基于本体和自动标注的网络邮票图像语义检索研究——以南京邮电大学数字邮票库为例

Research on Semantic Retrieval of Web Stamp Images Basing on Ontology and Automatic Annotation: Case Study of Digital Stamp Database of Nanjing University of Posts and Telecommunications

针对网络邮票图像的特点,提出邮票领域本体构建方法。根据网络邮票图像的视觉特征和描述文本,利用本体描述其语义特征,通过自动图像标注技术构建邮票图像本体库,并构建网络邮票图像的语义检索系统。实验表明,该系统解决了网络图像基于关键字检索和基于内容检索中的语义缺失问题,具有较高的图像检索准确率。

In view of characteristics of web stamp images, the paper puts forward constructing method of stamp domain ontology. According to the visual features and description texts of web stamp images, the paper uses ontology to describe its semantic features, builds a stamp image ontology database by automatic image annotation technology, and constructs a semantic retrieval system of web stamp images in the database. Experiment shows that this retrieval system solves problems of semantics deficiency in the keywords- based and content-based web image retrievals, and has high image retrieval accuracy.