Forum

Author Topic: Difficulty creating planar orthomosaic from complex surface  (Read 5984 times)

haglore

  • Newbie
  • *
  • Posts: 3
    • View Profile
Difficulty creating planar orthomosaic from complex surface
« on: September 12, 2017, 02:06:26 AM »
Hi, I am Al and I'm new here. I use photoscan a lot to create 3D models of rock outcrops (cliffs).

The surfaces of my cliffs are very rough, but I want to extract scaled orthophotos from my models so I can map fracture traces across the cliff faces.

I can generate great point clouds and texture maps with my photo coverage, but the orthomosaic created from the texture seems to only reconstruct part of the scene. You can see from the current view point the whole model is visible, but only a fraction is reconstructed in the orthomosaic (attached images).

I wonder if this has to do with the roughness of my surface and collapsing the model into a 2D planar orthomosaic? But I am not sure.

Any thoughts? This works fine with some of my other cliff models, but others, I have this issue. 

SAV

  • Hero Member
  • *****
  • Posts: 710
    • View Profile
Re: Difficulty creating planar orthomosaic from complex surface
« Reply #1 on: September 12, 2017, 08:29:50 AM »
Hi haglore,

If you want to map fracture traces of complex rock outcrops, I suggest this workflow:

1. Export the dense point cloud of your PhotoScan model (las or laz format)
2. Install CloudCompare (free and open-source) and open the point cloud
3. Trace your fractures using the COMPASS plugin (Plugins > Compass). It also computes an orientation measurement (strike and dip) for each trace. These orientation data can then be exported as a CSV file for further analysis (e.g., rose diagrams, stereonet plots, etc).

Here is a detailed description of the COMPASS plugin:

Trace Tool: Digitise and measure traces and contacts
The trace tool allows rapid computer-assisted digitisation of fracture traces and lithological contacts. It uses a least-cost path algorithm to 'follow' fracture traces and contacts between user defined start and end points. To use, select the tool and then left click the start and end of a fracture you would like to digitise/measure. The tool will then try to find a fracture trace that links these points. Generally this works remarkably well, though if the desired result is not achieved, intermediate 'waypoints' can be added (also using left click) to modify the trace. To finish a segment click 'Accept' (green tick) or press the Space Key. To cancel a segment, click 'Close' or press the Escape key


Let me know if you need help.

Note: There is also an 'equivalent' plugin for QGIS that can be used to trace fractures on raster datasets (i.e., orthophotographs), called GeoTrace.

Regards
SAV

This is the official publication for the Compass and GeoTrace plugins:
Thiele, ST., Grose, L., Samsu, A., Micklethwaite, S., Vollgger, SA. & Cruden, S., 2017, 'A computer-assisted approach to structural interpretation of point clouds and rasters', Journal TBA

Jeremiah_ROWE

  • Guest
Re: Difficulty creating planar orthomosaic from complex surface
« Reply #2 on: September 12, 2017, 04:57:29 PM »
I didn't even know such a thing existed! Thanks for sharing!

haglore

  • Newbie
  • *
  • Posts: 3
    • View Profile
Re: Difficulty creating planar orthomosaic from complex surface
« Reply #3 on: September 12, 2017, 06:26:38 PM »
Thanks for the response SAV. This looks like an excellent approach for some of the applications that I am trying to do and I am familiar with some of CloudCompare. 

But, the dense point cloud has less detail than the texture map (because the texture map wraps pixels from the original photos).  Can I use the same workflow with the texture model? I believe I have loaded these into CloudCompare before too, but I know the files are much larger typically.

If so this would be huge.  Thanks again!

-Haglore

SAV

  • Hero Member
  • *****
  • Posts: 710
    • View Profile
Re: Difficulty creating planar orthomosaic from complex surface
« Reply #4 on: September 13, 2017, 04:09:22 AM »
Hi Haglore,

The CC Compass plugin only works on point clouds. But you could also load the mesh of the same model 'in the background' in order to help you to visually identify fractures on the dense point cloud. That's how I normally do it.

