【摘要】 针对基于图像高层语义的检索方法问题,提出了一种基于自然场景图片语料库下的图片本体检索框架.该方法通过考虑图像内容之间的对应关系,引入人类正常图像检索过程.利用对图像内容的分块语义注释的描述,该框架解决了目前图像检索中的问题.构建好的领域本体可以用来完成定性语义描述和检索,实现了包含语义概念和空间关系的自然语言查询.最后,对该应用的有效性进行了评估,实验结果表明,该框架在检索的精确度和相关度上都表现得非常好.更多还原

【Abstract】 In order to solve the problem of image semantic retrieval based on high level image,this paper proposes an image ontology retrieval framework based on natural scene image corpus.In this method,the human normal image retrieval process is introduced by taking into account the correspondence between image contents.The proposed architecture addresses the problem of image retrieval through the use of image block representations over semantic annotation.The domain ontology can be used to complete qualitative semantic image descriptions and retrieval,and achieve natural language queries containing semantics concepts and spatial relationships.Finally,this paper evaluates the effectiveness of the architecture,and the experimental results show that,this framework performs very well in retrieval accuracy and relevance.更多还原