图象数据关联规则挖掘

ASSOCIATION RULES MINING OF IMAGE DATA

本文在介绍了多媒体数据挖掘基本思想和图象数据主要特征基础之上 ,对图象数据关联规则挖掘进行了讨论 .首先对关联规则挖掘有关问题加以描述和分析 ,针对图象数据 ,定义了新的关联规则及其有关概念 ,并给出考虑项数量的原子特征和空间关系关联规则挖掘算法 Max Occur和 MAR- Spatial。

Based on the introduction of the basic method of multimedia data mining and main features of image data, the paper discusses the mining of AR (association rule) in image data. The paper first describes and analyzes some issues about AR mining. Secondly, as for image data, the paper introduces the new definition of AR and its relevant concepts, and offers two AR mining algorithms about atomic features and spatial relationship, MaxOccur and MAR Spatial, which consider the number of items and solve the problem of information losing resulting from the ignorance of items recurrence.