Forum

Author Topic: Using truncated coordinates causing strange errors... Any help appreciated  (Read 2011 times)

SuperNewb

  • Newbie
  • *
  • Posts: 42
    • View Profile
Hi Photoscan community,

I am hoping to get help in wrapping my head around a problem I am experiencing.
I was given truncated coords for my GCPs, I converted these into standard UTM to assist in pinning the GCPs within Photoscan.

After pinning the targets, instead of importing the truncated target coords back in, I optimized the sparse point surface based on my converted UTM coords...  This is where I am lost.  The optimized surface based off the UTM coords has an average error of 2cm.  When I import the truncated coords back in, the error sits around 30cm...

I screwed up further by going ahead and building the dense point surface before importing the truncated coords and finding out about the error... Worse yet, I have spent a day filtering this Dense point surface that took 2 days to process.

I am assuming that I will need to go back and optimize the sparse point surface based on the truncated coords to correct this then rebuild the dense point surface and then start filtering again.

My question to the community is,  Where does this error come from?  when going back and forth between truncated and non-truncated coords, isn't the model just shifting in x/y space?  the elevations are unchanged between the two coord systems...  Photoscan is telling me that the error is in the elevation values however...  Why is error being introduced in elevation from a simple x/y shift?

If anyone can help me wrap my head around this it would be greatly appreciated.

Thanks


Outis79

  • Newbie
  • *
  • Posts: 18
    • View Profile
Hello,
I don't know if that can help you...
When I truncate coordinates I check whether the digits truncated are the same for all targets...
For example  I have 2 targets with est coordinates:
X1= 536700 and X2=536755.
I can truncate the first 4 digits because they are all the same. the points are within a range of 100 meters

Let's say I have a 3rd target with X3=536800.
The three points are within  range of 1000 meters.
 I can truncate only the first 3 digits otherwise X1 and X3 would have the same east coordinate.

Same thing is valid for north and z coordinates.
I hope this can help you.

 

SuperNewb

  • Newbie
  • *
  • Posts: 42
    • View Profile
My method for truncating the coords is to use excel. subtracting the same digit across all coords in a list.
Using 5820000 and 590000 for northing and easting respectively.

5825184.935   593030.281   933.993  would turn into 5184.935   3030.281       933.993

I think I know what you are saying though, you cant simply remove the first few digits to properly truncate.

I am still at a loss as to where the z error comes from in a simple xy shift though...

Alexey Pasumansky

  • Agisoft Technical Support
  • Hero Member
  • *****
  • Posts: 14927
    • View Profile
Hello SuperNewb,

Usually coordinate system have area of use, so if you are loading the coordinates that are millions of meters off the area of use, it will likely cause some issues.

If you are using such truncated coordinates approach, then I can suggest to switch to Local Coordinates system.
Best regards,
Alexey Pasumansky,
Agisoft LLC