tras,etal.“Content-based image retrieval With relevance feedback”in MARS.Proe.ICIP,1997,2:815-818.
[7]赵继东,鲁珂,吴跃.一种基于谱图理论的Web图像搜索方法[J].计算机应用研究.2008,22(2):23-24.
[8]周明全,耿国华,韦娜著.基于内容图像检索技术[M].北京:清华大学出版社, 2007:35～36.
[9]忠伟,章毓晋.综合利用颜色和纹理的特征检索[J].通信学报,1999,21(5):36-37
[10]沈云涛.基于视觉感知特性的图像检索研究[D].西北工业大学博士学位论文,2005,pp. 35～36
[11]L.Tao and G.xu.Color in Machine Vision and its APPlication.Chinese Science Bulletin,2001,46(17),1411-1421
[12]H.Palus.RePresentations of Colour Images in Different Colour SPaces.The colour imagehand book(eds.5.j.sangwine and r.e.n.horne),london:chapman&hall,1998,79-92.
[13]F. R. K. Chung. Spectral Graph Theory, volume 92 of Regional Conference Series in Mathematics. AMS, 1997.45(8):212-213
[14]张芳芳.基于内容的图像检索系统优化方法研究.武汉大学硕士学位论文.2004, pp. 21～22
[15]程珺.基于相关颜色的图像检索.华中科技大学硕士学位论文.2006, pp. 12～14
[16]史忠植,王文杰著.人工智能[M].北京:国防工业出版,2007, pp. 31～32
[17]甘赟,方志军.基于HSV空间的彩色图像直方图检索技术[J],科技广场,2010,3(3):86-87
[18]周明全,耿国华,韦娜著.基于内容图像检索技术[M].北京:清华大学出版社, 2007:17～18.
[19]Manik Varma and Andrew Zisserman. A Statistical Approach to Texture Classication from Single Images.IJCV(62), No. 1-2, 2005: 61-81.
[20]Stricker M A,Orengo M.Similarity of color Images.SPIE Proceedings,1995,2420:381-392
[21]王涛,胡事民,孙家广.基于颜色-空间特征的图像检索[J].软件学报, 2002,13(10):2032-2033
[22]蔡丹琳.基于SVM反馈和多特征融合的图像检索技术研究.华侨大学硕士学位论文.2007, pp. 33～34
[23]R.M.Haralick,K.Shanmngam,and I.Dinstein,“Texture Feature for Image Classification,”IEEETransactionsonSystems,ManandCybernetics,vol.3,no.6,pp.610-621,1973.
[24]魏翩翩.基于内容的图像检索技术研究及系统实现.厦门大学硕士学位论文.2009,pp.19-20
[25]邓诚强,冯刚.基于内容的多特征综合图像检索[J]计算机应用,2003:19(10):55-56.
[26]谢邦旺.基于形状的图像检索.苏州大学硕士学位论文.2007, pp.17-18
[27]周明全,耿国华,韦娜著.基于内容图像检索技术[M].北京:清华大学出版社, 2007:72～73.
[28]李俊山,李旭辉.数字图像处理[M].北京:清华大学出版社.2007:279-280.
[29]袁杰.基于内容的综合多特征图像检索方法研究.浙江师范大学硕士学位论文. 2009, pp. 23～24
[30]Bosch,A,Zisserman,A,Munoz,X Representing shape with a spatial pyramid kernel New York,NY,USA:ACM (2007)401-408
[31]周明全,耿国华,韦娜著.基于内容图像检索技术[M].北京:清华大学出版社, 2007:147～148
[32]宋炜,张铭.语义网简明教程[M].北京:高等教育出版社. 2004: 10-11.
[33]孙亮,任小康.基于本体的图像语义检索模型[J] .重庆工学院学报(自然科学版),2009,23(1):129-130.
[34] T Berners– Lee, J Hendler, O Lassila,“The semantic Web”,ScientificAmerican , 2001,284(5) :34～43
[35]张辉,张浩,陆剑峰.SVM在数据挖掘中的应用[J].计算机工程,2004,3(6):7-8.
[36]张华煜,邢丽萍.基于核函数的支持向量机分类方法[J].电脑开发与应用,2005,18(7):26-30.
[37]ChunFu Lin, ShengDe Wang. Fuzzy Support Vector Machines[J]. IEEE Transactions on Neural Networks, 2002, l3(2):464-471.
[38]基于语义的图像检索技术研究.西安电子科技大学硕士学位论文.2006,pp. 36～37
[39]袁杰.基于内容的综合多特征图像检索方法研究.浙江师范大学硕士学位论文. 2009, pp. 66～67
[2]张磊.图像分类和图像语义标注的研究.山东大学硕士学位论文.2008, pp. 22～33
[3]NiblackW.,etal.“The QBIC Projeet:Querying Images by Content Using Color, Texture and Shape”,in Proeeedihgs SPIE,San Jose,CA,1993,28(4):56-58.
