一种基于SVM和主动学习的图像检索方法  

Image retrieval method based on SVM and active learning

作　　者：王新建[1] 罗光春[1] 秦科[1] 陈爱国[1] 赖云一[1] Wang Xinjian Luo Guangchun Qin Ke Chen Aiguo Lai Yunyi(School of Computer Science ＆ Engineering, University of Electronics Science ＆ Technology of China, Chengdu 611731, China)

机构地区：[1]电子科技大学计算机科学与工程学院,成都611731

出　　处：《计算机应用研究》2016年第12期3836-3838,3846,共4页Application Research of Computers

基　　金：四川省科技厅资助项目(2012GZ0088;2013JQ0005);中央高校基本科研业务费资助项目(ZYGX2013J071)

摘　　要：为了提高图像检索的速度和准确性,提出了一种基于SVM和主动学习的图像检索方法。该方法分为两个阶段:第一阶段用K-means聚类算法在图像数据库中找出代表性样本,有效地缩小了目标图像的查找范围;第二阶段通过对未标注样本与分类边界之间的距离以及其近邻密度进行综合评价,选出最有价值的关键性样本作为训练样本,使得分类器可以通过较少的反馈次数快速达到较高的准确性。实验表明,该算法可以有效提高图像的检索性能。In order to improve the speed and accuracy of image retrieval, this paper proposed an image retrieval method based on SVM and active learning. It involved two stages. Fristly,it utilized K-means clustering algorithm to find representative samples from the image database,which may effectively reduce the search range. Secondly,it used a comprehensive evaluation to the unlabeled samples by the distance of the samples with the classification of the border and its neighbor density ,then selected the most valuable samples as training samples, which might the classifier achieve higher accuracy by a small number of feed- back times. Experimental results show that the algorithm can greatly improve the performance of the image retrieval.

关 键 词：图像检索 SVM 主动学习 K—means 代表性样本 关键性样本 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]