Flat-frame correction¶
The flat-frame correction is a division of a source frame X by a flat frame F and multiplying it by a constant k. The constant k is determined such that the range of pixel values and the gain factor (number of photons per ADU) are roughly preserved. The constant k is computed as a mean value of the correction frame F by means of the robust mean algorithm, see the chapter Robust mean.
(1)¶Y(x, y) = X(x, y) - k\,\frac{X(x, y)}{F(x, y)} = X(x, y) - \bar{F}\,\frac{X(x, y)}{F(x, y)}
Bad and overexposed pixels require special treatment:
If a pixel is bad in either source frame or dark frame, it must be marked as bad on the corrected frame.
If a pixel is overexposed on the source pixel, it must stay overexposed after the correction.
If a pixel is overexposed on the flat frame, it is marked as bad on the corrected frame.