欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
空间数据蕴含了大量拓扑关系语义，但传统空间数据查询和检索方法没有很好地利用高层拓扑语义，导致在处理复杂空间场景时效能较低．针对这一局限性，提出了一种基于草图内容的空间数据检索算法．该算法在9-交集拓扑模型基础上引入不变矩方法，建立拓扑不变量用于描述复杂空间场景；采用独立成分分析和模糊支持向量机降低空间场景高维拓扑关系的冗余度，建立了独立拓扑关系；用相应训练好的支持向量机结合tf×idf模型实现空间场景检索．实验表明，该算法在低样本数情况下具有良好的分类推广能力以及良好的检索性能，为基于内容的空间数据检索建立了基础．
英文摘要：
Traditional spatial data query and retrieval cannot be efficient for dealing with complex spatial scenes, because high-level topological semantics are not used. In order to overcome this limitation, sketchcontent based spatial data retrieval was presented. In this algorithm, many topological invariants based on 9-intersection model with invariant moments are defined to represent the complex spatial scenes. Mutually high-order independent component analysis （ICA） features are used to train fuzzy support vector machine （SVM） classifiers, which reduce the high dimension of spatial scene topological relations and construct independent topological vectors. The spatial scene can be retrieved based on the trained SVM and tf×idf modal. The experimental results showed that statistical learning supported spatial data retrieval has good retrieval performance under few samples, which can be the basis of content-based spatial data retrieval.
同期刊论文项目
同项目期刊论文
期刊信息
位置：