《计算机应用研究》|Application Research of Computers

基于SVM的图像低层特征与高层语义的关联

Correlation of Image Low level Features and High level Semantics Using SVM

免费全文下载 (已被下载 次)  
获取PDF全文
作者 成洁,石跃祥
机构 湘潭大学 信息工程学院,湖南 湘潭 411105
统计 摘要被查看 次,已被下载
文章编号 1001-3695(2006)09-0250-03
DOI 10.3969/j.issn.1001-3695.2006.09.081
摘要 在基于内容的图像检索中,针对图像的低层可视特征与高层语义特征之间的鸿沟,提出了一种基于支持向量机(SVM)的语义关联方法。通过对图像低层特征的分析,提取了颜色和形状特征向量(221维),将它们作为支持向量机的输入向量,对图像类进行学习,建立图像低层特征与高层语义的关联,并应用于鸟类、花卉、海洋以及建筑物等几个典型的语义类别检索。实验结果表明,该方法可适应于不同用户的图像检索,并提高了检索性能。
关键词 支持向量机(SVM);低层特征;高层语义;基于内容的图像检索
基金项目 国家自然科学基金资助项目(60234030);湖南省自然科学基金资助项目(02jjy2091)
本文URL http://www.arocmag.com/article/1001-3695(2006)09-0250-03.html
英文标题 Correlation of Image Low level Features and High level Semantics Using SVM
作者英文名 CHENG Jie, SHI Yue-xiang
机构英文名 College of Information Engineering, Xiangtan University, Xiangtan Hunan 411105, China
英文摘要 A new method for correlating image lowlevel feature with highlevel semantic based on SVM is proposed, aiming at overcoming the considerable gap between them in the field of contentbased image retrieval. Through analyzing the image lowlevel features, color and shape feature vectors are selected as SVM’s input vectors. Then make study of image classes to build the correlation from image lowlevel features to highlevel semantics. This semantic correlation method has been used in semantic retrieval, which concerns the following typical semantic categories: birds, flowers, sea and buildings. Experimental results demonstrate that the algorithm can adapt to the various users’ image retrieval and improve the retrieval function.
英文关键词
参考文献 查看稿件参考文献
 
收稿日期
修回日期
页码 250-252
中图分类号
文献标志码 A