Computing Curved Area Labels in Near-Real Time
This work addresses automated label placement in digital cartography, offering an incremental improvement over prior methods by broadening the search space and enhancing efficiency.
The paper tackles the Area Labeling Problem by proposing a method that uses a pruned skeleton to efficiently compute label placements for geographic areas, achieving near-real-time performance.
In the Area Labeling Problem one is after placing the label of a geographic area. Given the outer boundary of the area and an optional set of holes. The goal is to find a label position such that the label spans the area and is conform to its shape. The most recent research in this field from Barrault in 2001 proposes an algorithm to compute label placements based on curved support lines. His solution has some drawbacks as he is evaluating many very similar solutions. Furthermore he needs to restrict the search space due to performance issues and therefore might miss interesting solutions. We propose a solution that evaluates the search space more broadly and much more efficient. To achieve this we compute a skeleton of the polygon. The skeleton is pruned such that edges close to the boundary polygon are removed. In the so pruned skeleton we choose a set of candidate paths to be longest distinct subpaths of the graph. Based on these candidates the label support lines are computed and the label positions evaluated. Keywords: Area lettering \and Automated label placement \and Digital cartography \and Geographic information sciences \and Geometric Optimization.