[FOSS-GPS] Accurate timestamp in rtkpost

abmmusa amusa2 at uic.edu
Sun Dec 6 15:42:02 PST 2015


David, Thanks for your suggestion on the PPS output and SMPTE timecode. I'll
look more into it and see if we can use that easily. Also, I posted the
pos/solution file just below the the RINEX file in the earlier email. It
only shows milliseconds. But later I processed with increased decimal count,
and I get micro or nano seconds output as you said earlier. 

I looked into the inaccuracy for the iPhone end. We have approx. 8 minutes
long videos with 30fps and hence total frame is approx. 15000. I looked into
the difference between timestamp for every pair of adjacent frame. Because
of 30fps, the time-gap should be 0.033. For all time-gap values, our mean is
0.033 and standard deviation is also 0.033. So we can be off by 1 frame
because of the iPhone time drift. The accumulation of this drift is also
linear. 

Now to see if there is any offset between iPhone time and GPS time, I looked
visually between many pairs of frames, where a pair is formed for the
closest location between two frames (from two different driving) according
to the Topcon GPS, and corresponding timestamp matching for iPhone and
Topcon GPS. Now if there is a time offset, I should be seeing one frame
always behind or always after. However, I see they are at the same location
or before or after with random amount of time-gap. Sometime the time-gap is
even 2/3 seconds. So I don't see any fixed time-offset between iPhone and
Topcon GPS. It looks strange that the iPhone network time will be drifting
for few seconds within 8 minutes of video recording. And as I said above the
standard deviation of drift is 0.033 according to our data. 

I'll look more into this, specially the iPhone video timestamps, and come
back later if I find what might be going wrong. In the meantime, please let
me know if anyone here did this kind of video synchronization, and what
common pitfalls I should be careful about. 

-Musa



--
View this message in context: http://open-source-gps-related-discussion-and-support.1099874.n2.nabble.com/Accurate-timestamp-in-rtkpost-tp7573360p7573367.html
Sent from the Open Source GPS-related discussion and support mailing list archive at Nabble.com.


More information about the FOSS-GPS mailing list