Geometric Abstraction for Effective Visualization and Modeling

Abstract

In this cumulative thesis, I describe geometric abstraction as a strategy to create an integrated visualization system for spatial scientific data. The proposed approach creates a multitude of representations of spatial data in two dominant ways. Along the spatiality axis, it gradually removes spatial details and along the visual detail axis, the features are increasingly aggregated and represented by different visual objects. These representations are then integrated into a conceptual abstraction space that enables users to efficiently change the representation to adjust the abstraction level to a task in mind. To enable the expert to perceive correspondence between these representations, controllable animated transitions are provided. Finally, the abstraction space can record user interactions and provides visual indications to guide the expert towards interesting representations for a particular task and data set. Mental models of the experts play a crucial role in the understanding of the abstract representations and are considered in the design of the visualization system to keep the cognitive load low on the users side. This approach is demonstrated in two distinct fields of placenta research and in silico design of DNA nanostructures. For both fields geometric abstraction facilitates effective visual inspection and modeling. The Adenita toolkit, a software for the design of novel DNA nanostructures, implements the proposed visualization concepts. This toolkit, together with the proposed visualization concepts, is currently deployed to several research groups to help them in nanotechnology research.