ABSTRACT
Approaches for the estimation of the most informative viewpoints for scientific datasets are similar to those developed for polygonal data, with differences due to the different focus of scientific visualization with respect to more general computer graphics. One main difference from polygonal computer graphics is that the underlying data is in general more complex, as scientific data are generated by measurements and simulations that have a very heterogeneous output. Some data contain solely measurements in numerical values without a priori knowledge about structures, while other data contain information about the most relevant structures such as critical points in flow data or segmentation masks of anatomical objects. The heterogeneity among scientific data types corresponds to the heterogeneity in the visualization approaches for viewpoint quality evaluation.
