Visual Boosting in Pixel-based Visualizations

Abstract

Pixel-based visualizations have become popular, because they are capable of displaying large amounts of data and at the same time provide many details. However, pixel-based visualizations are only effective if the data set is not sparse and the data distribution not random. Single pixels - no matter if they are in an empty area or in the middle of a large area of differently colored pixels - are perceptually difficult to discern and may therefore easily be missed. Furthermore, trends and interesting passages may be camouflaged in the sea of details. In this paper we compare different approaches for visual boosting in pixel-based visualizations. Several boosting techniques such as halos, background coloring, distortion, and hatching are discussed and assessed with respect to their effectiveness in boosting single pixels, trends, and interesting passages. Application examples from three different domains (document analysis, genome analysis, and geospatial analysis) show the general applicability of the techniques and the derived guidelines.