Maps: Zooming should be more beautiful
New programming makes it easier to zoom in on mapsRead out
Road maps and city maps on the Internet could soon be even more beautiful and user-friendly. Computer scientists at the University of Würzburg work with maps whose scales do not change abruptly when moving in and out, but are sliding. The overview is retained and the users can orient themselves better.
A day trip by train should lead to Bamberg. To prepare for the Internet, take a look at the maps of Google or Bing. Where is the railway station in Bamberg, in which direction do you have to walk to the city center, how to get to the cathedral? Deeper and deeper, one zooms into the city map, discovers this and that - and at some point no longer has the station on the screen. Where was he again?
"Anyone who uses the zoom function of maps on the Internet does not get different views of one and the same map, but always other maps, so that the images change abruptly, " explains Alexander Wolff of the Institute of Computer Science of the University of Würzburg. As a result, users easily lose their orientation. Wolff and his colleague Jan-Henrik Haunert want to change that - with sliding-scale maps.
Enlarge individual passages from city maps
A sliding scale - the Würzburg computer scientists are thinking, inter alia, of interactive maps on which individual regions can be enlarged. Example: A tourist wants to take a closer look at the streets leading to the Löwenbrücke on the Internet map of Würzburg. He marks the spot and gets the desired area enlarged. The rest of the map he sees on the screen will continue to be displayed, preserving the overall view of the city center.Alexander Wolff (left) and Jan-Henrik Haunert are working on new algorithms to make interactive maps on the Internet even better. © Robert Emmerich
This type of presentation can be realized on the computer for some time, with the so-called fisheye technique. However, Jan-Henrik Haunert and computer science student Leon Sering have now developed a significantly improved solution. Because so far, the rest of the map is heavily distorted in the fisheye technique, if you zoom out a certain region. In the method of the Würzburg computer scientists, this distortion is much smaller by about 75 percent. In an even better variant, the computer takes only half as long to calculate the zooming process: only the direct environment of the enlarged region is distorted. And here too, the deformation is still 65 percent lower than in the fisheye technique. display
Calculation time must be even shorter
Speaking of computing speed: Several seconds are currently needed, for example, to recalculate the city map of Würzburg with a desired zoom. This may be acceptable for someone who wants to create a card for a printout once. However, this is far too slow for users surfing the Internet through a city map - and, of course, for real-time applications in navigation systems. So far, the technology is not suitable for smartphones either: "Behind it lies an extensive convex programming that is too powerful for smartphones, " says Professor Wolff.
This and more would like to reach the computer scientists in the next three years in their DFG project. You also want to eliminate other effects that confuse the user of interactive maps such as zooming in a city map, the label and the type of presentation abruptly change. That, for example, the station, which is first shown as a dot, suddenly becomes a large structure of railways and buildings.
"We want to design animations that make it possible to make smooth transitions between individual maps and different dimensions, " says Jan Haunert. GeneralThe generalization of the data is the basic problem. Generalization is in this case: For the animation, a detailed map and labeling must be simplified step by step. In the end, the user should get the impression when zooming, steplessly in a single card to slide back and forth instead of jerking from card to card and from scale to scale. The planning of a day trip to Bamberg should then be much more comfortable. (IEEE Transactions on Visualization and Computer Graphics, 2011, DOI 10.1109 / TVCG.2011.191)
(Julius-Maximilians-University W rzburg, 15.02.2012 - NPO)