Alternatively, you can manually trace a fracture on your mesh using CloudCompare's Trace Polyline tool. Here is the description: http://www.cloudcompare.org/doc/wiki/index.php?title=Trace_polyline

Regards,
SAV



SAV

  • Hero Member
  • *****
  • Posts: 710
    • View Profile
Re: Difficulty creating planar orthomosaic from complex surface
« Reply #5 on: September 13, 2017, 04:13:33 AM »
It is brand new; only available since a few weeks. The associated paper has been submitted but is not published yet.

The Compass and GeoTrace plugins both allow one to quickly and consistently trace fractures (and other features, as will be shown in the paper).

Regards,
SAV


I didn't even know such a thing existed! Thanks for sharing!

clemson_geospatial

  • Newbie
  • *
  • Posts: 6
    • View Profile
Re: Difficulty creating planar orthomosaic from complex surface
« Reply #6 on: September 13, 2017, 09:01:29 PM »
It is brand new; only available since a few weeks. The associated paper has been submitted but is not published yet.

The Compass and GeoTrace plugins both allow one to quickly and consistently trace fractures (and other features, as will be shown in the paper).

Regards,
SAV


I didn't even know such a thing existed! Thanks for sharing!

This is really interesting, thank you for sharing. Can the GeoTrace plugin operate on any raster (i.e. a DEM rather than an orthophoto)? I'm working with a team who want to digitize rice fields in coastal South Carolina and use linear features in DEMs derived from LiDAR (remnant dikes are the linear features). Might GeoTrace work for that purpose?

SAV

  • Hero Member
  • *****
  • Posts: 710
    • View Profile
Re: Difficulty creating planar orthomosaic from complex surface
« Reply #7 on: September 14, 2017, 05:50:55 AM »
Hi clemson_geospatial,

The GeoTrace plugin works on ANY raster.  :)

You just have to choose/use the right 'cost layer', which can be computed using the GeoTrace plugin. There are quite a few options, such us Darkness, Sobel, Prewitt Transform, .....

But you can also create your own cost layer (for example, based on slope or curvature of your DEM)  which is then used by the GeoTrace plugin.
The GeoTrace plugin simply needs a single-band (grey) raster which it can use to intelligently 'grow a path' along the darkest pixels (minimum values). Note that it is designed to follow 'narrow' objects such as fractures or ridges.

For example, we have used the GeoTrace plugin to trace linear features on geophysical raster datasets. Worked really well.

Regards,
SAV

It is brand new; only available since a few weeks. The associated paper has been submitted but is not published yet.

The Compass and GeoTrace plugins both allow one to quickly and consistently trace fractures (and other features, as will be shown in the paper).

Regards,
SAV


I didn't even know such a thing existed! Thanks for sharing!

This is really interesting, thank you for sharing. Can the GeoTrace plugin operate on any raster (i.e. a DEM rather than an orthophoto)? I'm working with a team who want to digitize rice fields in coastal South Carolina and use linear features in DEMs derived from LiDAR (remnant dikes are the linear features). Might GeoTrace work for that purpose?
« Last Edit: September 14, 2017, 07:34:02 AM by SAV »

cadm8

  • Jr. Member
  • **
  • Posts: 74
    • View Profile
Re: Difficulty creating planar orthomosaic from complex surface
« Reply #8 on: September 17, 2017, 03:19:02 PM »
SAV, you blew us all away with that one :o

SAV

  • Hero Member
  • *****
  • Posts: 710
    • View Profile
Re: Difficulty creating planar orthomosaic from complex surface
« Reply #9 on: September 18, 2017, 09:17:03 AM »
Haha.  ::)
These plugins are magic  ;)
Should speed up some of the tedious digitisation.

Regards,
SAV

SAV, you blew us all away with that one :o