DEM and orthomozaic per points approximation accuracy

[mmikhail]

Hello,

We are using photoscan to reconstruct ground surface from UAV footage, it's a great tool, thank you guys providing it, definitely best on the market.

For our task it is crucial to access accuracy of DEM and orthophoto computation for every given point of the model, so footage can be recaptured for areas where average quality of approximation (per surface of per volume with specific characteristics) is too low. So we are looking for a solution to access approximation error or standard deviation for every particular X, Y point of model, not only for GCPs.

We did find some workarounds with projecting random points from UAV footage photos and approximated resulting projection errors, but it seems to be not the easiest way for computing this parameter.

Is there a generic way to compute approximation accuracy for every pixel of say DEM or orthomozaic?

Thank you in advance:)