节点文献

基于语义矩阵的图像检索相关反馈技术

Relevance Feedback Technology for Image Retrieval Based on Semantic Matrix

  • 推荐 CAJ下载
  • PDF下载
  • 不支持迅雷等下载工具,请取消加速工具后下载。

【摘要】 针对基于语义网络的图像检索中的过反馈问题,本文提出了将语义网络转换为语义矩阵,采用基于多用户相关反馈的投票机制,可有效地克服过反馈。在此基础上提出了一种新颖的新图添加策略和处理语义向量查询的有效方法,最后指出了还需进一步研究的问题。

【Abstract】 To solve the problem of overfull feedback during image retrieval based on semantic network, this paper proposes to convert semantic network into semantic matrix, and utilizes a voting mechanism based on multi-user relevance feedback to effectively overcome overfull feedback. Based on the semantic matrix and voting mechanism, a novel strategy of adding new images into the retrieval system is brought forward, and an effective method of dealing with queries in the form of semantic vector is proposed. Finally, the paper puts forward the issues for further study.

【基金】 国家自然科学基金资助项目“基于语义网的多媒体知识元发现与挖掘”研究成果之一,项目编号:70503022
  • 【文献出处】 情报理论与实践 ,Information Studies:Theory & Application , 编辑部邮箱 ,2008年01期
  • 【分类号】G354
  • 【被引频次】3
  • 【下载频次】337
节点文献中: