融合相关反馈和流形学习的语义图像检索方法研究  

Research on Semantic Image Retrieval Methods Integrating Relevance Feedback with Manifold Learning

作　　者：刘利[1] 

机构地区：[1]惠州学院计算机科学系,惠州516007

出　　处：《现代计算机》2013年第18期17-20,共4页Modern Computer

基　　金：惠州市科技计划项目(No.2012-7;No.2012-5;No.2011B020006009)

摘　　要：融合相关反馈和流形学习的图像检索方法,既可以解决基于内容图像检索的"语义鸿沟"问题,又可以解决因为用户反馈标记样例较少所导致的较难学习用户语义概念问题。深入研究近年来比较有代表性的方法,包括ARE和MMP,并在具体的系统中比较二者的性能;此外,进一步分析此类方法面临的挑战和实际应用时需迫切解决的问题。Image retrieval methods which integrate relevance feedback with manifold learning attract a lot of attention of researchers. These methods can narrow semantic map and gain semantic conception from little labeled samples difficultly. Studies the representative methods in recent years including ARE and MMP, and compares their performance in an image retrieval system. In addition, analyzes the challenges and problems of such methods in practical application.

关 键 词：相关反馈 流行学习 基于内容的图像检索 语义图像检索 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术]