Forum

Author Topic: point cloud variance  (Read 6184 times)

wim van calster

  • Guest
point cloud variance
« on: April 24, 2019, 11:15:28 AM »
Could anybody tell me how i can use the pointcloud variation tool?

Alexey Pasumansky

  • Agisoft Technical Support
  • Hero Member
  • *****
  • Posts: 14816
    • View Profile
Re: point cloud variance
« Reply #1 on: April 24, 2019, 01:57:28 PM »
Hello wim van calser,

Tie points variance values are used only for statistical assessment and wouldn't have any effect on the result.

Variance could be visualized in the Model view for the tie points in a form of vectors.
Best regards,
Alexey Pasumansky,
Agisoft LLC

wizprod

  • Jr. Member
  • **
  • Posts: 84
    • View Profile
    • Dronographica
Re: point cloud variance
« Reply #2 on: November 24, 2019, 12:44:47 PM »
Hello wim van calser,

Tie points variance values are used only for statistical assessment and wouldn't have any effect on the result.

Variance could be visualized in the Model view for the tie points in a form of vectors.

What can be determined from the view?
What units is this view displayed in ?
How to get a more uniform distribution?

Dragline

  • Newbie
  • *
  • Posts: 34
    • View Profile
Re: point cloud variance
« Reply #3 on: January 29, 2020, 04:26:08 AM »
If I understand correctly, the question being asked by the first poster is how to get the point cloud variance to be visible when looking at the sparse cloud.



How do I get this feature to work? I've built both a sparse and dense cloud, however neither allowed for this.

What checkbox am I missing?

Mak11

  • Sr. Member
  • ****
  • Posts: 374
    • View Profile
Re: point cloud variance
« Reply #4 on: January 29, 2020, 05:10:44 AM »
Dragline,

You have to run Optimize Camera Alignement once after your initial camera alignement while selecting Estimate tie point covariance.



Mak

afischbach

  • Newbie
  • *
  • Posts: 4
    • View Profile
Re: point cloud variance
« Reply #5 on: March 04, 2020, 01:49:38 AM »
I would like to use the point cloud variance to assess orthoimage spatial distortion.  I am counting objects on the ground across our study area using a 10m x 10m grid laid over the survey area and geospatially clipped to the outline of the objects.  I would like to utilize the aerial uncertainty of each grid cell to better describe the uncertainty in counting these objects across the study area.  See Battallie et al 2017 for the method (DOI 10.1007/s00300-016-2060-5).
Please advise on how the point cloud variance is estimated.  Is there a publication that may be referenced?
Please advise if this point cloud variance may be used to asses the spatial registration of the model and it's orthoimage product.
Please advise on how to export this point cloud variance for import to a GIS.  When exporting the point cloud to XYZ text file, could this information be retained and exported?

Benny

  • Newbie
  • *
  • Posts: 13
    • View Profile
Re: point cloud variance
« Reply #6 on: March 13, 2020, 07:31:39 PM »
Hi,

Does anyone know what is the unit of the point cloud variance? When displayed, there is a colorbar with the min-max values, but no unit.

Is it in metre by default? Is it related to the unit of the coordinate system?

Thank you in advance!

BenoƮt

specyfick

  • Newbie
  • *
  • Posts: 5
    • View Profile
Re: point cloud variance
« Reply #7 on: January 27, 2021, 02:44:26 PM »
@Alexey Pasumansky, what is the mathematics beyond "estimate tie points covariance"
How the sigmas are calculated ?
Can you provide reference or description ?

Joe Adams

  • Newbie
  • *
  • Posts: 8
  • UAS for Science
    • View Profile
    • USGS NUSO
Re: point cloud variance
« Reply #8 on: January 27, 2021, 07:39:20 PM »
I will give this a bump also.

We very much would like the option to export the Point Cloud Variance values to a raster. This was also asked for in this post:

https://www.agisoft.com/forum/index.php?topic=11255.msg50541#msg50541

Is it possible to access and save this information via the Python API? I don't know this well at all but I did find:

class Metashape.PointCloud     (Sparse point cloud instance)
   class PointCloud.Point
      cov
          Point coordinates covariance matrix.
            Type matrix

Very helpful information we would like to include in our reporting.

Thank you.
Joe Adams
UAS Operator
USGS National Uncrewed Systems Office
Denver, Colorado

Joe Adams

  • Newbie
  • *
  • Posts: 8
  • UAS for Science
    • View Profile
    • USGS NUSO
Re: point cloud variance
« Reply #9 on: January 27, 2021, 07:47:06 PM »
My Bad. I found Alexey already posted a code snippet which includes extracting the covariance value from a point:

https://www.agisoft.com/forum/index.php?topic=11218.msg50653#msg50653

Thank you Alexey.
Joe Adams
UAS Operator
USGS National Uncrewed Systems Office
Denver, Colorado