Forum

Author Topic: Aligning two overlapping sets of pictures (urban) taken one year apart  (Read 1129 times)

lemieszek

  • Newbie
  • *
  • Posts: 12
    • View Profile
Hi all!

I'm studying the application of Metashape for urban studies and I'm trying to compare different properties of two sets of pictures of the same urban area taken 1 year apart from each other.

I've processed both separately using the same GCPs (I have maybe 30 of them but it's a huge area - 600 hectares). They mostly coincide to a great degree. If I measure the difference between DEMs I can see buildings that have been demolished or constructed and even how much the trees have grown or have been cut down.

But there are tiny differences that are a little bit annoying. If a given building (say, 20m high) hasn't changed but it's dislocated by a few centimeters from one set to the other, then the DEM difference will tell me there's a slice of height gain that's 20m high on one side and an equal height loss on the other side.

So my question is: is there a best practice regarding this kind of alignment process? Since maybe 95% of the area hasn't changed at all, one would think it's possible to align both of them perfectly, with maybe ever so slight variations on the remaining 5%, which wouldn't matter since that bit has changed anyway.

Thanks in advance!
« Last Edit: March 28, 2021, 09:27:43 PM by lemieszek »