Research On The Anti-aliased Method Of Density Distortion And Ghost Clusters In Parallel Coordinates

The rapid development of data communication technology using the Internet and World Wide Web has led to the existence of a large number of shared information online.In addition,enterprises and government organizations create large amount of data that contain both structured and unstructured information during operation,which needs to be processed,analyzed and visualized.Therefore,with the increasing application of demographic computation and digital libraries,multi-dimensional data analysis becomes more and more common.Parallel coordinate plots is a well-researched and widely used technology to visualize multidimensional data.Dimensions are represented by vertical and equidistant axes,and data records are encoded by polylines,which connect the corresponding attribute values on each axis.Parallel coordinates have been applied in various fields,such as finance,traffic safety,and network analysis.However,it is also affected by the visual clutter problem as well as computational complexity problem,and many methods have been developed to solve these problems.However,there is no empirical study of how the slope-dependent rendering improves the recognition of cluster.To solve the above problems,an anti-aliasing method for density distortion and ghost clusters of parallel coordinates were included in the thesis.In this thesis,it is further studied and improved on the basis of the existing formal description of the two image distortion phenomena of parallel coordinates.Firstly,different visual structures are studied in the based-pixel screen space,and the data values are mapped to pixel coordinates and get the formal description of the quality of the rendered image,including overplotting and other metrics of parallel coordinates.Then,the measurement matrix is constructed by the measurement index of parallel coordinates rendered by different axes.Finally,the best axis pair sequence is used to reorder the dimensions of the parallel coordinate plot,and the improved parallel coordinate plot is obtained by rendering the broken line based on the slope.The visualization effect of different clutter reduction methods on the same data set were showed,which showed the effectiveness of this visualization method.Experimental results show that this method can effectively improve the user's recognition of clusters in the data when the parallel coordinate plot is rendered based on the slope.