Forum

Author Topic: Dense cloud artifact  (Read 512 times)

mhnewton

  • Newbie
  • *
  • Posts: 11
    • View Profile
Dense cloud artifact
« on: January 09, 2024, 02:41:08 AM »
Hey, I have been having issues with artifacts showing up in my dense cloud models for erosion modelling. It appears to be an alignment issue, but I am not sure which parameters I need to change in order to fix it?? I currently use highest setting to align the cameras, remove all cameras with quality <0.7, add the GCPs and create dense cloud on ultra high with mild depth filtering. Below is a screenshot of one of the artifacts as well as part of the processing report with the parameters used.

Any help is greatly appreciated
Cheers!

JMR

  • Hero Member
  • *****
  • Posts: 502
    • View Profile
Re: Dense cloud artifact
« Reply #1 on: January 09, 2024, 08:07:58 PM »
I can not see whatever you call an artifact, can you elaborate a bit more? (add a close-up, please)
The first question is why dense cloud and not a mesh directly after alignment?
Have you refined the alignment by any recursive cleanup method?

Regards

José Martínez
Accupixel Ltd

mhnewton

  • Newbie
  • *
  • Posts: 11
    • View Profile
Re: Dense cloud artifact
« Reply #2 on: January 09, 2024, 09:27:13 PM »
Hi José

Maybe "artifact" is the wrong term to use... I tried to show a close up in the screenshots below, but essentially when the dense cloud is built, it appears like one of the cameras is tilted, like a tile being lifted slightly above the rest of the point cloud to create this noise in the DEM when constructed. You can see in the Pre-DEM (flight before erosion) that it is a continuous surface, while in the Post-DEM (flight after erosion) there is the distinguished lifted corner that is created from some of the points being slightly above the rest of the dense cloud. It is not lifted by much, only a few centimeters at the most, but is enough to throw off erosion calculations when comparing the 2 DEMs. Classifying the dense cloud by point confidence, you can see these lifted regions have very low confidence (red=low, blue=high)... It might be as simple as just removing the low confidence points from the dense cloud, but I'm curious if there is something else that was causing the low confidence to begin with

As for why I didn't use a mesh, I am creating the DEM from the dense cloud directly so that it is created by IDW interpolation and retains as much resolution as possible to be able to run the erosion calculations.