# Seeking Generalisation Strategies for Building Outlines in PostGIS?

I have a dataset of building footprints vectorised from a raster dataset. The dataset is currently just a vector representation of the raster data. I would like a polygon dataset that only has vertices at the corners of the building. I have attempted to use the Simplify command in PostGIS with limited success.

Is there a better way?

There are many algorithms dedicated to building simplification. You may have a look at this website for an overview. See also this question. You could use:

• Building outline simplification: It consist in recursively deleting too short edges of the outline and lengthening the neighbor edges. Example:

• Smallest surrounding rectangle algorithm: It consist in replacing the building geometry by the smallest surrounding rectangle. This rectangle can also be scaled to the building initial area. Example:

• Squaring algorithm: It consists in applying a light rotation to edges to make the corners perfectly squared. Example:

These transformations may be enough for your need. It can be interesting to combine them.

I am almost 100% sure none of these algorithms are implemented in PostGIS. Some are implemented in the Opencarto java library I am developing. Let me know if you want to test it!

• Can you give a short explanation of what algorithms you refer to in Opencarto for the mentioned "Building outline simplification". I couldn't find direct info in the wiki or code in Opencarto. Sep 9, 2016 at 8:51
• It is "ShortEdgesDeletion". You can find it on github there: github.com/jgaffuri/OpenCarto/blob/master/src/main/java/org/… Sep 12, 2016 at 7:48
• Jan 24, 2018 at 13:52
``````geometry ST_SimplifyPreserveTopology(geometry geomA, float tolerance);
``````

Will avoid creating derived geometries (polygons in particular) that are invalid http://www.postgis.org/documentation/manual-1.5SVN/ST_SimplifyPreserveTopology.html

should be a better output than Simplify

check your geometry first and after with `ST_IsSimple(geometry geomA);`

Something like this might do the trick:

"If we buffer by a large amount, then reverse the buffer by the same amount, we’ll get something that has a similar shape to the original, but without the crinkly bits."

http://blog.opengeo.org/2010/11/22/removing-complexities/

• that's an interesting approach and it seems relatively effective. Thanks for the pointer, it's something I'll be able to use. Your answer would be better if it contained a simple phrase containing the core idea being referenced and wasn't just a link. Jan 12, 2011 at 0:51