Georeferencing with the new LogDump

[Yups]

Hello!

I am really happy with the new LogDump in qgroundcontrol and the new features in the firmware.

But there ist still one thing missing, to have a perfect platform for taking aerial pictures.

Sync the triggerevent coordinates with the pictures.

Select the folder with the source jpgs, select the log, check jpg and log time to avoid wrong selections. (with a definable time range).
Very important is a customable delay for the camera trigger. Some cameras have an delay of 0,5s.

It would be great to have very precise referenced pictures.

Lat, Long and Alt must be written in the EXIF-Data.
This can be done with the free EXIF-Tool by Phil Harvey: http://www.sno.phy.queensu.ca/~phil/exiftool/


best regards from Germany
Yannick

[Max]

Hi Yannick,

There are plenty of photo tagging apps out there that are very full featured. LogDump gives you a log file to feed into those programs. There IS a camera trigger delay in the latest logDump (also in QGC 1.3 RC1). So you can narrow down the export only to triggered events, and match against a folder of photos on a consecutive basis (photo 1 matches 1st trigger point, photo 2 goes with the 2nd, and so on). Or use LogDump to determine the actual time offset of all your pictures, and use "traditional" tagging based on photo time stamp (still accurate to the second with most cams).

I don't see how logDump would be any better at matching your photos to the log than any other decent tagging software. Am I missing something? The only thing better I can think of is if either the camera could store external data with a photo (feed it lat/lon from AQ) or if a photo ID, from the camera, could be stored in the AQ log somehow. Neither sound very realistic.

Cheers,
-Max

[Cookiemonsta]

I can recommend geosetter for doing this
It's free and very good.

[Yups]

Thanks for your relplies!

I already know Geosetter.

The problem i have in mind is very simple. Two things could be a problem:

1st: The camera misses one shot, so the complete alignment is wrong
2st: The camera takes one picture more than expectet, same problem

Both problems are common. One possibility is the alignment by time, as Max already mentioned. But it is not precise.
When the Copter flies with ~ 10m/s, the resolution is 10m...

So we need a combination. The Time must be in a certain range, maybe about 3-5 seconds, but the definite alignment should be made with the TriggerEvent.


Best regards
Yannick

[Max]

Well my point is that the AQ doesn't know when your camera actually fired, or if it fired at all. W/out some kind of sensor that can detect what the camera is doing, or a way to feed position data to the camera, I'm not seeing how the data can be improved.

Using the trigger delay options one can get as accurate as possible right now. The timetamp of the photo doesn't have to be precise, just needs to match uniquely with an exported log record (there can't be 2+ photos taken within the same triggering moment). It should work as long as there is sufficient time gap between photos (eg. if you're shooting every second, that could get confusing).

Accurately geotagging photos taken at 10 m/s is going to be challenging. You need a camera that fires with a very predictable delay (fixed focus, etc). Also GPS isn't super accurate to begin with, so some error will be introduced anyway.

Actually I think feeding precise position data into the camera would be ideal, especially since the photos will then come out already tagged. I'm not sure if there are any cameras out there right now capable of that. I know you can get cameras with built-in GPS, but those typically aren't very accurate due to size/antenna/cost constraints (I do have one and it's pretty cool for vacations and such, but it's not a precise instrument).

Cheers,
-Max

[Max]

Here are some cams with data input (position) capability: http://www.tetracam.com/Products1.htm Not cheap though, and fairly specific to vegetation surveying.

One way to detect when a camera fires would be to use a hotshoe or other remote flash interface.

-Max

[Yups]

[Yups]

Has anybody thought about this solution?

Yannick

[Kisssys]

[Yups]

Dear Steve,

thanks for your tests! I use a Sony NEX-5N for my aerial shots because of its weight. The DSLRs are to heavy for long flights
Yes, with the normal flash connector you will not get any relationship to the shutter or flash speed.
But thats not necessary!

Do you know the Piksi RTK GPS System?

50hz update rate and cm-level precision. With this receiver, you would get a new position every 20ms. So it would play a role if you sync it with the camera delay or not. For several cameras, there is only an IR-shutter available -> the delay is not fixed!

The camera angle doesn't matter. You could even fly with a hard mounted cam. Besides Dronemapper thera are agisoft photoscan or Pix4D. All of them use Aerial Triangulation, so its no problem when the cam does not point directly to the ground.

Groundcontrolpoints are one possibility. But you need an extra high precision GPS device wich costs some thousand bucks and the referencing signal costs per minute...

So it would be very useful, to get Aerial Shots with half-meter or 1/4 meter precision (this should be possible with the AQ GPS) WITHOUT additional GCPs.

Best regards from Germany,
Yannick