基于约减支持向量机的相关反馈图像检索算法  

IMAGE RETRIEVAL ALGORITHM WITH RELEVANCE FEEDBACK BASED ON REDUCED SUPPORT VECTOR MACHINE

作　　者：齐亚莉[1] 李业丽[1] 

机构地区：[1]北京印刷学院计算机科学与技术系,北京102600

出　　处：《计算机应用与软件》2011年第8期149-151,共3页Computer Applications and Software

基　　金：北京市教育委员会科技发展计划面上项目(KM200910015007);北京市人才强教计划项目(PXM2010_014223_095557)

摘　　要：针对支持向量机在大规模数据集上的低效率,提出了基于约减支持向量机的相关反馈图像检索算法。首先采用约减支持向量机训练初始分类器,以该分类器作为检索模型,根据检索结果进行相关反馈,从而进行再检索。实验结果表明,随着反馈次数的增加,检索到的相关图像也会增加;另外相对传统的基于向量机的方法,数据集规模越大,基于约减支持向量机的算法在时间上的优势越明显。Aiming at the inefficiency of support vector machine applied on large data sets,an image retrieval algorithm with relevance feedback based on reduced support vector machine（RSVM） is presented.Firstly,the RSVM is used to train primary classifier,and then such classifier is taken as the retrieval model to make relevance feedback according to the retrieved result followed by further re-retrieval.Experiments show that along with the increase of the feedback times,the relevant images will be increased too.More importantly,in contrast to traditional algorithm based on SVM,the larger the database size is,the more manifest the advantage in time-consuming aspect the algorithm based on RSVM has.

关 键 词：约减支持向量机 相关反馈 图像检索 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]