Deriving maps from small to large scale automatically

[Watnuts]

Hello everyone!

I am looking into possibilities of deriving topographic maps from large scales to small scales. I work in an agency that mass produces maps in several scales - 1:2000, 1:10 000 with Microstation and 1:50 000, 1:100 000 and up with ArcGIS. All of the information is stored in databases (CAD objects are converted) so there is a central database for all the information. But as you can probably see, the scales are being produced separately. I am doing a research on whether it is possible to create maps only in one scale, for example 1:10 000 and later automatically derive the information to create a map for smaller scales like 1:25 000 and 1:50 000.

I am just starting this research so I am looking for any similar examples. If anyone has any information, maybe their own experience or researches that have been done on this topic, maybe there are some companies that are doing this already, please share.

Feel free to ask any questions and give your own insight, I will appreciate any kind of information on this topic.

Cheers
Watnuts

[Redrobes]

A lot of aspects of map making is about what information to include and what to discard. As a mainly fantasy mapping site we generally have mostly artistic maps here and so to turn a large scale map into a smaller scale we would just resample the image. When using CAD like ArcGIS then the vectors do not shrink so it becomes a job of saying that these four lines making up a building are now so small that maybe we can drop them. Or it could be that becuase that building is close to many other small buildings you should merge them together and generate some new shape based on the collection of buildings. Its a matter of classification and reimaging which is an arbitrary topic based on style. When your in google maps and you zoom out it keeps some of the placenames and enlarges the font so its still readable but at some point it will drop that name because its a place too small to bother labling. So there is no exact right or wrong thing to perform to do it.

Most mapping agencies have a big database of features and use some app with a programmed style to generate the map at different scales. I didn't think it was the general rule to keep databases per scale or to have one database with features and mark them all up for which scales they would pertain to. The same kind of situation occurs in games programming where some of the characters are 3D models but when the character moves away from the camera you dont want to render the whole 3D model with the same kind of detail that you would if it were close up. Lots of graduated level of detail algorithms exist for that kind of thing. You can use the same algorithms in 2D on the vector shapes - one that springs to mind is the Quadric Edge Collapse. There are others in 3D based on Octrees and Marching Cubes which is 2D would be Quadtrees and Marching Squares. So have a look around for those kinds of things whcih may help.

[waldronate]

Good search terms will include words like "generalization" or "abstraction". For example: https://en.wikipedia.org/wiki/Cartog...generalization