基于聚类和用户兴趣模型的个性化图像搜索

郭璐; 宋京; 杜晶晶; 许哲; 黄鹤

doi:10.3788/OPE.20172513.0199

您当前的位置：

首页 >

文章列表页 >

基于聚类和用户兴趣模型的个性化图像搜索

信息科学 | 更新时间：2020-08-13

- 基于聚类和用户兴趣模型的个性化图像搜索
- Personalized image searching based on clustering analysis and user interest model
- 光学精密工程 2017年25卷第10s期页码：199-204
- 作者机构：
  
  1. 中电科第二十研究所, 陕西西安 710068
  2. 西安爱生技术集团公司, 陕西西安 710065
  3. 无人机系统国家工程研究中心,陕西西安,710072
  4. 长安大学电子与控制工程学院,陕西西安,710064
- 作者简介：
- 基金信息：
  
  中国博士后科学基金资助项目（No.2013M542310）();国家自然科学基金资助项目（No.41101357）()
- DOI：10.3788/OPE.20172513.0199
  中图分类号： TP391.4
- 收稿日期：2017-05-24，
  
  修回日期：2017-06-19，
  
  纸质出版日期：2017-11-25
- 稿件说明：
移动端阅览
郭璐, 宋京, 杜晶晶等. 基于聚类和用户兴趣模型的个性化图像搜索[J]. 光学精密工程, 2017,25(10s): 199-204

GUO Lu, SONG Jing, DU Jing-jing etc. Personalized image searching based on clustering analysis and user interest model[J]. Editorial Office of Optics and Precision Engineering, 2017,25(10s): 199-204
郭璐, 宋京, 杜晶晶等. 基于聚类和用户兴趣模型的个性化图像搜索[J]. 光学精密工程, 2017,25(10s): 199-204 DOI： 10.3788/OPE.20172513.0199.

GUO Lu, SONG Jing, DU Jing-jing etc. Personalized image searching based on clustering analysis and user interest model[J]. Editorial Office of Optics and Precision Engineering, 2017,25(10s): 199-204 DOI： 10.3788/OPE.20172513.0199.

摘要

针对现有图像搜索方法难以完整地评价用户查询目的及图像检索质量较低的问题，提出了一种基于聚类和用户兴趣模型的个性化图像搜索方法。该方法首先将输入的检索图像拆分成9个子块，并将图像转换到HSV颜色空间中，利用图像的颜色分布直方图提取颜色特征信息；然后采用Gabor小波提取图像的纹理特征，将获取到的颜色特征和纹理特征进行融合以构成图像的多特征融合相似度矩阵，计算图像间的相似性。再以多特征融合相似度矩阵为多核的动态聚类的输入，对数据库中图像进行聚类，将聚类图像送入LSSVM网络，确定分类面并构建个性化用户兴趣模型；最后按照与用户兴趣模型相似度高低程度进行比较，将检索结果提供给用户自主选择。实验结果表明：与单一的颜色和纹理特征搜索算法相比，所提方法的平均查全率、查准率分别提升了8.2%、11.42%和19.7%、26.08%，可以有效地提升图像搜索的质量，应用价值明显。

Abstract

Aimed at the problem that existing image searching algorithms were difficult to evaluate query purpose of users completely and the quality of image retrieval was low

a personalized image searching method based on clustering analysis and user interest model was proposed in the thesis. Input retrieval image was divided into 9 sub-blocks and transformed into HSV color space

and color distribution histogram of image was used to extract color feature information. And then

Gabor wavelet was used to extract texture features of the image

and obtained color features and texture features were fused to form multi-feature fusion similarity matrix of the image to calculate similarity among images. Then multi-feature fusion similarity matrix was used as input of multi-core dynamic clustering to cluster the images in the database. The clustering image was sent to LSSVM network to determine the classification surface and construct personalized user interest model. Finally

retrieved results were provided to users for independent choice according to comparison with similarity degree of user interest model. Experimental result shows that:average recall ratio and precision ratio of method in the thesis are improved by 8.2%

11.42% and 19.7%

26.08% compared with search algorithm of single color and texture features. It can promote quality of image searching effectively and has obvious application value.

