摘要
如何跨越图像低层视觉特征到高层语义特征的"语义鸿沟"已成为语义图像检索问题的关键,首先将待分类图像分成五个区域;然后在提取图像底层特征的基础上,采用基于支持向量机组(SVMS)的方法建立图像低层视觉特征到高层语义特征之间的映射,将一幅图像同时归入一类或几类图像语义。实验结果表明,该方法具有较好的检索查全率和准确率。
The solution of ＂semantic gap＂ which existence between the low-features and the high-level semantic features had become the key in problems of the semantic image retrieval, First separated the image into five part, then extracted low-level features, used a new approach to establish a link from image low-level feature to high-level semantic based on support vector machines. Finally, the images were classified as one or several classes. The experiment proves that it has obtained the high accuracy.
出处
《计算机应用研究》
CSCD
北大核心
2008年第2期452-454,共3页
Application Research of Computers
基金
国家自然科学重点基金资助项目(60234030)
国防科工委资助项目
关键词
语义图像检索
底层特征
高层语义
支持向量机
semantic image retrieval
low-level feature
high-level semantic
support vector machine （SVM）
