Labeling Vario-scale Maps

Abstract

Map labels play an important role in helping users interpret geographic information. Traditional label placement algorithms are typically designed for static or fixed-scale maps, where labels appear at discrete, predefined zoom levels. However, vario-scale maps change content continuously during zooming, requiring labels to also transition smoothly to maintain readability and visual coherence.
This research explores how to place and adjust labels on vario-scale maps, ensuring legibility and usability throughout scale transitions. Due to limited existing work in this area, a set of requirements, both hard and soft, are defined to guide the methodology and evaluate results.
Different strategies are applied based on geometry types. For elongated features such as roads and rivers, label anchor points and orientations are derived from the medial axis (skeleton). For compact geometries such as buildings, the centroid and pole of inaccessibility are used to determine anchor points, while the minimum rotated bounding rectangle guides label rotation.
Two polygonization methods are developed using the tGAP structure: slice-based method extracts geometry at fixed scales and computes label positions per feature type; event-based method reacts to changes in geometry, computing label anchors whenever a change occurs. Label anchor points could be represented as continuous trajectories, supporting smooth interpolation. During interactive visualization, label positions, orientations, and opacities are interpolated across scales. To prevent clutter, real-time collision detection is implemented.
The methodology was applied to the TOP10NL dataset in the Delft region. Results showed that the event-based method allows more precise responsiveness to geometry changes but can produce occasional abrupt label movements. In contrast, the slice-based method offers more stable transitions but may cause unnecessary label shifts even when geometries do not change. Event-based labeling also resulted in a higher number of anchor points. While hard requirements were generally satisfied, some soft requirements were not fully met due to intentional trade-offs or implementation limitations.
Overall, this research introduces a novel framework for integrating dynamic labeling into vario-scale maps using the tGAP structure and demonstrates its feasibility through an interactive, real-time prototype with smooth label transitions and collision handling.