摘要
为了在图像语义标注领域能更好地反映标注之间的关系,通过对已标注图像的标注进行分析来建立标注之间的关系,在此基础上将叙词查询的概念引入到图像语义标注中,并提出了基于叙词查询的图像语义标注方法,把语义标注问题统一在叙词查询与图像的语义关系相结合的统一的框架下。最后通过在Corel图像数据库中的验证表明,所提出的方法是有效的,并且标注率得到了明显的提高。
To better represent the correlation among keywords in image annotation domain, this paper analyzed the annotated images and built the correlation of keywords. Furthermore, studied the traditional image annotation framework and presented a novel image annotation framework based on thesaurus retrieval, which used the visual features to retrieve the related images from image database, then embedded the correlation among keywords in the proposed image annotation framework, it is the first time to introduce thesaurus retrieval in image annotation domain, experimental results show that the performance of the annota- tion is greatlv imoroved on Corel datasets.
出处
《计算机应用研究》
CSCD
北大核心
2011年第8期3174-3177,共4页
Application Research of Computers
基金
国家教育部科研重点资助项目(107021)
关键词
语义标注
叙词查询
图像检索
in image annotation domain
this paper analyzed the annotated images and built the correlation of Key words.Furthermore
studied the traditional image annotation framework and presented a novel image annotation framework based on thesaurus retrieval
which used the visual features to retrieve the related images from image database
then embedded the correlation among Key words in the proposed image annotation framework
it is the first time to introduce thesaurus retrieval in image annotation domain
experimental results show that the performance of the annotation is greatly improved on Corel datasets. Key words： semantic annotation
thesaurus retrieval
image retrieval
