After last update of fiber thermal shift measurements, some issues have been found:
Noisy fiber profile, yielding unreliable cross-correlation with fiber model (see figure below)
Unreliable shift sub-pixel measurement
The first issue was fixed by setting a threshold on the initial fiber locations, whereby fibers cannot be closer than 5 pixels (consistent with fiber centroid measurements). The second issue was fixed by fitting a Gaussian + constant term around the maximum cross-correlation.
After last update of fiber thermal shift measurements, some issues have been found:
The first issue was fixed by setting a threshold on the initial fiber locations, whereby fibers cannot be closer than 5 pixels (consistent with fiber centroid measurements). The second issue was fixed by fitting a Gaussian + constant term around the maximum cross-correlation.
QA plots were also improved