I have coordinates of about 13 points from different pictures in UTM. Per point I can have 2, 3, 4 or 5 coordinates of these points from 2, 3, 4 or 5 pictures. I would like to quantify the quality of the set of pictures statistically but I don't know how. Below you find a screenshot from EXCEL to clarify the situation. If anyone knows how I could define some kind of accuracy of these images based on these point locations please let me know.

I thought about creating polygons for each cluster in QGIS (I am using 2.18) and comparing the area and width but that seems so rudimentary to me.

enter image description here

  • k-means clustering?
    – Taras
    Commented May 13, 2020 at 8:50
  • I don't have to cluster the observations anymore, I know which observations correspond to the same points. This is indicated by the groups in the excel table.
    – Jan-Pieter
    Commented May 13, 2020 at 8:52

1 Answer 1


You could simply use the average (or also the RMS) of the distances of each point to the cluster center. The cluster center is computed as the average position of the cluster points.

  • Yeah that was another option but I'm not sure if it will be useful in the context. The problem is I have to quantify the quality without having a "ground truth" so by using the cluster center it seems as if I'm assuming it to be the "ground truth" which isn't the case. Thanks for the suggestion though.
    – Jan-Pieter
    Commented May 13, 2020 at 12:25

Your Answer

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

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