一种基于概率粗糙集模型的图像语义检索方法*

南京大学学报(自然科学版) ›› 2011, Vol. 47 ›› Issue (4): 438–455.

一种基于概率粗糙集模型的图像语义检索方法*

徐久成，李晓艳**，孙林

出版日期:2015-04-21 发布日期:2015-04-21
作者简介: (河南师范大学计算机与信息技术学院，河南新乡，453007)
基金资助:
国家自然科学基金(60873104,61040037),河南省省教育厅自然科学基金(20088B20019)

An image semantics retrieval method based on probability rough set model

Xu Jiu-Cheng，Li Xiao-Yan，Sun Lin

Online:2015-04-21 Published:2015-04-21
About author: (College of Computer and information Technology, Henan Normal University, Henan Xinxiang,453007，China)

摘要/Abstract

摘要： 针对目前图像数据信息量大、检索不易和人们对图像检索习惯于对图像概念语义进行检索的难题，木文将概率粗糙集理论和图像的语义标注技术引入图像的信息检索中，提出了一种基于朴素贝叶
斯理论和概率粗糙集模型的图像语义信息检索模型.首先，针对图像库中的图像构造精确标注词空间，并通过朴素贝叶斯理论对图像进行精确标注和模糊加权标注.将概率粗糙集模型和朴素贝叶斯理论的
后验概率相结合，计算每对图像标注词的条件概率和模糊条件概率，并求得每个标注词的支持集和被支持集，在此基础上，计算每个标注词的支持集和被支持集的上、卜近似，并通过上、卜近似构造图像的语
义相似度计算方法，之后计算待查询图像的查询特征与图像库中图像之间的语义相似度，并根据相似度的大小给出检索的排序和输出.最后，给出一个简单的仿真实验，实验结果表明该方法是有效可行的.

Abstract: In view of the problem that the present image data contains too much information, not easy to retrieve and people is used to searching the image semantic concept in the process of image retrieval，this paper introduces
the probabilistic rough set model and image semantic annotation technology into the image information retrieval and puts out a kind of image information retrieval model based on Naive Baycsian and probability rough set model.
Firstly, the images arc crisp annotated and weighted fuzzy annotated by Naive Baycsian after building crisp annotation space for the image library information. Secondly, the conditional probability and fuzzy conditional
probability arc computed for each pair of annotations by combining the posterior probability of Naive Baycsian with the probabilistic rough set model，after that the support set and supported set of each annotation arc computed，the
upper and lower approximation of each annotation arc calculated for the support set and supported set，and the image semantic similarity calculation method is built by the upper and lower approximation, then the semantic similarity
between query characteristics and the characteristics of image library is calculated，and the retrieval results arc output by ordering according to similarity. Finally, taking the images in Corel image library for an example，the
paper give a simple simulation experiment which shows the algorithm is feasible and effective in practice.

徐久成，李晓艳**，孙林
. 一种基于概率粗糙集模型的图像语义检索方法*[J]. 南京大学学报(自然科学版), 2011, 47(4): 438–455.

Xu Jiu-Cheng，Li Xiao-Yan，Sun Lin
. An image semantics retrieval method based on probability rough set model
[J]. Journal of Nanjing University(Natural Sciences), 2011, 47(4): 438–455.

参考文献

[1]Wu N, Song F M. An image retrieval method based on higlrlevcl image semantic information. Journal of magc and Uraphics, 2006，11(12):
1774-1780.(吴楠，宋方敏.一种基于图像高层语义信息的图像检索方法.中国图象图形学报，2006，11(1 2):1774一1980).
[2]Li M, Wang T.A method for semantic image retrieval based on rough set. Journal of Lanzhoc University of Technology, 2006，32(2):98一 101.(李明，土彤一种基于Rough集的语义图像检索方法.兰州理工大学学报，2006,32 (2):98一101).
[3]Gao Y Y，Yao F T，Zhang S Y. 3D model se mantic classification and retrieval with Uaussiar processes. Journal of Zhejiang University(E n- ginccring Scicncc)，2010，44(12):2261一2266.
(高波涌，姚伏天，张三元.基于高斯过程的三维模型语义分类和检索.浙江大学学报(工学版)，2010，通通(12) :2261一2266).
[4]Zhu W Q, Liu Q. New method to automatically annotate and retrieve images. Application Re search of Computers, 2007，2’1(7):318一320. (朱文球，刘强.一种新的图像语义自动标注与检索算法.计算机应用研究，2007, 24(7); 318一320.
[5]Tao W S,Tao X P，Lv J. Rule and similarity based semantic web service matchmaking. Jour- nal of Nanjing University(Natural Sciences)，2010,46(2);159一168.陶望胜，陶先平，吕
建.基于规则与相似度的语义wcl〕服务匹配.南京大学学报(自然科学),2010,46(2);159~ 168).
[6]Guivacla V N，Raghavan V V，Grosky W l，et al. Information retrieval on the world wide web. IEEE Internet Computing, 1997:58一68.
[7]Pawlak Z. Rough sets, International Journal of Computer and information Sciences，1982，11 (5):341一356.
[8]Nguye J. Classification based on optimal feature extraction and the theory of rough sets. New- townabbcy; University of Ulster at Jordans- town, 1995. 131一137.
[9]Intan R，Mukaidono M. Gencralizcd fuzzy rough sets by conditional probability relations. international Journal of Pattern Recognition and Artificial lntclligence, 2002，16(7):865一881.
[10]HuJ，Cao X B, Wang X F. Uraphics and image pre-indexing based on tolerance rough sets. Journal of Computer Aided Design and Comput- er Uraphics, 2002，l4l(3) ;212一246.(胡静，
曹先彬，土煦法.基于相容粗糙集的图形图像信息预检索.计算机辅助设计与图形学学报，2002，14(3):242一2 r1 6}.
[11]Huang Z U, Zhu C X. information retrieval based on probability rough set model. Computer Engineering, 2008，34(23):193一195.(黄治国，朱承学.基于概率粗糙集模型的信息检索.计算机工程，2008,34(2 3):193一195).
[12]Zhang Y Q, Bao J J. image annotation and re tricval based on Baycsian theory. Computer Sci- ence, 2008,35(8);229-231.(张元清，包骏杰. 基于贝叶斯理论的图像标注和检索.计算机科学，2008,35 (8):229一231).
[13]WuJ，LinGK，Lu M Y，et al. Enhancing rel- evance feedback in image retrieval using asym- metric Baycsian learning. Journal of Nanjing U- niversity( Natural Sciences)，2009,45(5):604一
612.乌日俊，林正奎，鲁明羽等.基于不对称贝叶斯学习的图像检索相关反馈算法研究.南京大学学报(自然科学),2009,45(5):604~612.
[14]zhong H，Xia L M. Ontology-based image retrieval. Computer Engineering and Applica- tions,2007, X3(17);37~40.(钟洪，夏利民.基于木体图像检索.计算机工程与应用，2007, 43(17):37~40).
[15]httl://download. csdn, net/down/1504820/ gaojin8502.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed