Today, users can access maps at any time and at any place via the Internet. Services like Google Maps or Bing Maps offer multiple possibilities for interaction. In particular, users can zoom or pan in order to define the area mapped on screen. Depending on the scale (or zoom factor), a map is selected from a small set of maps; the selected map is shown to the user. When zooming in, another map may become selected and displayed, which implies abrupt changes. This often causes users to lose orientation.
In a project funded by the German Research Foundation (DFG) (Ha 5451/3-1; 2012–2015), we develop algorithms for automatic generation of variable-scale maps. A variable-scale map may either be (i) a continuum of maps that covers a certain interval of scale or (ii) maps whose scale varies over the plane. When generating such maps, special quality criteria and constraints have to be taken into account. Algorithms for continuous generalization, dynamic label placement, and map schematization are needed. We will formalize these tasks as optimization problems and, thereby, we will make the quality of variable-scale maps measurable. In order to solve these problems, we will develop exact algorithms and heuristics for real-time applications.
Labeling streets along a route in interactive 3d maps using billboards.
In Proc. 18th AGILE Int. Conf. Geograph. Inform. Sci. (AGILE’15), Lecture Notes in Geoinform. and Cartography, pages 269–287. Springer-Verlag, 2015.
N. Schwartges, B. Morgan, J.-H. Haunert, and A. Wolff.
Labeling streets in interactive maps using embedded labels.
In Proc. 22nd ACM SIGSPATIAL Int. Conf. Advances in Geograph. Inform. Systems (ACM-GIS’14), pages 517–520, 2014.
N. Schwartges, A. Wolff, and J.-H. Haunert.
How to eat a graph: Computing selection sequences for the continuous generalization of road networks.
In Proc. 22nd ACM SIGSPATIAL Int. Conf. Advances in Geograph. Inform. Systems (ACM-GIS’14), pages 243–252, 2014.
M. Chimani, Th.C. van Dijk, and J.-H. Haunert.
Interactive focus maps using least-squares optimization.
International Journal of Geographical Information Science, 28(10):2052–2075, 2014.
Th.C. van Dijk and J.-H. Haunert.
Drawing Road Networks with Focus Regions.
IEEE Transactions on Visualization and Computer Graphics (Proc. Information Visualization 2011), 17(12):2555–2562, 2011.
J.-H. Haunert und L. Sering.
Constrained set-up of the tGAP structure for progressive vector data transfer.
Computers & Geosciences, 35(11):2191-2203, 2009.
J.-H. Haunert, A. Dilo und P. van Oosterom.
[doi] [PDF] [BibTeX]
Algorithms for Interactive Variable-Scale Maps.
DFG Project Report, 2015.
J. Haunert, Th.C. van Dijk, and A. Wolff.
A promising approach to submit a vector map from a server to a mobile client is to send a coarse representation first, which then is incrementally refined. We consider the problem of defining a sequence of such increments for areas of different land-cover classes in a planar partition. In order to submit well-generalised datasets, we propose a method of two stages: First, we create a generalised representation from a detailed dataset, using an optimisation approach that satisfies certain cartographic constraints. Second, we define a sequence of basic merge and simplification operations that transforms the most detailed dataset gradually into the generalised dataset. The obtained sequence of gradual transformations is stored without geometrical redundancy in a structure that builds up on the previously developed tGAP (topological Generalised Area Partitioning) structure. This structure and the algorithm for intermediate levels of detail (LoD) have been implemented in an object-relational database and tested for land-cover data from the official German topographic dataset ATKIS at scale 1:50 000 to the target scale 1:250 000. Results of these tests allow us to conclude that the data at lowest LoD and at intermediate LoDs is well generalised. Applying specialised heuristics the applied optimisation method copes with large datasets; the tGAP structure allows users to efficiently query and retrieve a dataset at a specified LoD. Data are sent progressively from the server to the client: First a coarse representation is sent, which is refined until the requested LoD is reached.