计算机与现代化

• 应用与开发 • 上一篇    下一篇

基于融合多特征的社会网上图像检索方法

郭海凤,李广水,仇彬任   

  1. 金陵科技学院信息技术学院,江苏南京211169
  • 收稿日期:2013-07-26 修回日期:1900-01-01 出版日期:2013-12-18 发布日期:2013-12-18

Method of Social Website Image Retrieval Based on Multi-feature Fusion

GUO Hai-feng, LI Guang-shui, QIU Bin-ren   

  1. College of Information Technology, Jinling Institute of Technology, Nanjing 211169, China
  • Received:2013-07-26 Revised:1900-01-01 Online:2013-12-18 Published:2013-12-18

摘要: 由于用户标注的非专业性及随意性,社会图像网站在图像检索过程中,其检索结果一般不够理想,针对这一现状,提出融合多特征的图像检索方法。依据用户输入的初始检索词,由系统提供给用户候选图像集,依据用户选择的目标图像,结合候选集中图像底层特征、社群主题、标签顺序等多属性构建出某一候选图像的多特征向量,依据计算不同向量与目标图像之间的距离,提交给用户距离最近的图像集,完成图像检索。仿真实验依据Flickr网站上真实数据展开,试验结果表明本方法的有效性。

关键词: 社会网, 图像检索, 颜色特征, 社群主题, 标签顺序

Abstract: ecause of users’ non-professional and randomness, in the process of retrieval image, social website retrieval can not ideally generate sufficient result for people. In view of this situation, this paper proposes a method of image retrieval based on multi-feature fusion. According to the initial term of users input, the system offers users a set of candidate images, based on the user’s choice of target image, combining the low-level features of candidate images sets, community themes and label order, this paper constructs a candidate image multi-feature vector. The system can calculate the distance between the vector and the target image, submits the nearest image collection to user. Simulation experiments are done based on real data from Flickr website, and the test results prove that this method is effective.

Key words: social website, image retrieval, color feature, community themes, label order