Would it be possible to add the ability to optimize camera orientation by including z values for "check points" in the optimization step? I could see that being a challenge because reconstruction is still at the sparse cloud stage, but I would rather fix alignment to start with than subtract the systematic error from poor alignment later...
I was reading a paper (Brasington et al 2003) that discussed using a linear trend surface analysis to remove systematic error from a photogrammetrically derived surface, and I was thinking about doing a similar correction (moving window kriging) to create an error surface and apply it to my DSMs in a region of my study area where I have poor z control for multiple flights because I don't have any good GCPs. Then I thought - how cool would it be if PS allowed me to use checkpoints to improve alignment and remove bias, taking advantage of the knowledge of camera orientation?!
To elaborate more, I have reasonably well-distributed GPS check points on good surfaces (flat, bare, unchanging ground), and I have several hundred more that I've picked from two lidar flights (also compared with each other). I can use these check points to estimate uncertainty and systematic bias of the DEM, and for linear trend analysis, kriging, or some other error estimation, but it seems like I would get better results if I did the error correction within PS, since error generally seems low frequency/amplitude ("smiles and frowns" because of a drift from true Z due to loss of control) and check points could improve alignment between control points.
Reference:
Brasington, J., Langham, J., Rumsby, B., 2003. Methodological sensitivity of morphometric estimates of coarse fluvial sediment transport. Geomorphology 53, 299–316. doi:10.1016/S0169-555X(02)00320-3
Topic: (aerial mapping) allow use of check points in model optimization step (Read 3783 times)
