Hi everyone
I am trying out pix4D for the first time as I would like to calculate forestry canopy top levels. We currently use a medium format camera (80mm which included GPS and Roll, Pitch and Yaw data). I have had moderate success with some trial runs done with 25cm GSD imagery and 70/60 forward and side overlaps (moderate success = 0.5m to 5m variations on ground and tree canopy top levels when compared to LiDAR data of the area). I would appreciate it if any of you could point me in the directions of a custom tutorial or advise on settings/workflows to use in order to improve accuracies.
Thanking you in advance.
Hi Rory,
Probably you can take a look on the following publication
Tree height quantification using very high resolution imagery acquired from an unmanned aerial vehicle (UAV) and automatic 3D photo-reconstruction methods
P.J. Zarco-Tejada , R. Diaz-Varela, V. Angileri, P. Loudjani
All the best,
Rory,
This may or may not help you, but I found good canopy results in a 4 mile road project I performed. I setup the GCP’s and overlap for Corridor Mapping as suggested in https://support.pix4d.com/hc/en-us/articles/202557459.
I used 90% + Frontal overlap since the minimum suggested is 85% for a corridor.
After Flying down the center-line, I decided to fly parallel paths over both road ROW’s, giving me 3 flight paths instead of the originally planned 1 path. The Sidelap was 70 - 75%.
Even though I was not interested in the Canopy, Pix4d did a great job modeling the trees. Much better than I normally see in my work.
Of course ground contours matched the LiDAR data I had for a portion of that project, due to using Survey Grade Ground Control Points.
You may want to try 90% Frontal, 80% Side overlap, and enough Ground Control Points as these will substantially help the reconstruction. If you are not satisfied, you will need to fly lower for a better GSD. Since my final output is normally Ortho’s, I generally fly for a GSD of 1-2 cm.
I usually end up collecting way too much data in the field, but I’ve found that’s easier and cheaper than having to go back and re-fly a portion of any project. Hard Drives are cheap and it’s easy to adjust the resolution of the final deliverable. We can always reduce final output resolution in the office, but cant increase it. That’s why I changed my mind in the field and flew 3 flight paths instead of the One Path I had planned for in the referenced 4-mile road project. Now I fly all corridor projects this way.
Please let us know what you determine to be your best workflow. This is how we all learn, from each other.
Good Luck,
Ryan Fontaine, P.E.
Hi Ryan
Did you use nadir or oblique images or both? At wich altitude was the fly?
Thank’you and best reguards.
Stefano Romani
Stefano,
I use nadir for corridor projects flying a DJI Inspire @ 30-40 meter altitude.
Hi Ryan,
Are you able to determine the ground level below trees canopy?
Many thanks in advance!
Carlos,
Only if the ground is photographed through openings in the canopy, which in my case - the road obviously is.
I’m not normally interested in the canopy, I was just sharing my experience with good canopy reconstruction resulting from the High Overlaps and GCP’s I used for flying a roadway using the recommended corridor techniques from Pix4D.
If someone needed to survey large areas of dense canopy, the Ebee (RTK version) would be a great choice.
Many thanks Ryan.
How many GCPs do you normally take for a 2km highway reconstruction?
Carlos,
At least 10.
See :
https://support.pix4d.com/hc/en-us/articles/202559299
I will normally survey more GCP’s and use some of them as check points (not used as GCP’s in the Processing)
Yes but they talk about 8-10 GCPs without relation to corridor length.
Is the same 8-10 GCPs for 2km or 10km?