7

I have a point pattern (see below) in which I assume to find "significant" lines of points. However, I did not find any method that is integrated in a GIS or in R that searches for such line pattern in a spatial point pattern. There is one method that is implemented in the free software called PAST (http://folk.uio.no/ohammer/past) called point alignment.

the method is described in :

Hammer, Ø. (2009). New statistical methods for detecting point alignments. Computers & Geosciences, 35(3), 659-666.

beyond this method I am not aware of any tool that finds lines in point patterns. Any hints on possible tools, r-packages, etc. are most welcome.

Spatial pattern of individual plants

Improve this question
2
  • can you expand on what you consider to be significant? just colinear?
    – Ian Turton ♦
    Sep 3, 2013 at 8:44
  • good question. I am not sure how to describe "significant" here. "colinear" is a good word though. I mean I "see" many lines but are these points really lined up or not? and if so, how many lines are there? as Oyvind stated below, it depends on the search radius.
    – Jens
    Sep 3, 2013 at 17:55

2 Answers 2

Reset to default
7

This paper: Christophe & Ruas 2002, Detecting Building Alignments for Generalisation Purposes, ISPRS, Ottawa. describes an operational method for the detection of small surfaces (buildings) alignements - it should work even better with points! (this method is rather robust since it is used for the production of 1:25000 maps in France).

Improve this answer
1
  • dear julien, great I will try to implement this approach in a r-function. look's interesting.
    – Jens
    Sep 3, 2013 at 17:37
6

I'm the author of Past and that paper you refer to. From your plot it seems like you have used a too large search radius. Try to decrease the Radius value. Also, if you would like to detect more lines, increase the alpha value (significance level).

Improve this answer
1
  • Oh, wow, that´s a surprise. :) In the plot I just draw a line by hand in order to make my message more clear. increasing alpha indeed gives more lines and I found many patterns. however, when do I know that I found the optimal solution? and should alpha be more like 0,05 or 0,1? one more problem is that lines appear/disappear when switching the radius from small to large. should I just notify all significant lines at all radii? one last question: what´s next after lines are identified? display them in a GIS? thanks for your comment and PAST !
    – Jens
    Sep 3, 2013 at 17:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.