摘要： 通过分析图像数据对语义信息的检索需求，提出一种表达图像信息的结构化方法，并定义一种根据图像中隐含语义信息来判定相似度的算法。该方法在一定程度上克服了基于内容的图像检索和基于文本的元数据图像检索这2种常用图像检索方法缺乏语义信息匹配上的不足。原型系统根据结构化协同标注中语义信息的丰富程度与图像检索的精确程度，证明了该方法的有效性。

关键词: 协同标注, 图像检索, 结构化, 相似度

Abstract: By analyzing image data’s demands for semantic information, a structural method is adopted to express information in images, and a similarity algorithm which contains semantic information to help retrieve images is defined. To a certain extent, this method overcomes limitations of two common image retrieval methods, i.e., content based image retrieval, and text based metadata image retrieval, both of which lack semantic information for match. The prototype system proves its effectiveness through fruitful semantic information of collaborative annotation and accuracy of image retrieval.

Key words: collaborative annotation, image retrieval, structural, similarity

中图分类号: 

• TP391