摘要： 介绍一种基于空间相似的图像数据库检索模型,提出了空间相似度计算的算法,对对象的空间方位、拓扑关系、数量和空间关系局部高度相似等因素都作了全面考虑,详细阐述空间相似度与局部相似对象簇的求解过程,建立的模型不仅能识别经平移、缩放和旋转变换的图像,而且能识别图像的对称变换,通过实验验证,表明检索模型的图像搜索效率高。

关键词: 图像检索, 空间相似度, 对象空间关系, 对象簇, 图像变换

Abstract: A novel model for retrieving image by spatial similarity in image database is presented. In this model, an effective spatial similarity algorithm is defined, which considers spatial position and topological spatial constraints, object number and high similarity of part objects' spatial relationship. Moreover, the calculation of similarity and the course of solving high similarity object clusters have been discussed in detail. Study shows that the retrieval model not only recognizes images undergoing translation, scaling and rotation transformation but also identifies symmetry transformation. Experiment testified that this model is effective.

中图分类号: 

• TP391.4