一种结合地理知识的遥感影像目标实体关联方法北大核心CSCD

遥感影像目标识别技术已广泛应用于目标动态监测与定位等领域。但影像目标识别的结果缺乏与目标属性信息的链接,导致分析人员只能依据影像特征进行分析,难以进行更复杂的目标数据关联分析与挖掘。针对遥感影像目标识别语义属性信息缺失的问题,本文利用知识图谱相关技术将影像判别的目标信息与知识语义网链接。首先,提出了一种遥感影像目标知识图谱构建框架;其次,针对遥感影像目标不同的数据类型,构建遥感影像目标知识抽取模型,提出了基于相似度目标实体识别和预定义模式的关系抽取方法;然后,基于多特征Logistic模型的影像目标实体链接方法,实现了遥感影像目标实体与百科知识库的知识关联;最后,针对预定试验区域进行试验,验证了本文方法的可行性。

A remote sensing image object knowledge association method based on geographic knowledge

Remote sensing image target recognition technology has been widely used in the field of target dynamic monitoring and positioning. However, the result of image target recognition lacks the link with the target attribute information, which makes it difficult for analysts to carry out more complex target data association analysis and mining. In view of the lack of semantic attribute information in remote sensing image target recognition, the target information of image discrimination is linked to the semantic web by using knowledge graph technology. Firstly, the framework of building remote sensing image target knowledge graph is proposed. Secondly, according to different data types of remote sensing image targets, a remote sensing image target knowledge extraction model is constructed. The method of object entity recognition based on similarity and relationship extraction of predefined patterns are proposed. Then, based on the image object entity linking method considering the multiple features logistic model, the knowledge association between the remote sensing image object entity and the encyclopedia knowledge base is realized. Finally, experiments are carried out in the predetermined experimental area to verify the feasibility of this method.