V-Map analysis A;ways a skeptic with such tests from a vendor....... Scroll to bottom of analysis. And I quote..."This may in fact indicate that the GPS survey was not sufficiently accurate to serve as a bench mark for SfM accuracy estimating or that the targets of the CPs have deteriorated in the time span between GPS survey (January 2015) and date of image acquisition (August 2015)."
So, if the company was skeptical about the GPS, what gives? Unless the parking lot was on the San Andreas fault, what would 7 months do to bust this test??
So many questions.
GISJoel_GetItSurveyed, post: 415569, member: 11867 wrote: So many questions.
It has been a long time since I've taken statistics, but the fact that their "average error" is a negative value seems odd to me. The mean of a normal distribution should be zero, shouldn't it? Using this as an assessment of accuracy seems odd.. ?
Even more obviously, they claim to be measuring the center of 6" parking stripes to within one third an inch.. which seems like quite a remarkable feat. If you want to be performing high accuracy assessment, is a 6" parking stripe really the best you can do for control? You're claiming an accuracy 20 times better than the width of your "target."
I requested some information from them. They are right up the street from me in Gainesville. Their test bed is the Santa Fe College Gymnasium parking lot. There may be a University of Florida Geomatics graduate associated with this. I know one of the professors (see www.geomatics.us) up there has been developing an alternative low cost GNSS units (like around $3,500). Looks like there might be quite a bit of GNSS research development in these parts.
I don't buy it. Without omega, phi, and kappa orientation of each camera station you don't have all the math. You would need a super accurate IMU along with RTK and event logging to measure orientation directly. Hence ground control to estimate these variables through a bundle block adjustment.
