Issue No. 05 - September-October (2006 vol. 12)
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/TVCG.2006.170
Focus+context visualization integrates a visually accentuated representation of selected data items in focus (more details, more opacity, etc.) with a visually deemphasized representation of the rest of the data, i.e., the context. The role of context visualization is to provide an overview of the data for improved user orientation and improved navigation. A good overview comprises the representation of both outliers and trends. Up to now, however, context visualization not really treated outliers sufficiently. In this paper we present a new approach to focus+context visualization in parallel coordinates which is truthful to outliers in the sense that small-scale features are detected before visualization and then treated specially during context visualization. Generally, we present a solution which enables context visualization at several levels of abstraction, both for the representation of outliers and trends. We introduce outlier detection and context generation to parallel coordinates on the basis of a binned data representation. This leads to an output-oriented visualization approach which means that only those parts of the visualization process are executed which actually affect the final rendering. Accordingly, the performance of this solution is much more dependent on the visualization size than on the data size which makes it especially interesting for large datasets. Previous approaches are outperformed, the new solution was successfully applied to datasets with up to 3 million data records and up to 50 dimensions
data structures, data visualisation, feature extraction, rendering (computer graphics),outlier detection, context visualization, parallel coordinate, focus visualization, data representation, small-scale feature detection, data abstraction, output-oriented visualization approach, rendering technique,Data visualization, Humans, Navigation, Computer vision, Visual system, Information processing, Concurrent computing, Jamming, Multidimensional systems, Costs,Parallel coordinates, focus+context visualization, outliers & trends, large data visualization.
"Outlier-Preserving Focus+Context Visualization in Parallel Coordinates", IEEE Transactions on Visualization & Computer Graphics, vol. 12, no. , pp. 893-900, September-October 2006, doi:10.1109/TVCG.2006.170