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
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
|
.help centerpars May00 noao.digiphot.daophot
.ih
NAME
centerpars -- edit the centering algorithm parameters
.ih
USAGE
centerpars
.ih
PARAMETERS
.ls calgorithm = "none"
The centering algorithm. The "gauss" and "ofilter" centering algorithms
depend critically on the value of the fwhmpsf parameter in the DATAPARS task.
The centering options are:
.ls none
The initial positions are assumed to be the true centers. Users should
select this option if the initial centers are known to be accurate,
e.g. they were computed by DAOFIND task.
.le
.ls centroid
The object centers are determined by computing the intensity weighted means
of the marginal profiles in x and y. Centroid is the recommended centering
algorithm for users running PHOT interactively and selecting objects
with the image display cursor, or when the input coordinates may be inaccurate.
.le
.ls gauss
The object centers are computed by fitting a Gaussian of fixed fwhmpsf,
specified by the DATAPARS fwhmpsf parameter, to the marginal profiles in
x and y using non-linear least squares techniques.
.le
.ls ofilter
The object centers are computed using optimal filtering techniques,
a triangular weighting function of half width equal to fwhmpsf as
specified by the DATAPARS fwhmpsf parameter, and the marginal distributions
in x and y.
.le
.le
.ls cbox = 5.0 (scale units)
The width of the subraster used for object centering in units of the
scale parameter. Cbox needs to be big enough to include sufficient
pixels for centering but not so large as to include a lot of noise.
Reasonable starting values are 2.5-4.0 * FWHM of the PSF.
.le
.ls cthreshold = 0.0 (sigma units)
xels cthreshold * sigma above (emission features) or below (absorption
features) the data minimum or maximum respectively are used by the centering
algorithms where sigma is equal to the value of the DATAPARS sigma parameter.
features) the data minimum or maximum are used by the centering algorithms.
DAOPHOT users should leave this value at 0.0 which invokes the appropriate
default thresholding technique for each centering algorithm. Setting
cthreshold to INDEF turns off thresholding altogether for all the centering
algorithms.
.le
.ls minsnratio = 1.0
The minimum signal to noise ratio for object centering. If the estimated signal
to noise ratio is less than minsnratio the computed center will be returned
with an error flag.
.le
.ls cmaxiter = 10
The maximum number of iterations performed by the centering algorithm.
All the centering algorithms use this parameter.
.le
.ls maxshift = 1.0 (scale units)
The maximum permissible shift of the center with respect to the initial
coordinates in units of the scale parameter. If the shift produced by the
centering algorithms is larger than maxshift, the computed center is returned
with an error flag.
.le
.ls clean = no
Symmetry-clean the centering subraster before centering? DAOPHOT users should
leave clean set to "no".
.le
.ls rclean = 1.0 (scale units)
The cleaning radius for the symmetry-clean algorithm in units of
the scale parameter.
.le
.ls rclip = 2.0 (scale units)
The clipping radius for the symmetry-clean algorithm in units of
the scale parameter.
.le
.ls kclean = 3.0 (sigma)
The number of standard sky deviations for the symmetry-clean algorithm.
.le
.ls mkcenter = no
Mark the fitted centers on the displayed image ?
.le
.ih
DESCRIPTION
The centering algorithm parameters control the action of the centering
algorithms. The default parameters values have been proven to produce
reasonable results in the majority of cases. Several of the centering
parameters are defined in terms of the DATAPARS parameter \fIscale\fR,
the scale of the image, and \fIsigma\fR the standard deviation of
the sky pixels.
For each object to be measured a subraster of data \fIcbox\fR / \fIscale\fR
pixels wide around the initial position supplied by the user is extracted
from the IRAF image. If scale is defined in units of the number
the half-width half-maximum of the psf per pixel, then a single value of
cbox can be used for centering objects in images with different psfs.
