Research And Improvements On The Methods Of Multi-dimensional Uncertainty Visualization

In the real world,the level of credibility and accuracy that people make decisions and conclusions from the data obtained in analysis,is mainly depended on quality of the analyzed data.However,real-world data is known to be imperfect,suffering from various forms of uncertainty such as statistical variation or spread,estimation errors,minimum-maximum range values,or missing data in data entry.Analysis and making decision without considering variable uncertainty can lead to inaccurate or even incorrect results.Especially,with the development of science,massive,complicated,heterogeneous and multidimensional data are required to analysis.It becomes increasingly important to discover this kind of potential uncertainty.By reading a lot of foreign literatures,exploration of uncertainty is known as an active research field and an ongoing unresolved problem,including definition,measurement methods and how to visualize it.For this reason,under the condition of no identified topology about uncertainty,the conclusions of uncertainty definition,five categories of how to measure uncertainty are presented.Besides,the basic process of uncertainty visualization,mapping methods and three theoretical evaluations of uncertainty visualization are put forward.Secondly,existing uncertainty visualization techniques are mainly developed for the fields of cartography,flow or volume data.Prior work has primarily focused on spatio-temporal or univariate data.Little research has been reported on conveying data quality information within multivariate visualization techniques.Therefore,the main contribution in this literature focuses on exploring how to utilize and modify parallel coordinates,scatterplot and star glyph to make users aware of quality of their data by conveying not only the actual data,but also its corresponding uncertainty attributes.Three uncertainty visualization methods are presented: the modified parallel coordinate,the parallel scatterplot,and the integration of parallel coordinate and star glyph.At the same time,three mentioned theoretical analysis are utilized to evaluate them and their advantage and disadvantage are also analyzed.Eventually,MFC and OPENGL technologies are combined to construct an experimental platform to validate our proposed new ideas.Meanwhile,Brushing,Hiding,Zooming in/out,Linking,Hiding five HCI technologies are imposed into the system to help users to focus on what they find interesting.