摘要
提出一种适应图书馆特点的视觉特征和高层语义相结合的图像检索算法,通过相关反馈构建了动态的相似性度量方程。实验结果表明,综合视觉特征和语义特征的检索比仅利用视觉特征的检索能获得更高的检索率。
In this paper, a new image retrieval algorithm in digital library is proposed. It is based on the combination of visual features and semantic contents, in addition to giving a dynamic similarity metric eguation by relevance feedback technique. Experiments on our prototyping system showes that it can get higher retrieval rate than only by visual features.
出处
《现代图书情报技术》
CSSCI
北大核心
2005年第12期30-33,共4页
New Technology of Library and Information Service
关键词
图像检索
相关反馈
动态相似性度量
Image retrieval Relevance feedback Dynamic similarity metric
