Dr. Wouter Meulemans (giCentre, City University, London)

Presentation slides

Schematic maps reduce details to a minimal level, while still supporting the main purpose of the map. This is in stark contrast to traditional generalization for topographic maps, that show detail at a maximal level, constrained by legibility. Furthermore, distortion and stylization are often applied to clarify structures in the information and communicate the schematic nature of the map.

A central question for computing such maps is what is means for a schematic shape to “resemble” the geographic one. We review prominent similarity measures, assessing their suitability as an optimization criterion for the schematization process. Based on this, we present a heuristic algorithm, guided by a similarity measure, that can effectively and efficiently compute schematic maps following the angular-restriction criterion as pioneered by Beck’s map of the London underground. Inspired by both traditional schematic maps and recent trends, we also show how to compute schematic maps that restrict geometry to circular arcs, followed by a brief comparison between straight-line and circular-arc schematization.

We’ll then take a step back and look at a more fundamental question. If we discard the constraints arising from geography, we are left with just a combinatorial graph as input. The question that arises is how many line segments we need, to draw a given graph. We introduce and review three algorithms for planar cubic 3-connected graphs, and show the results of an experiment comparing these methods.

Bio

After a PhD in computational geometry from TU Eindhoven and a postdoc at WWU Münster, Wouter Meulemans is now a Marie-Curie research fellow at the giCentre, City University London. His research interests lie in developing geometric algorithms and applying these to problems arising in information visualization, automated cartography and GIS.