Parallel Scatterplots: Visual Analysis with GPU

We propose an information visualization method, the parallel scatterplots, to analyze the relationships among multidimensional datasets. By combining the methods of parallel coordinates and scatterplots, the relationship join analysis among multi-demensional data sets is effectively carried out via a variety of techniques such as focus+context, multi-perspective multi-view, brush etc. Using the unified shader GPU to render and output particles, line segments and billboards, we have achieved strong depth cue and interactive performance on rendering tens of millions of data. To reduce the visual cluttering on large data, we propose a highly efficient clustering algorithm based on space-filling curves on the GPU. We have integrated joining, clustering and visualization on a GPU-based system which is capable of performing interactive visual analysis on tens of millions of data.