一种基于概念层次的图像检索方法  

A Novel Concept-based Image Retrieval Method

作　　者：陈世亮[1] 李战怀[1] 袁柳[1] 

机构地区：[1]西北工业大学计算机学院,西安710072

出　　处：《计算机科学》2008年第4期139-141,153,共4页Computer Science

基　　金：国家自然科学基金项目资助(60373108)

摘　　要：针对现有图像检索方法中存在的不同用户间因对图像的理解差异而存在的歧义问题,通过引入概念抽象的思想,将图像标注和检索中语义相关的关键词联系起来,统一不同用户对图像语义的理解和描述;结合相关反馈技术,确定检索关键词所代表语义概念信息,并实现概念层次的图像检索。实验结果表明,该方法能有效地提高图像的检索性能。To bridge the semantic gap between user＇s description and annotated labels, much interesting work has been attempted. Relationships between image annotation keywords and user used in retrieval are built by introducing the abstraction of concept, and then the description of image annotation is unified. By adopting correlative feedback technologies, the concept of the query keyword is selected, and then the concept-based image retrieval method is proposed. The experiments show that this method can improve the performance of retrieval

关 键 词：概念语义 图像检索 语义层次 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TP311.13[自动化与计算机技术—计算机科学与技术]