Will a GPS camera improve the accuracy of DTMs?

[stefan]

I want to create as accurate as possible DTMs from pictures from an UAV. I will use some kind of rover to get GCPs. How should I think about camera positions? Is it unnecessary to get a camera with GPS, or will it increase the accuracy of my DTM?

Also interested in literature about this.

[JMR]

I think having on-camera GPS can't improve results over decent GCP but CAN help to speed up alignment stage and  even to avoid not-aligned images. It would also help during GCP marking because model is roughly geo-located prior to entering them, so when you import points, they should appear floating somewhere near their actual position and you just have to drag them on each photo to their right projections.
But let me add that better cameras usually don't boast on-board GPS, so you'd better extracting rough(<2m?) photo coordinates out from the UAV flight log. That makes unnecessary any on-camera crappy(<5m?) gps and gives you better and broader camera choice.
Regards!

Jos?
GEOBIT

[stefan]

Thank you for your reply.

Its good to know that it probably wont affect the accuracy. However speeding up the process is always interesting. Any idea how much of a difference it could make?  Are we talking 0,5 % or 20 %?

[Alexey Pasumansky]

Hello Stefan,

Using Ground Control pre-selection based on the camera position coordinates is significantly faster even compared to Generic pre-selection. Extremely useful it is for big datasets hundreds and thousands of photos.
It could be even times.

Later, after camera alignment is finished you can place GCPs on photos, input precise coordinates, uncheck all the cameras in the Ground Control pane and optimize/georeference the model based on this data.

[dvdsam0]

Hi All
How good is the GPS fix on a GPS enabled camera stuck to the bottom of a multi rotor?
Would it be beneficial to have part of the camera sticking out past the body (not good for the CG)?
Thanks

[Jack_in_CO]

The United State Bureau of Land Management has technical Bulletin 428 that covers this aspect. If you are seeking high-level precision for aerial photogrammetry, you will likely need to have surveyed ground control points.

[RalfH]

Hi Jack-in_CO,

do you have a link to that document?

[gEEvEE]

Hi Ralph,

I suppose he means this document: http://www.blm.gov/nstc/library/pdf/TN428.pdf

For all you GCPs and accuracy freaks out there: keep in mind that in photogrammetry, the following rule is used: If you need a model with x cm accuracy, you need GCPs which have at least 1/3x accuracy. So, if you need to deliver a model with 6 cm accuracy, your GCPs should measured with at least 2 cm accuracy.

Additionally, the ground sampling distance (GSD) of the camera has to be taken into account as well as its real resolving power. A GSD of 10 cm does not mean that the true resolving power is 10 cm: this depends on many variables, the MTF of the optical system and the contrast of the subject being the prime variables.

Also, keep in mind that the georeferencing errors PS delivers do NOT say anything about the accuracy of your model, they just inform your on its global georeferencing accuracy. Obviously, a good georeferening is more likely to be more accurate. A true accuracy assessment can only be obtained with an independent set of ground control measurements against which your model can be validated.

Cheers,

Geert

[tony88]

How about if we employ RTK GPS base/rover system such as the kickstarter project from Swift Nav.
Has anyone tried a similar RTK system?

[Jack_in_CO]

gEEvEE, thanks for adding the link. That is the document I was referencing to.