I have been mapping some forestry for a while with varying levels of success. I’m looking for any advice you can give on best practices for image capture of dense tree coverage without many defining landmarks from which to create tie points.
The output is a 2D ortho mainly for representation of areas of more and less sparse canopy. The client gets a stitched JPG as well.
At the moment I’m getting a lot of artefacts in the trees and the software is obviously finding it difficult to create the necessary tie points.
I fly for a forestry company all the time. My standard settings are 400ft altitude, default the rest. I fly for GIS stand delineation purposes mostly, so I’m not aiming for showpiece quality. There’s some smudging and deformation in the fine details of the end result, but nothing that detracts from the purpose of the image. Depending on your purposes, this may still be unacceptable to you. I’m including a sample of some dense trees at 1:250 zoom. I’m not exactly sure what you mean by artifacts, but what you see in this sample is what my client calls perfectly acceptable, and there’s nothing special I did to achieve it besides increase altitude to the max. The resolution on this data is about 1.2in/px. Someone else here may have suggestions for how to increase the crispness of forest images down to a hair, but it could also be that this is just where the technology is at this time. DroneDeploy documentation does warn that monotone landscapes like row crops or forests may be a challenge to their stitching software, and gives some suggestions for how to optimize the results by adjusting altitude, image overlap, and other factors. DroneDeploy also recommends turning on terrain following if your forest landscape has a lot of elevation change.
