基于XML文本片段的图像检索实现与评价  

XML-Text-Fragment-Based Image Retrieval:Implementation and Evaluation

作　　者：陆伟[1,2] 张宓[2] 刘丹[2] 

机构地区：[1]武汉大学信息资源研究中心 [2]武汉大学信息管理学院,430072

出　　处：《中国图书馆学报》2009年第2期53-58,共6页Journal of Library Science in China

基　　金：国家社科基金项目"基于XML的多媒体信息检索模型及其实现研究"(编号:06CTQ006)研究成果之一

摘　　要：为了研究文本片段与图像位置关系对图像检索结果造成的影响,本文采用同在域范围和RK两种层级体系来表示文本与图像的关系,并对两种体系的不同层级分别进行检索评价。实验结果表明,就整体而言,层级越靠近图像,检索效果越好;越靠近图像的文本,越能描述图像的语义信息。RK层级能更细致地表达文本片段与图像的位置关系,随着层级以图像实体为中心向外扩展,检索效果整体呈下降趋势。表1。图3。参考文献11。To analyze the influence of the position relation of text fragment and image on the retrieval efficiency, the field level they are both in and Region Knowledge are respectively used to present the different relations in this paper, and retrievals and evaluations are proposed on different levels. The results of these experiments indicate that closer distance between text fragment and image can bring more accurate retrieval result, and when presenting the relation of text fragment and image by RK, the direct operatloh on field name is avoided and the presentation of the position relation is more delicate. Among all levels centered on an image entity, the retrieval aceuxacy are declining from inside to outside.

关 键 词：XML检索 图像检索 TBIR 

分 类 号：G354.4[文化科学—情报学]