北京邮电大学学报

  • EI核心期刊

北京邮电大学学报 ›› 2016, Vol. 39 ›› Issue (5): 16-19.doi: 10.13190/j.jbupt.2016.05.004

• 论文 • 上一篇    下一篇

相关反馈结合鲁棒局部二值模式的CBIR算法

孙挺1,2, 耿国华2   

  1. 1. 周口师范学院 网络工程学院, 河南 周口 466001;
    2. 西北大学 可视化研究所, 西安 710069
  • 收稿日期:2015-07-13 出版日期:2016-10-28 发布日期:2016-12-02
  • 作者简介:孙挺(1972-),男,博士,副教授,E-mail:suntingzk@126.com.
  • 基金资助:
    国家重点基础研究发展计划(973计划)前期研究专项(2011CB311802);河南省科技厅科技发展计划科技攻关项目(122400450356);河南省科技厅科技发展计划软科学项目(132400410927)

CBIR Algorithm Based on Relevance Feedback and Robust Local Binary Patterns

SUN Ting1,2, GENG Guo-hua2   

  1. 1. College of Network Engineering, Zhoukou Normal University, Henan Zhoukou 466001, China;
    2. Visualization Institute, Northwestern University, Xi'an 710069, China
  • Received:2015-07-13 Online:2016-10-28 Published:2016-12-02

摘要: 提出了一种鲁棒局部二值模式结合相关反馈的基于内容的图像检索算法.鲁棒局部二值模式是一种性能较好的特征提取算子,对噪声和光照变化具有较强的鲁棒性,且不会造成原始数据的改变,可提高特征提取的准确性.相关反馈使系统可获知用户的偏好,对检索结果具有导向作用.多个纹理数据库的实验结果表明,提出算法的检索准确性和鲁棒性优于同类算法.

关键词: 鲁棒局部二值模式, 图像检索, 相关反馈, 特征提取

Abstract: A algorithm of content based image retrieval (CBIR) based on robust local binary patterns (RLBP) and relevance feedback (RF) was proposed. RLBP is a kind of feature extraction operator with good performance, which has strong robustness to the noise and illumination changes. The original data will not be changed with RLBP, and the accuracy of feature extraction can be improved. RF enables the system to learn the user's preferences and guide the search results. Experiments with several texture databases show that the accuracy and robustness of the proposed algorithm is better than that of the similar algorithms.

Key words: bobust local binary pattern, image retrieval, relevance feedback, feature extraction

中图分类号: