Cross-talk correction for the Y4KCam
-------------------------------------

J. Subasavage
07 Dec. 2011

The cross-talk on Y4KCam is mirrored about the center and thus, a
single saturated star affects pixels in all four quadrants.
Therefore, it is necessary to define coefficients for three "victim"
quadrants for each "source" quadrant.  To correct images for
cross-talk, each victim quadrant must be rotated/flipped to match the
orientation of the source quadrant.  Then, follow this formula.

victim = victim - (source*coefficient)

The cross-talk correction is best performed prior to any other
processing as one can set digital saturation to equal 65535
identically.  Otherwise, some saturation threshold for calibrated
images may need to be determined on a case-by-case basis.  Below are
the matrices of coefficients, both for saturated and unsaturated
images.

These coefficients were determined in February 2011 using a dataset of
49 frames with considerably saturated stellar images.  In general, the
correction works quite well except near the edges of the CCD.  In the
future, we hope to determine the coefficients with an increased number
of frames to better sample these regions.

-----------------------------------------------------------------------
(for all pixels between 0 and 65534 counts)
   xtcoefflo = [[0.000000000, 0.002367490, 0.000425089, 0.000310884],$
                [0.002377290, 0.000000000, 0.000352606, 0.000270416],$
                [0.000374571, 0.000269293, 0.000000000, 0.001581960],$
                [0.000300836, 0.000303373, 0.001707190, 0.000000000]]

(for all saturated pixels == 65535 counts)
    xtcoeffhi = [[0.000000000, 0.002785190, 0.000432938, 0.000365557],$
                 [0.002784870, 0.000000000, 0.000453587, 0.000423568],$
                 [0.000371918, 0.000251799, 0.000000000, 0.001829820],$
                 [0.000297175, 0.000319047, 0.002619930, 0.000000000]]
-----------------------------------------------------------------------

The matrices are defined as follows:

[ (ul:ul), (ul:ur), (ul:ll), (ul:lr)
  (ur:ul), (ur:ur), (ur:ll), (ur:lr)
  (ll:ul), (ll:ur), (ll:ll), (ll:lr)
  (lr:ul), (lr:ur), (lr:ll), (lr:lr) ]

Here, the first listing is the victim (e.g., ur:) and the second
listing is the source frame (e.g., :ul).

The pixel definitions are as follows:

       x1   x2   y1   y2
ul = [   0:2051,2052:4103]
ur = [2052:4103,2052:4103]
ll = [   0:2051,   0:2051]
lr = [2052:4103,   0:2051]