We (www.gmxconsulting.co.uk) are developing a large-scale agriculture project in Nigeria (5,000ha) and started to use drones to survey and map topography.
The intention was then to use the Digital Elevation Model for irrigation planning and design. However, when we post processed the images we got very large x y z errors - please see the photos below.
1. Could anyone tell us where we have done wrong? Can we correct NOW (after the aerial survey)?
2. Also we look to outsource the drone surveying and post processing for our projects in Africa - Anyone interested please let us know.
Here are some more details:
Platform: E384
AGL: 120 - 180m
Overlap: Front 60% Side 60%
GCPs: No
Flight time: <60 mins
Cameras: Canon S110 RGB and NGB
Number of photos: 300-400
Photo taken out before processing: None
Software: Photoscan
Many thanks in advance!
Replies
May I suggest you do the survey at the end of the dry season, there will be no grass and I guess you may be able to correct for the trees.
Good luck
Yes, Paul, we will be doing it again next month or so when most vegetation would be have been burnt off by the scotching sun.
Best
Quan
By the way, we would love to assist you in future endeavors in Africa. We are based in South Africa but operate all over the continent. Our specialization are topographic surveys and GIS data extraction from low level photogrammetry. With use our own flavor of fixed wing mapping aircraft with Leica survey grade GPS for pre-marking ground control.
Hi Luke, saw you work in Uganda. please contact me at quan.le@gmx.com
Best
Quan
Hi Quan
From what I see, the terrain is very undulating. This will seriously impact the set overlaps and side laps that you set relative to your design or home location. This seems to have reduced the over and sidelaps significantly, resulting in a "loose" model.
I can guarantee you that if you had to do a formal, terrain based survey over this project, you will find very large inconsistencies, especially in the vertical. I have seen poor sidelaps generate errors over a few meters at times.
While you have no information to substantiate this, having properly surveyed ground control will quickly highlight this. Ground control will never be able to fix bad flight planning as the side and overlaps are already captured.
Having control is essential for anything to considerd a topographic survey. You have to be able to quantify you errors relative to the terrain. The sidelap and overlap errors that photogrammetry software issues are not relative to the ground and these need to be confirmed before any planning and development takes place.
Quan, My first guess based on the high errors in the geotags is that the geotags are skipping pictures or pictures are skipping tags somewhere in the middle. I can help take a look at that if you upload the pictures and raw .tlog to DDMS at app.event38.com.
Secondly, all the areas outside the flight area are washing out the elevation changes within the mission area because of elevation errors caused by having low overlap at the edges - nothing you can do about that, Photoscan exports everything it can calculate and it's up to you to crop out the borders. I'm not sure that will fix everything in this case but the small elevation changes are not visible in your DEM because of the ~100m change caused by the edges.
Lastly, if you can collect any kind of GCPs that will always improve the accuracy of the resulting mosaic/DEM especially vertically. If that capability is available then definitely use it if your primary concern is drainage modeling.
Jeff, We will certainly upload the images along .tlog.
Re geotagging are you looking to get a camera that automatically does that? It seems quite straightforward to me.
Well done to your E384. We were quite nervous at first but our trust grew greatly after a few flights.
We did crashed it too! but we managed to put it back to fly again without any obvious harm! Well done.
Best
Quan
Thanks, we'll take a look and see what we can do.
We are exploring ways to make geotagging automatic - one way right now for our users who also use DDMS is to just upload the .tlog with their images and have us do it. If you're using Photoscan or Pix4d on your own, our solution will not be available for at least several more months, I think.
Great, I'm glad to hear you're happy with it!
@Jeff, why don't you offer geotagging by default ? All you get uploaded to your server are 2 or more timestamped tables of records (timestamped images and timestamped geolocations). Server-side run script simply establishes relation between table of timestamped images and table of timestamped geolocations, so you can inject geotags into Exif by basic correlation. Job can be done by Exif editors, Flickr, Panoramio and other on-line services. If you don't have Exif editor CLI run on your server, you can amend image file names appending geotags to file name FileName1_geotag1.jpg so the job is really easy and straightfoward.
Sorry if that was misleading, we do of course offer geotagging by default! Most cameras can't be purchased with GPS onboard nowadays, so we tie in the cameras to Pixhawk instead, and we built this utility to add the geotags: http://www.event38.com/Articles.asp?ID=270. This utility is free for any Pixhawk user to use, by the way, and it doesn't require an Event 38 drone.
What I was referring to before is that this utility is an automated part of DDMS, which also produces orthomosaics (for free as well).
-
1
-
2
-
3
-
4
-
5
of 6 Next