blob: 742fa7cb86f3ca3360c6f3c41caa4213263d952d (
plain) (
blame)
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
|
.help center1d May93 xtools
.ih
NAME
center1d -- One dimensional centering
.ih
SYNOPSIS
.nf
center = center1d (initial, data, npts, width, type, radius, threshold)
real initial # Initial guess
real data[npts] # Data points
int npts # Number of data points
real width # Feature width
int type # Feature type
real radius # Centering radius
real threshold # Detection threshold
.fi
.ih
ARGUMENTS
.ls initial
Initial guess for the center of the feature.
.le
.ls data[npts]
Pixel data vector.
.le
.ls npts
Number of points in the data vector.
.le
.ls width
Width used to define the convolution function. If the width is 1 or less
then the nearest minimum or maximum is returned. If the width is greater
than 1 then a minimum with of 3 is used in the algorithm.
.le
.ls type
Type of feature. The feature types are defined in the file <xtools/center1d.h>.
Currently the types are emission and absorption features.
.le
.ls radius
Centering radius or error limit about the initial guess.
.le
.ls threshold
Minimum difference between the maximum and minimum pixel value in the
region around the initial guess allowed for detecting a feature. For
data which is all positive and the type is emission then the
threshold is also used as an absolute cutoff.
.le
.ih
DESCRIPTION
If the width is 1 or less then the nearest minimum or maximum is found.
The centering radius is still applied as is the threshold. If the width
is greater than 1 then a minimum width of 3 is used in the algorithm.
The one dimensional position of a feature is determined by solving the equation
(1) integral {(I-I0) f(X-XC) dX} = 0
where I is the intensity at position X, I0 is the continuum intensity, X is the
pixel coordinate, and XC is the desired feature position. Figure 1 shows
the range of pixels used in determining the continuum intensity, the feature
threshold, and solving the equation.
.ks
.nf
Figure 1: Data Feature Vector
+-----------------------------------------------------------+
-| * |
S| * * |
| * * |
t| * ** |
| * * |
r| * * |
| * * |
e| * * |
| * * |
n| * * * *|
| * * * * * |
t| * * |
| * * |
h| * |
-| * |
+---------+-----------------+---------------------+---------+
-B -A 0 A B
X-XC
A = radius + 0.5 width B = radius + 1.5 width
.fi
.ke
The range -A to A is used to determine the continuum intensity and
the strength of the feature. For absorption features the continuum
intensity is the maximum point in this range while for emission
features the continuum is set to zero. Admittedly these are not real
measures of the continuum but they contain the fewest assumptions
and are tolerant of nearby contaminating features. The feature strength
is the difference between the maximum and minimum values. If the feature
strength is less than the specified detection threshold then a value of
INDEF is returned for the feature position.
.ks
The range -B to B includes the range of allowed feature positions plus the
half-width of the feature. This range is used in solving equation (1).
The convolution function f(X-XC) is a sawtooth as shown in figure 2.
For absorption features the negative of this function is used.
.nf
Figure 2: f(X-XC)
+-------------------+-------------------+
| | * |
| | * * |
| | * * |
0 +-*-*-*-*-----------*-----------*-*-*-*-+
| * * | |
| * * | |
| * | |
+-------+-----------+-----------+-------+
-width/2 0 width/2
X-XC
.fi
.ke
The two figures graphically define the parameter \fIwidth\fR. Generally
it should be set to a value near the actual width of the emission or absorption
feature. If the width is too wide then the center will be affected by blending
from nearby lines while if it is too narrow the accuracy of the centering is
decreased. The parameter \fBradius\fR determines how far from the initial
estimate for XC the interactive solution of (1) may go.
Equation (1) is solved iteratively starting with the initial position.
When successive positions agree within 0.1% of a pixel the position is
returned. If the position wanders further than \fIradius\fR from the
initial guess or outside of the data vector then the procedure returns
the value INDEF. If more than 100 iterations are required or the corrections
per iteration exceed the minimum correction reached after 3 further iterations
then the solution has failed to converge and INDEF is returned. Note that
this latter condition may occur if the width is too small in a flat topped
profile.
This task uses the one dimensional image interpolation package \fBiminterp\fR
in solving equation (1).
.ih
BUGS
Though the algorithm does not fail if the width is made very small the
results become unreliable. Therefore a silent limit of 3 is imposed
by the algorithm. If there is ever a need to allow smaller widths
then the procedure can be changed and the applications relinked.
.endhelp
|
