I am working on automated land parcel identification and would like to extract road/river/thin and linear-like polygons from a massive dataset. For now, the only index I can get some answers from is the ratio between the perimeter and the area of the polygons. The results are Ok, but I need to "clean" them.

How do I find the shape index, compactness or assymetry with PostgreSQL/PostGIS? I need a solution for vector data only.

  • 1
    GIS SE uses a "Focused question / Best answer" model. This question seeks to start a discussion and compile a list of options. Such a request would be appropriate in the Geographic Information Systems Chat feature, but not here in the main site. – Vince Sep 25 '17 at 21:58
  • 1
    Well roads and rivers have colours too, so if you have that information, it would help. From my experience working with land parcels, there is no typical land shape type. You are likely to get much better results if you can add in additional remote sensed data, as land parcels have a tendency to change over the year as things grow/die. However, as it stands the question is probably too vague for this format, and many Postgis users will not have used eCognition. – John Powell Sep 25 '17 at 22:54
  • Hi John, thanks for the reply. Yes, remote sensing data could help in for such problems, but rasters are not suitable for this project (talking about multimillion data and processing times + storage). – mazinga Sep 26 '17 at 5:22
  • Hi Mazinga, multi million isn't really that big these days, and even if it were billions, there are cloud solutions for this kind of analysis. If you want to automatically extract land parcels from millions of objects with a high degree of accuracy then you should use as much information as possible. If you are prepared to accept a lower accuracy, please edit the question to make this clear. Have you thought of using something like Google Earth Engine to help with the analysis? – John Powell Sep 26 '17 at 19:21
  • 1
    I want to extract road/river/thin and linear-like polygons and clean the results after extracting those. There is no need to extract all of the mentioned polygons. I already cleaned around 90% of them and would like to push it a little bit further with involving the mentioned parameters. So the question remains: shape index, compactness or asymmetry for vector data in Pg/PostGIS . The accuracy and the environment are part of other topics that would make this question too broad again. No other data must be involved. I process it via a fast server. – mazinga Sep 26 '17 at 19:36

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.