基于平行坐标的多视图协同可视分析方法  

Coordinated Visual Analytics Method Based on Multiple Views with Parallel Coordinates

作　　者：陈谊[1] 蔡进峰[1] 石耀斌[2] 陈红倩[1] 

机构地区：[1]北京工商大学计算机与信息工程学院,北京100048 [2]北京特种车辆研究所,北京100072

出　　处：《系统仿真学报》2013年第1期81-86,共6页Journal of System Simulation

基　　金："十二五"国家科技支撑计划项目(2012BAD29B01);北京市属高等学校科学技术与研究生教育创新工程建设项目(PXM2012_014213_000079;PXM2012_014213_000037)

摘　　要：平行坐标和散点图矩阵技术是多维数据可视化和可视分析的主要技术,然而在数据较为杂乱时,这些技术都存在局部信息无法清晰展现的缺陷。结合平行坐标和散点图矩阵等技术,提出了一种简单、快捷的多视图协同可视分析方法。该方法用在平行坐标中嵌入直方图等统计方法来解决局部可视化的缺陷,通过融合不同技术的优点,从不同的角度对多维数据进行可视化和可视分析,从中挖掘出有价值的信息。将该方法应用到农药残留检测数据的分析中,取得了良好效果。Parallel coordinates and scatter-plot matrix are the main visualization and visual analysis techniques for multidimensional data. However, these techniques have defects that local information can not be shown clearly when data set is large and complex. A simple and flexible visual analysis method called multiple coordinated views based on parallel and scatter-plot matrix was proposed. This method combined with the advantages of the parallel coordinates and scatter-plot matrix, and embedded some statistical analysis techniques such as histograms in Parallel Coordinates to compensate those defects. Users could analyze multidimensional data in different perspectives simultaneously, and mine valuable information from the multidimensional data set with this method. The results of application in the pesticide residue detection data set show that this method can implement visual analysis to multidimensional data flexibly and effectively.

关 键 词：多维数据 平行坐标 多视图协同可视分析 农药残留检测数据 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]