1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
|
.help apbackground Aug90 noao.twodspec.apextract
.ce
Background Determination
Data from slit spectra allow the determination and subtraction
of the background sky using information from regions near the object
of interest. Background subtraction may also apply to cases of
scattered light though other techniques for scattered light removal
may be more appropriate. The APEXTRACT package provides for determining
the background level at each wavelength (line or column along the dispersion
axis) from a set of regions and extrapolating and subtracting the
background at each pixel extracted from the object profile. The
type of background used during extraction is specified by the parameter
\fIbackground\fR. If the value "none" is used then no background is
subtracted and any background parameters defined for an aperture are
ignored. If the value is "average", "median", "minimum" or "fit" then a
background is determined, including a variance estimate when using variance
weighted extraction (see \fIapvariance\fR), and the subtracted background
spectrum may be output if the \fIextras\fR parameter is set.
The basic aperture definition structure used in the APEXTRACT package
includes associated background regions and fitting parameters. The
background regions are specified by a list of colon delimited ranges
defined relative to the center of the aperture. There are generally
two ranges, one on each side of the object, though one sided or more
complex sets may be used to avoid contaminated or missing parts
of the slit. The default ranges are defined by the parameter
\fIb_sample\fR. Often the ranges are better set graphically using a
cursor by invoking the 'b' option of the aperture editor.
If the background type is "average", "median", or "minimum" then pixels
occupying these regions are averaged, medianed, or the minimum found to
produce a single background level for all object pixels at each wavelength.
Note that the "average" choice does not exclude any pixels which may
yield a background contaminated by cosmic rays. The "median" or "minimum"
is recommended instead.
If the background type is "fit" then a function is fit to the pixels in the
background regions using the ICFIT options (see \fBicfit\fR). The
parameter \fIb_naverage\fR may be used to compute averages or medians of
groups or all of the points within each sample region. The fit is defined
by a function type \fIb_function\fR; one of legendre polynomial, chebyshev
polynomial, linear spline, or cubic spline, and function order
\fIb_order\fR (number of polynomial terms or spline pieces). An
interactive rejection of grossly deviant points from the fit may also be
used. The fitted function can define a constant, sloped, or higher order
background for the object pixels.
Note that the background setting function, the 'b' key in \fBapedit\fR,
may be used to set the background regions for all the background options
but it will always show the result of a fit regardless of the background
type.
After determining a background by averaging, medianing, minimizing, or
fitting, a box car smoothing step may be applied. The box car size is
given by the parameter \fIskybox\fR. When the number of available
background pixels is small, due to a small slit for instance, the noise
introduced to the extracted object spectrum may be unsatisfactorily large.
By smoothing the background one can reduce the noise when the background
consists of a smooth continuum. The trade-off, however, is that near sharp
features the smoothing will smear the features out and give a poorer
subtraction of these features. One could extract both the object and
background separately and apply a background smoothing separately using
other image processing tools. However, this is not possible for variance
weighted extraction because of the intimate connection between the
background levels, the profile determination, and the variance estimates
based on both. Thus, this smoothing feature is included.
The background determined by the methods outlined above is actually
subtracted as a separate step during extraction. The background
is also used during profile fitting when cleaning or using variance
weighted extraction. See \fBapvariance\fR and \fBapprofile\fR for
further discussion.
.ih
SEE ALSO
approfile apvariance apdefault icfit apall apsum
.endhelp
|
