Structural analysis of spatial data: a control-freak's perspective

From this point, it may be interesting to take a step back to "re-realize" that visualization and graphics in general are probably more about using computational resources to provide the users with insights about their data, than to replace them in decision making. Still, enriching existing approaches with user controls supporting interactive-time feedback remains a challenge in several areas.

Topology-based approaches have shown to be very useful in several visualization and graphics tasks, due to their ability to extract the high- level structure of a phenomenon, hence dividing an overall problem into simpler pieces where solutions can be easily and efficiently derived. In this talk, I will show how controlling the combinatorial and structural properties of a solution can provide important computation insights for the design of time-efficient, yet highly flexible, algorithms, even for initial automatic suggestions. In particular, I will illustrate the benefits of topology analysis in several, surprisingly diverse, real-life problems I have been working on recently with my research collaborators, such as scalar field multi-resolution representation computation, interactive surface quad-meshing or image mosaic composition. Parts of this work were presented at SIGGRAPH and VIS earlier this year.