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
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
|
include <mach.h>
include <math.h>
# APTOPT - One-dimensional centering routine using repeated convolutions to
# locate image center.
define MAX_SEARCH 3 # Max initial search steps
int procedure aptopt (data, npix, center, sigma, tol, maxiter, ortho)
real data[ARB] # initial data
int npix # number of pixels
real center # initial guess at center
real sigma # sigma of Gaussian
real tol # gap tolerance for sigma
int maxiter # maximum number of iterations
int ortho # orthogonalize weighting vector
int i, iter
pointer sp, wgt
real newx, x[3], news, s[3], delx
real adotr(), apqzero()
begin
if (sigma <= 0.0)
return (-1)
# Allocate working space.
call smark (sp)
call salloc (wgt, npix, TY_REAL)
# Initialize.
x[1] = center
call mkt_prof_derv (Memr[wgt], npix, x[1], sigma, ortho)
s[1] = adotr (Memr[wgt], data, npix)
#if (abs (s[1]) <= EPSILONR) {
if (s[1] == 0.0) {
center = x[1]
call sfree (sp)
return (0)
} else
s[3] = s[1]
# Search for the correct interval.
for (i = 1; (s[3] * s[1] >= 0.0) && (i <= MAX_SEARCH); i = i + 1) {
s[3] = s[1]
x[3] = x[1]
x[1] = x[3] + sign (sigma, s[3])
call mkt_prof_derv (Memr[wgt], npix, x[1], sigma, ortho)
s[1] = adotr (Memr[wgt], data, npix)
#if (abs (s[1]) <= EPSILONR) {
if (s[1] == 0.0) {
center = x[1]
call sfree (sp)
return (0)
}
}
# Location not bracketed.
if (s[3] * s[1] > 0.0) {
call sfree (sp)
return (-1)
}
# Intialize the quadratic search.
delx = x[1] - x[3]
x[2] = x[3] - s[3] * delx / (s[1] - s[3])
call mkt_prof_derv (Memr[wgt], npix, x[2], sigma, ortho)
s[2] = adotr (Memr[wgt], data, npix)
#if (abs (s[2]) <= EPSILONR) {
if (s[2] == 0.0) {
center = x[2]
call sfree (sp)
return (1)
}
# Search quadratically.
for (iter = 2; iter <= maxiter; iter = iter + 1) {
# Check for completion.
#if (abs (s[2]) <= EPSILONR)
if (s[2] == 0.0)
break
if (abs (x[2] - x[1]) <= tol)
break
if (abs (x[3] - x[2]) <= tol)
break
# Compute new intermediate value.
newx = x[1] + apqzero (x, s)
call mkt_prof_derv (Memr[wgt], npix, newx, sigma, ortho)
news = adotr (Memr[wgt], data, npix)
if (s[1] * s[2] > 0.0) {
s[1] = s[2]
x[1] = x[2]
s[2] = news
x[2] = newx
} else {
s[3] = s[2]
x[3] = x[2]
s[2] = news
x[2] = newx
}
}
# Evaluate the center.
center = x[2]
call sfree (sp)
return (iter)
end
# AP_TPROFDER -- Procedure to estimate the approximating triangle function
# and its derivatives.
procedure ap_tprofder (data, der, npix, center, sigma, ampl)
real data[ARB] # input data
real der[ARB] # derivatives
int npix # number of pixels
real center # center of input Gaussian function
real sigma # sigma of input Gaussian function
real ampl # amplitude
int i
real x, xabs, width
begin
width = sigma * 2.35
do i = 1, npix {
x = (i - center) / width
xabs = abs (x)
if (xabs <= 1.0) {
data[i] = ampl * (1.0 - xabs)
der[i] = x * data[i]
} else {
data[i] = 0.0
der[i] = 0.0
}
}
end
# MKT_PROF_DERV - Make orthogonal profile derivative vector.
procedure mkt_prof_derv (weight, npix, center, sigma, norm)
real weight[ARB] # input weight
int npix # number of pixels
real center # center
real sigma # center
int norm # orthogonalise weight
pointer sp, der
real coef
real asumr(), adotr()
begin
call smark (sp)
call salloc (der, npix, TY_REAL)
# Fetch the weighting function and derivatives.
call ap_tprofder (Memr[der], weight, npix, center, sigma, 1.0)
if (norm == YES) {
# Make orthogonal to level background.
coef = -asumr (weight, npix) / npix
call aaddkr (weight, coef, weight, npix)
coef = -asumr (Memr[der], npix) / npix
call aaddkr (Memr[der], coef, Memr[der], npix)
# Make orthogonal to profile vector.
coef = adotr (Memr[der], Memr[der], npix)
if (coef <= 0.0)
coef = 1.0
else
coef = adotr (weight, Memr[der], npix) / coef
call amulkr (Memr[der], coef, Memr[der], npix)
call asubr (weight, Memr[der], weight, npix)
# Normalize the final vector.
coef = adotr (weight, weight, npix)
if (coef <= 0.0)
coef = 1.0
else
coef = sqrt (1.0 / coef)
call amulkr (weight, coef, weight, npix)
}
call sfree (sp)
end
define QTOL .125
# APQZERO - Solve for the root of a quadratic function defined by three
# points.
real procedure apqzero (x, y)
real x[3]
real y[3]
real a, b, c, det, dx
real x2, x3, y2, y3
begin
# Compute the determinant.
x2 = x[2] - x[1]
x3 = x[3] - x[1]
y2 = y[2] - y[1]
y3 = y[3] - y[1]
det = x2 * x3 * (x2 - x3)
# Compute the shift in x.
#if (abs (det) > 100.0 * EPSILONR) {
if (abs (det) > 0.0) {
a = (x3 * y2 - x2 * y3) / det
b = - (x3 * x3 * y2 - x2 * x2 * y3) / det
c = a * y[1] / (b * b)
if (abs (c) > QTOL)
dx = (-b / (2.0 * a)) * (1.0 - sqrt (1.0 - 4.0 * c))
else
dx = - (y[1] / b) * (1.0 + c)
return (dx)
#} else if (abs (y3) > EPSILONR)
} else if (abs (y3) > 0.0)
return (-y[1] * x3 / y3)
else
return (0.0)
end
|