Header offsets

Offsets can be acquired from the FITS header for each file (this prevents the use of a single FITS file containing all jitter data, for simple implementation purposes). The user needs to specify what are the keywords to look for in each frame header to get the offset information in x and y, in pixels.

NTT offsets as found in SOFI data have been shown to be reliable to subpixel accuracy in the best cases, usually in the range of about 5 pixels in error, and never more than 10. Differences are mainly due to the optics and wavelength range. The default search area in this case is set to $\pm 10$ pixels in x and y, which should cover more than really needed. The default measurement area is set to $\pm 25$ pixels, which amounts the total amount of pixel operations to:

$(2\times10+1)^2 \times (2\times25+1)^2 = 1,147,041$