I’m looking for your help regarding best practices for measuring the height of recent, therefore, small, vegetation (about 70 cm, to 1 meter).
We have tried using a fixed wing drone, flying at 120m altitude, camera Sony RX1RXII, GSD of 1.55 cm/px, and regarding overlap: Side 82% and Forward 78%.
These parameters were not enough to have a representation of small plants in the cloud point.
We could easily calculate height of trees and bigger objects, about 10cms wide but not at all the small plants.
We tried cross flight and single row. Neither worked.
The lowest we can go is 80m, without upsetting the camera trigger.
Do you have any suggestions? Using a multi-rotor that can go lower, maybe, or a different camera?
The central problem here is going to come down to your ground sampling distance (GSD). It’s important to keep in mind that the accuracies of your project will be 1-3 times your GSD. Let’s say you have a GSD of 5cm. The accuracy could be off by as much as 15cm. If your plant is 70cm then you are looking at over a 20% discrepancy in plant heights.
In my personal opinion using a fixed wing to map a small object isn’t practical. They are designed for very large projects in the hundreds of acres. If you have access to a multi-rotor that can fly lower and thus keep your GSD smaller then you will have much better results. Along these lines, if you have access to a higher resolution camera then it will also give you a smaller GSD.
Another thing to keep in mind is that vegetation is inherently difficult model. It has a very complex structure that tends to move in the wind. Because of this the software can have a difficult time creating keypoints which allows the images to be tied together. We recommend significantly increasing your overlap in these areas to compensate. Perhaps in the 80-90% range.
Hope this helps.