质量度量指标驱动的数据聚合与多维数据可视化  

Quality-metrics driven multi-dimensional data aggregation and visualization

作　　者：李杨[1,2] 郝志峰[2,3] 谢光强[1,2] 袁淦钊[3] 

机构地区：[1]广东工业大学自动化学院,广东广州510006 [2]广东工业大学计算机学院,广东广州510006 [3]华南理工大学计算机科学与工程学院,广东广州510006

出　　处：《智能系统学报》2013年第4期299-304,共6页CAAI Transactions on Intelligent Systems

基　　金：国家自然科学基金资助项目(61070033);广东省自然科学基金资助项目(9251009001000005);广东省科技计划资助项目(2010B050400011)

摘　　要：以多维数据可视化为研究对象,在质量度量模型下,采用数据聚合为基本手段,来提高多维数据可视化的图像质量.在质量度量指标驱动的框架下提出了均分K-means++数据聚合算法,在传统K-means算法的基础上,专门以数据可视化为目的对算法进行了改进,使得算法聚合得到的数据既能够较好地保持原数据的大部分特性,又能显著地提高可视化后的图像质量.仿真实验证明,在不同的数据抽象级别DAL下,无论是图像质量指标还是质量度量指标HDM(直方图差值度量)、NNM(最近邻距离度量),算法都表现出了较好的仿真结果.For the purpose of this research paper,we examined multi-dimensional data visualization with the quality metrics model; taking data aggregation as a basic means in order to improve the multi-dimensional visualization image quality. Under the quality-metrics driven framework,we put forward a data aggregation algorithm called equipartition K-means ＋ ＋ based on conventional K-means,and thus,were able to improve the algorithm especially as it pertains to data visualization. The aggregated data obtained by equipartition K-means ＋ ＋ may not only preserve most features of the original data,but also improve the image quality after visualization. Our simulation experiments show that at each value of data abstraction level（ DAL）,equipartition K-means ＋ ＋ get good results,not only in visualization image quality but also quality metrics of histogram difference measure（ HDM） and nearest neighbor measure（ NNM）.

关 键 词：质量度量 数据空间 数据聚合 K-均值 多维数据可视化 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]