If \fIclean\fR is "yes" the symmetry-clean algorithm is applied to the
centering subraster prior to centering. The cleaning algorithm attempts
to correct defects in the centering subraster by assuming that the image
is radially symmetric and comparing pixels on opposite sides of the center
of symmetry. The center of symmetry is assumed to be the maximum pixel
in the subraster, unless the maximum pixel is more than \fImaxshift /
scale\fR from the initial center, in which case the initial center is used
as the center of symmetry. Pixels inside the cleaning radius are not edited.
Pairs of pixels in the cleaning region, r > \fIrclean\fR / \fIscale\fR
and r <= \fIrclip\fR / \fIscale\fR and diametrically opposed about the
center of symmetry are tested for equality. If the difference between the
pixels is greater than \fIkclean * sigma\fR, the larger value is replaced
by the smaller. In the cleaning region the sigma is determined by the
noise model assumed for the data. Pairs of pixels in the clipping region,
r > \fIrclip\fR / \fIscale\fR are tested in the same manner as those in
the cleaning region. However the sigma employed is the sigma of the
sky background. DAOPHOT users should leave clean set to "no".
New centers are computed using the centering algorithm specified by
\fIcalgorithm\fR, the data specified by \fIcbox / scale\fR, and pixels
that are some threshold above (below) an estimate of the local minimum
(maximum). \fICthreshold\fR values of 0.0, a positive number, and INDEF
invoke the default thresholding algorithm, a threshold equal to the
local minimum (maximum) plus (minus) \fIdatapars.sigma * cthreshold\fR,
and a threshold exactly equal to the local minimum (maximum) respectively.
After thresholding the signal to noise ratio of the subraster is estimated.
If the SNR < \fIminsnratio\fR the new center is still computed but an error
flag is set.
The default centering algorithm is \fInone\fR is which case the initial
centers are assumed to be accurate and no recentering is done.
The simplest centering algorithm is \fIcentroid\fR. Centroid computes the
intensity weighted mean and mean error of the centering box x and y marginal
distributions using points in the marginal arrays above (below) the minimum
(maximum) data pixel plus (minus) a threshold value. The threshold value is
either the mean, \fIdatapars.sigma * cthreshold\fR above (below) the local
minimum (maximum) if \fIcthreshold\fR is greater than zero, or zero above
(below) the local minimum (maximum) if \fIcthreshold\fR is INDEF. The centroid
algorithm is similar to that by the old KPNO Mountain Photometry Code.
Note that centroid is the only centering algorithm which does not depend
on the value of \fIdatapars.fwhmpsf\fR.
The centering algorithm \fIgauss\fR computes the new centers by fitting a
1D Gaussian function to the marginal distributions in x and y using a
fixed fwhmpsf set by \fIdatapars.fwhmpsf\fR. Initial guesses for the fit
parameters are derived from the data. The gauss algorithm iterates until
a best fit solution is achieved.
The final centering algorithm choice \fIofilter\fR employs a variation of the
optimal filtering technique in which the profile is simulated by a triangle
function of width \fIdatapars.fwhmpsf\fR.
The default thresholding algorithm for all centering algorithms other
than "centroid" is no thresholding.
If the computed shift in either coordinate > \fImaxshift\fR / \fIscale\fR,
the new center is returned but an error flag is set.
1. List the centering parameters.
.nf
da> lpar centerpars
.fi
2. Edit the centering parameters.
.nf
da> centerpars
.fi
3. Edit the CENTERPARS parameters from with the PHOT task.
.nf
da> epar phot
... edit a few phot parameters
... move to the centerpars parameter and type :e
... edit the centerpars parameters and type :wq
... finish editing the phot parameters and type :wq
.fi
4. Save the current CENTERPARS parameter set in a text file ctrnite1.par.
This can also be done from inside a higher level task as in the
above example.
.nf
da> epar centerpars
... type ":w ctrnite1.par" from within epar
.fi
.ih
BUGS
.ih
SEE ALSO
epar,lpar,datapars,phot
.endhelp
|