[4]A.Pentland,R.W.Pieard,Selaroff,“Content-based Manipulation of image databases”, In International Journal of Computer Vision,1996,12(2):345-248.
[5]J.R.Bach,C.fuller,A.Gupta,A.Hampapur,“The Virage images search engine:An open framework for image management”,in Proc.SPIE Storage and Retrieval for Image and Video Databases,1989,37(11):122-124.
[6]RuiY,HuangTS,Meht.tras,etal.“Content-based image retrieval With relevance feedback”in MARS.Proe.ICIP,1997,2:815-818.
[7]赵继东,鲁珂,吴跃.一种基于谱图理论的Web图像搜索方法[J].计算机应用研究.2008,22(2):23-24.
[8]周明全,耿国华,韦娜著.基于内容图像检索技术[M].北京:清华大学出版社, 2007:35～36.
[9]忠伟,章毓晋.综合利用颜色和纹理的特征检索[J].通信学报,1999,21(5):36-37
[10]沈云涛.基于视觉感知特性的图像检索研究[D].西北工业大学博士学位论文,2005,pp. 35～36
[11]L.Tao and G.xu.Color in Machine Vision and its APPlication.Chinese Science Bulletin,2001,46(17),1411-1421
[12]H.Palus.RePresentations of Colour Images in Different Colour SPaces.The colour imagehand book(eds.5.j.sangwine and r.e.n.horne),london:chapman&hall,1998,79-92.
[13]F. R. K. Chung. Spectral Graph Theory, volume 92 of Regional Conference Series in Mathematics. AMS, 1997.45(8):212-213
[14]张芳芳.基于内容的图像检索系统优化方法研究.武汉大学硕士学位论文.2004, pp. 21～22
[15]程珺.基于相关颜色的图像检索.华中科技大学硕士学位论文.2006, pp. 12～14
[16]史忠植,王文杰著.人工智能[M].北京:国防工业出版,2007, pp. 31～32
[17]甘赟,方志军.基于HSV空间的彩色图像直方图检索技术[J],科技广场,2010,3(3):86-87
[18]周明全,耿国华,韦娜著.基于内容图像检索技术[M].北京:清华大学出版社, 2007:17～18.
[19]Manik Varma and Andrew Zisserman. A Statistical Approach to Texture Classication from Single Images.IJCV(62), No. 1-2, 2005: 61-81.
[20]Stricker M A,Orengo M.Similarity of color Images.SPIE Proceedings,1995,2420:381-392
[21]王涛,胡事民,孙家广.基于颜色-空间特征的图像检索[J].软件学报, 2002,13(10):2032-2033
[22]蔡丹琳.基于SVM反馈和多特征融合的图像检索技术研究.华侨大学硕士学位论文.2007, pp. 33～34
[23]R.M.Haralick,K.Shanmngam,and I.Dinstein,“Texture Feature for Image Classification,”IEEETransactionsonSystems,ManandCybernetics,vol.3,no.6,pp.610-621,1973.
[24]魏翩翩.基于内容的图像检索技术研究及系统实现.厦门大学硕士学位论文.2009,pp.19-20
[25]邓诚强,冯刚.基于内容的多特征综合图像检索[J]计算机应用,2003:19(10):55-56.
[26]谢邦旺.基于形状的图像检索.苏州大学硕士学位论文.2007, pp.17-18
[27]周明全,耿国华,韦娜著.基于内容图像检索技术[M].北京:清华大学出版社, 2007:72～73.
[28]李俊山,李旭辉.数字图像处理[M].北京:清华大学出版社.2007:279-280.
[29]袁杰.基于内容的综合多特征图像检索方法研究.浙江师范大学硕士学位论文. 2009, pp. 23～24
[30]Bosch,A,Zisserman,A,Munoz,X Representing shape with a spatial pyramid kernel New York,NY,USA:ACM (2007)401-408
[31]周明全,耿国华,韦娜著.基于内容图像检索技术[M].北京:清华大学出版社, 2007:147～148
[32]宋炜,张铭.语义网简明教程[M].北京:高等教育出版社. 2004: 10-11.
[33]孙亮,任小康.基于本体的图像语义检索模型[J] .重庆工学院学报(自然科学版),2009,23(1):129-130.
[34] T Berners– Lee, J Hendler, O Lassila,“The semantic Web”,ScientificAmerican , 2001,284(5) :34～43
[35]张辉,张浩,陆剑峰.SVM在数据挖掘中的应用[J].计算机工程,2004,3(6):7-8.
[36]张华煜,邢丽萍.基于核函数的支持向量机分类方法[J].电脑开发与应用,2005,18(7):26-30.
[37]ChunFu Lin, ShengDe Wang. Fuzzy Support Vector Machines[J]. IEEE Transactions on Neural Networks, 2002, l3(2):464-471.
[38]基于语义的图像检索技术研究.西安电子科技大学硕士学位论文.2006,pp. 36～37
[39]袁杰.基于内容的综合多特征图像检索方法研究.浙江师范大学硕士学位论文. 2009, pp. 66～67