Hi there,
I've been running a fair share of low-altitude, high-GSD photorgrammerty projects so far with good results. Now I'm working on data taken from higher altitude. Many of the
images are partly blurred or show some lens distortion at the edges, leading to issues when palcing markers, as the image quality of some GCPs is pretty bad.
GCPs were measured with a highly accurate DGPs system
(~2cm).
GSD is at about only
5.4cm / pixel.
The markers used are way too reflective metal crosses (30x30cm, struts about 5cm wide) with a white sheet of paper. On bad images this ends up in a reflective mush (see pictures).
As there's no way to resurvey the area, I'll have to try to get the best results with the data I have. How would you recommend dealing with this?
Setting markers: - Should I place markers on the blurry images, despite the obvious lack of accuracy due to image quality?
- Should I delete images/cameras that are partly blurry (blurry edges but otherwise fine) or mask the blurry areas?
- Should I just leave those blurry bits out and set no marker on these images?
- Or should I set all markers on not blurry images first, let photoscan set the others for me and just confirm the automatically set ones on blurry bits?
Setting accuracy:- Should I set accuracy lower for blurred markers? If so, by how much?
So far I've been removing the worst images, not setting markers in images where they appear too blurred and setting overall marker accuracy to about 5cm[/b], which seems very low to me, given the manual recommends 0 for DGPS GCPs. How exactly does marker accuracy work? I would expect it to be not only RTK-error, but rather GSD-based, as in lower GSD leading to worse accuracy - if a single pixel is 5cm wide, is it even possible to come up with an accuracy better than 2.5cm (given I manually set the marker and it's set within one pixel=5cm)?
So, how would you recommend dealing with my blurry markers?
Sorry for the long post. I'm thankful for any info and will report back on the results.
Cheers.