ESO
SINFONI Pipeline User Manual
Doc:
VLT-MAN-ESO-19500-3600
Issue:
Issue 1.0
Date:
Date 2005-10-19
Page:
68 of 88
clean mean of all polynomials is computed. The pixels whose intensity constant (
+
'
) or slope (
+
'
)
coefficients exceed the mean by thresh_sigma_factor times the standard deviation of the pixel intensity
are rejected. Pixels whose non quadratic coefficient exceeds non_lin_threshold are rejected.
Thus a map of the pixels which do not have a linear sensitivity is generated, together with a table contain-
ing the polynomial coefficients: the non-linear terms give an estimate of the detector non-linearity.
10.1.4
Nearest neighbours bad pixel cleaning
This method is applied in the recipes si_rec_mflat and si_rec_distortion to clean the bad pixels of a flat frame.
The algorithm is controlled by the parameters method_index and factor, and, if method_index is 4, the bad
pixel cleaning is repeated iterations times.
(b)
(a)
Figure 10.1.0: A bad pixel is indicated in black. If method_index is 1,2 or 4 this is corrected evaluating the
eight nearest neighbours pixel intensities (a). If method_index is 3, the four closest pixels along the dispersion
direction are considered (b).
If method_index is 1, for each image pixel intensity,
6
, the eight nearest neighbours pixels intensities,
, are considered and the corresponding median,
6
, is computed excluding from this set those
pixels whose intensity is NAN.
6
$
5
!
:
:
:
$B
