摘要
通过建立一个三级影像内容描述模型,挖掘出隐含在不同影像不同区域中的语义特征;假定每一幅影像必包含感兴趣区域和非感兴趣区域,在此基础上重新组织用来描述影像内容的语义特征,利用此特征进行影像检索。实验结果表明,该模型能很好地提取出影像内隐含的语义特征;同时使用感兴趣影像和非感兴趣影像进行检索,能够很好地提高检索精度,并且降低了检索的复杂度。
A three-level image content representation model is built to mine semantic features that hide in different images and regions. Suppose that each image contains interesting region and uninteresting region, and reorganize semantic features that are used to represent image content. We utilize semantic features to retrieve image. The results show that this model can extract implied semantic feature from remote sensing images. Simultaneously, using inter esting image and unin,teresting image to retrieve image can get a good retrieval precision.
出处
《武汉大学学报(信息科学版)》
EI
CSCD
北大核心
2009年第6期684-687,共4页
Geomatics and Information Science of Wuhan University
基金
国家863计划资助项目(2009AA12Z114
2007AA12Z148)
国家973计划资助项目(2009CB723905)
国家自然科学基金资助项目(40771139)
关键词
遥感影像
基于内容的影像检索
语义挖掘
remote sensing image
content-based image retrieval
semantic mining