一种新的在图像关联规则挖掘中产生频繁项集的方法
A New Approach to Generate Frequent Itemsets in Mining Image Association Rules
-
摘要: 提出了一种进行图像关联规则提取时产生频繁项集的方法——频繁项树.为便于频繁项树的运用,使用了bSQ的图像数据格式来重新组织图像数据,并在此基础上提出了频繁项树的截断、半深度优先、图像掩模和多层次灰度范围自动生成等优化技术,降低了算法的时间和空间复杂度,使其具有较高的运行效率和实用价值.Abstract: This paper proposes an approach to generate frequent itemsets in mining image association rules——Frequent Item Tree.We utilize bSQ image format to re-organize the image data to apply frequent item tree.Moreover, this paper propose several optimization techniques, including frequent item tree pruning, semi-depth-first search, image mask, and multi-level gray generation, to decrease the time and space complexity.