Extraction of coregistration tie-point infromation

48 views
Skip to first unread message

Mitro Müller

unread,
Mar 15, 2021, 9:41:38 AM3/15/21
to FORCE
Greetings,

I would be interested to investigate existence of spatial variability in accuracy of Sentinel-2 multi-temporal co-registration. FORCE log-files provide information of coregistration results in form of number of identified tie points, the shift in x and y direction and the RMSE of the retrived image transformation. 

Is there a way to extract information on the direction and magnitude of shifts per recorded tie-point? 

Best regards,
Mitro Müller

david frantz

unread,
Mar 19, 2021, 11:50:43 AM3/19/21
to FORCE
Hi Mitro,

Although this information is available within the code, there is no out-of-the-box possibility to extract this.
If it is an option for you to look into the code, and re-compile the source-code, I could point you to the relevant lines. Some simple "printf" would suffice, I guess.
Please let me know if this works for you. 
Cheers,
David

Mitro Müller

unread,
Mar 22, 2021, 8:34:58 AM3/22/21
to FORCE
Hi,

I myself do not have strong background with C. However, I believe I can get help with this.
Thus, I would be grateful if you could point out the relevant lines where to implement changes.  

With best regards,
Mitro Müller

david frantz

unread,
Mar 26, 2021, 11:45:44 AM3/26/21
to FORCE
Hi Mitro,

you will need to look into src/lower-level/coreg.c

In the function transform_from_dm, the final transformation will be computed on the final set of tie points.

In the for-loop beginning in line 1193, you could print the pixel locations for target and base by adding an "else" clause in line 1208.

Let me know if that helps!

Cheers,
David
Reply all
Reply to author
Forward
0 new messages