Creating 2D ortho mosaic under large covered area

Joined
Oct 19, 2016
Messages
154
Reaction score
54
[Phantom 4 Pro]

A client wants to create an ortho mosaic under a very large covered area - approx 500' x 100' (assume altitude 30' AGL). Assume no GPS. How do I create the ortho without the aid of GPS (other than hand-flying and estimating image overlap)? Is this a job for a Ground Control Point system? Recommendations for an inexpensive GCP setup?

Thanks in advance...
 
[Phantom 4 Pro]

A client wants to create an ortho mosaic under a very large covered area - approx 500' x 100' (30' high). Assume no GPS. How do I create the ortho without the aid of GPS (other than hand-flying and estimating image overlap)? Is this a job for a Ground Control Point system? Recommendations for an inexpensive GCP setup?
Another (big) issue will be that without GPS there is no way for the orthophoto software to know where each image was taken.
Only 30 ft high is going to be a bit of a problem as well.
If there's a roof and maybe some supporting structure, you will have to fly even lower than 30 ft, limiting how much you can cover per frame.
 
Another (big) issue will be that without GPS there is no way for the orthophoto software to know where each image was taken.
Only 30 ft high is going to be a bit of a problem as well.
If there's a roof and maybe some supporting structure, you will have to fly even lower than 30 ft, limiting how much you can cover per frame.
The spatial awareness issue is the basis of my question. That's why I mentioned, "Assume no GPS.". Also understand the demands of the low altitude and number of images required. I'm trying to find a system where you can physically put a positioning transmitter in the center of the area and communicate positioning data to the drone, thereby creating an artificial GPS system under the covered area. I also understand it would probably involve attaching a receiver to the drone.
 
I've actually done something like that while still working at the AF Research Lab. We were deploying pallets with GPS trackers and wanted the receivers the locked and tracking before we pushed them out the back door of the airplane. Solution was receiver that got GPS info from the plane's external antenna then rebroadcast it through another antenna inside the cargo area. No modifications needed on the pallet receivers, but the retransmitter wasn't cheap! Unfortunately, it's been a few years and I couldn't tell you who made the system we used.
If you look through the symposium archive of meetings like SPIE or MSS (Military Sensing Symposium), you might find some of the papers we presented on the program. Look for anything like "Ballistic Winds".
GCPs wouldn't help. They are just accurately located makers that are visible in multiple images and are used to determine the 'true' location of your ortho
 
Last edited:
Programs like Agisoft Photoscan and Pix4D offer image matching with or without GPS in the EXIF. I've not used Pix4D, but Agisoft matches pretty well without GPS, even just printing off a few GCD sheets and scattering them around the area should help a lot with accuracy.

Both offer free trials, but are several thousand dollars to buy outright, or with monthly subscription. The former can be found on places outside the scope of this forum for those whose moral compass is a little 'greyer', if you catch my drift.
 
I've actually done something like that while still working at the AF Research Lab. We were deploying pallets with GPS trackers and wanted the receivers the locked and tracking before we pushed them out the back door of the airplane. Solution was receiver that got GPS info from the plane's external antenna then rebroadcast it through another antenna inside the cargo area. No modifications needed on the pallet receivers, but the retransmitter wasn't cheap! Unfortunately, it's been a few years and I couldn't tell you who made the system we used.
If you look through the symposium archive of meetings like SPIE or MSS (Military Sensing Symposium), you might find some of the papers we presented on the program. Look for anything like "Ballistic Winds".
GCPs wouldn't help. They are just accurately located makers that are visible in multiple images and are used to determine the 'true' location of your ortho
This is very interesting. However, I'm not on an Air Force budget. Wondering if someone has boiled it down to a commercial system that I could hang my hat on. But remember, at no time would GPS be available.
 
GCPs wouldn't help. They are just accurately located makers that are visible in multiple images and are used to determine the 'true' location of your ortho

Agisoft can line up the photos OK provided there's enough detail on the ground to work from. The GCPs are there to correct the warping in the generated point clouds before you move on to a dense cloud + mesh.
 
@Goof is correct, Photoscan and (to a lesser degree) Pix4D can model interiors without geolocated imagery. Try following these tips that Agisoft offers for interior models. You can use scale bars and markers to help you with your model.

How high is your interior? Turn on your interval camera and fly your lines as close as you can. You can use any of the flight planning apps to approximate the distance between lines and then go for it. Odds are that you'll collect more imagery than needed. If you're trying to model equipment, you'll want to shoot oblique imagery, as opposed to all nadir.
 
@Goof is correct, Photoscan and (to a lesser degree) Pix4D can model interiors without geolocated imagery. Try following these tips that Agisoft offers for interior models. You can use scale bars and markers to help you with your model.

How high is your interior? Turn on your interval camera and fly your lines as close as you can. You can use any of the flight planning apps to approximate the distance between lines and then go for it. Odds are that you'll collect more imagery than needed. If you're trying to model equipment, you'll want to shoot oblique imagery, as opposed to all nadir.
I have enough head room to comfortably fly at 30'. I suspect that's a boatload of images.
 
Check out the DJI RTK on the M200. It can do positioning from a base unit.
This is exactly what I want. Cost is probably a deal killer for now, but if it converts to revenue, then it may work. Do you have any scoop on the base unit?? I'm assuming the base unit and the Matrice would talk to each other....
 
Last edited:
I suspect that's a boatload of images.
Using a footprint of about 30'x20' at 30'AGL and a front and side overlap of ~80%, you're looking at around 1400 images. That's a very rough estimate.

Do you have any scoop on the base unit?
The M200 is expected to be released by September(?), but you're looking at a base price of $16k. It'll be a great mapping unit, if it does everything they claim. Check this out for a bit more information regarding the existing RTK system: D-RTK GNSS - Specs - DJI
 
Using a footprint of about 30'x20' at 30'AGL and a front and side overlap of ~80%, you're looking at around 1400 images. That's a very rough estimate.


The M200 is expected to be released by September(?), but you're looking at a base price of $16k. It'll be a great mapping unit, if it does everything they claim. Check this out for a bit more information regarding the existing RTK system: D-RTK GNSS - Specs - DJI
This is pretty cool, but I'm still not convinced it can do an automated mission that is completely devoid of GPS - per my original post.
 

Members online

No members online now.

Forum statistics

Threads
143,054
Messages
1,467,297
Members
104,919
Latest member
BobDan