关键词

Keywords

references

GAO Y, PENG J, LUO H, et al.. An interactive approach for filtering out junk images from keyword-Based goolge search results[J]. IEEE Transaction on Circuits and Systems for Video Technology, 2009,19(12):1851-1865.

方爽,殷俊杰,徐武平. 基于相似图片聚类的Web文本特征算法[J]. 计算机工程,2014(12):161-165,171. FANG SH, YIN J J, XU W P. Web Text Feature Algorithm based on similar image clustering[J]. Computer Engineering, 2014(12):161-165,171.(in Chinese)

YAN K, FENG X Y, HUANG H, et al.. A novel algorithm for filtering out junk images interactively from web search results[C]. IEEE International Conference on Computer Science and Information Technology, 2010,8:195-199.

段娜,王磊,等. 全局及其个性化区域特征的图像检索[J]. 计算机科学, 2016, 43(s2):205-207. DUAN N, WANG L, et al.. Image retrieval of global and personalized ROI adjustment of features[J]. Computer Science,2016, 43(s2):205-207.(in Chinese)

谷瑞军,陈圣磊,陈耿,等. 图像搜索结果的重叠层次聚类与代表点展现[J]. 计算机应用,2012(04):1097-1100. GU R J, CHEN SH L, CHEN G, et al..Hierarchical overlapping clustering and exemplar visualization of images returned by search engine[J]. Journal of Computer Applications,2012(04):1097-1100.(in Chinese)

张玉兵,宋威. 视觉特征的分块加权图像检索方法[J]. 计算机科学与探索, 2017,11(3):468-477. ZHANG Y B, SONG W. Block weighted image retrieval method based on visual features[J]. Journal of Frontiers of Computer Science & Technology,2017,11(3):468-477.(in Chinese)

吴晓雨, 何彦, 杨磊,等. 基于改进形状上下文特征的二值图像检索[J]. 光学精密工程, 2015, 23(1):302-309. WU X Y, HE Y, YANG L, et al.Binary image retrieval based on improved shape context algorithm[J]. Opt. precision Eng.,2015, 23(1):302-309. (in Chinese)

谢莉,成运,曾接贤,等. 基于颜色和梯度方向共生直方图的图像检索[J]. 计算机工程与应用, 2016, 52(10):181-186. XIE L, CHENG Y, ZENG J X, et al.. Image retrieval based on color and motif gradient direction co-occurrence histogram[J]. Computer engineering and applications, 2016, 52(10):181-186. (in Chinese)

翟铭晗,高玲. 基于加权颜色分层和纹理单元的图像检索算法[J]. 计算机应用,2016, 36(6):1668-1672. ZHAI M H, GAO L. New image retrieval method based on weighted color stratification and texture unit[J]. Journal of Computer Applications,2016, 36(6):1668-1672.(in Chinese)

FAN J, GAO Y, LUO H. Integrating concept ontology and multi-task learning to ahieve more effective classifier training for multi-level image annotation[J]. IEEE Transaction on Image Processing, 2008,17(3):407-426.

汪华章,何小海,宰文姣. 基于局部和全局特征融合的图像检索[J]. 光学精密工程, 2008, 16(6):1098-1104. WANG H ZH, HE X H, ZAI W J.Image retrieval based on combining local and global features[J]. Opt. Precision Eng.,2008, 16(6):1098-1104.(in Chinese)

赵洪伟,谢永芳,曹斌芳,等. 基于Gabor小波和LPP的浮选过程泡沫纹理特征提取及应用[J]. 上海交通大学学报,2014(7):942-947. ZHAO H W, XIE Y F, CAO B F, et al..Extraction and application of froth texture feature based on gabor wavelets and LPP in flotation process[J]. Journal of Shanghai Jiaotong University,2014(7):942-947.(in Chinese)

浏览量

487

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

利用空间分布熵的改进VLAD图像检索

结合主体检测的图像检索方法

结合深度卷积网络与加速鲁棒特征配准的图像精准定位

基于改进形状上下文特征的二值图像检索

压电陶瓷轮胎发电机的设计