KEY WORDS: CCD Star Images, image co-ordinates, astro-geodetic co-ordinates, precise geoids, psf fitting, image moment
analysis, centroid estimation, sub-pixel accuracy
ABSTRACT:
The new microchip and sensor technology, charged couple device (CCD), has assumed a permanent position as the natural
transducer for optical input to a computer. The CCD combined with the microcomputer have revolutionised the whole
discipline of observational astronomy at all stages from data gathering to data analysis, presentation and use. Not only has it
brought the convenience of digital imaging to field astronomy but has also great potential in determining very accurate
astronomic co-ordinates of a point at great speed and ease, compatible to GPS for the determination of deflection of the
vertical and local or precise geoids. Despite the numerous advantages of the CCD in astronomical work most CCD users
agree that individual devices have imperfections or various problems that have to be dealt with. In using CCD in geodetic
astronomy one major problem that has to be considered is whether the measuring accuracy of the image co-ordinates will be
enough for the determination of astronomic latitude or longitude. Just clicking on a star object in a digital image to determine
its position in the image yields pixel accuracy. But to meet geodetic accuracies, sub-pixel accuracies are required. In this
work we employed least squares smoothing techniques on data obtained from astronomical observations using CCD zenith
camera. The star image co-ordinates were estimated using the two main methods of centroid estimation, viz., moment
analysis and PSF fitting. This paper compares these two methods of estimating image co-ordinates to sub-pixel level and
their accuracies. Results so far indicate that star image measurement accuracies better than 0.3 arcsecond can be obtained.
They also show that PSF fitting method is more adaptive to automation.
1.
Introduction
Geometric Features
L = d 1
(1)
PERIMETER:
the side lengths.
P=
side lengths
(2)
x B( x , y ) B00
B10 =
y B( x , y ) B00
B01 =
AREA (A):
This is equal to the sum of all the pixels
covered by the object. That is, area of an object in a digital
image is the number of points in the object. Thus we can
compute the area of the object by simply scamnning these
points
A = total number of pixels
(3)
Principles of Centroiding
C
KL
x direction.
ORIENTATION ( ): This is defined as the angle of axis of
the least moment of inertia. It determines how the object lies
in the field of view (Horn 1986 pp. 49). The orientation of an
image is given by (ibid. pp. 54):
tan( 2 ) =
2 B11
B20 B02
IMAGE MOMENTS
BKL =
x K y L B( x, y)
x
B00 . We find
sin(2 ) =
cos(2 ) =
2 B11
PSF Fitting
(9)
(10)
r
is the radius with respect to the centre of
the star image
ln I ( r ) = ln I ( 0) + br 2
(11)
.
The equation above is of the form Y = A + BX. The
coefficients (A,B) can be estimated by linear regression from
(X,Y) pairs.
S X S XY S XX SY
S X2 nS XX
S S nS XY
B = X2 Y
S X nS XX
A=
S XY =
i =1
n
i =1
X i Yi
(17)
(12)
3.
n
i =1
X i2 ; SY =
n
i =1
Yi
ri 2 = ( xi x 0 ) + ( yi y 0 )
2
(13)
ln I ( ri ) ln I (0) = b ( x i x 0 ) + ( y i y 0 )
2
(14)
E i = c0 + c1 xi + c2 yi + c3 xi2 + c4 xi2
ImageCo-ordinate Measurements
and
X i ; S XX =
c1
2 c3
c
y0 = 2
2 c4
x0 =
where:
(16)
r2
I ( r ) = I (0) exp 2
2
SX =
E
= c1 + 2c3 x = 0
x
E
= c2 + 2 c4 y = 0
y
(15)
Ic( x , y , t ) = K
Io( x , y , t ) N ( x , y , t )
F ( x , y , t ) N ( x , y , t )
(18)
where
Figure 1: CCD250_2 a 2.5 sec. exposure image after calibration and median filtering
total columns
4.1
, y mj
Computations
xj
ai
=
yj
bi
bi
xij'
ai
yij'
bi
xijc
ai
c
ij
ei
fi
(20)
Similarly we have
x mj
m
j
ai
bi
ei
fi
(21)
or
xijc
It is pertinent to note that the images were taken
independently and secondly the stars were in motion. In
order to analyse these images they were registered or
transformed to the same datum using the two-dimensional
similarity transformation. The transformation also gives
m
j
(19)
row measured
(x
new co-ordinate system.These serve as approximate coordinates in the least squares computations described
below.
Analysis of Results
y sensorco ordinate=
mean positions
yijc
ai
bi
x mj ei
bi
ai
y mj f i
(22)
since
is
b a
orthogonal.
(x
c
ij
,y
c
ij
are
ATWAdX = ATWb
xj
x mj + dx
= m
yj
x j + dy
Let
yij' yijc
ai
bi
bi
ai
dx j
dy j
(25)
AdX
where
(24)
L L
(27)
(23)
xij' xijc
AdX ( L L ) AdX b
2
0
= v Wv
(28)
nu
u = 2 p .
v
is vector of residuals
dX
is vector of corrections or unknowns
A
is the design matrix
L, Lc
are measured and computed image
co-odinates
b
is the vector of reduced measurements
( " observed computed " )
xx
= 20 Q
(29)
xx
C C D 250_1
C C D 250_2
C C D 250_3
CC D 250_4
0 .9 9 9 9 8 9
1 .0 0 0 7 6 4
1 .0 0 1 0 9 5
1 .0 0 1 1 4 7
0 .0 0 0 0 0 0
0 .0 0 0 6 5 0
0 .0 0 1 2 6 9
0 .0 0 1 2 5 3
0 .0 0 2 4
6 3 .1 4 1 6
1 2 5 .7 4 7 6
1 8 6 .3 8 6 7
0 .0 0 0 8
1 .2 1 4 6
2 .1 1 3 8
3 .0 4 4 3
N .
a p p r o x im a te c o - o r d in a te s
x
y
- 8 9 .5 0 8 5 0 4
1 6 1 .7 0 2 2 3 5
7 7 5 .3 7 7 3 6 7
- 2 5 0 .0 6 7 4 8 8
- 4 6 .6 0 3 0 5 6
- 1 0 7 .2 2 6 8 5 3
- 1 8 1 .7 0 9 2 2 6
- 2 5 4 .3 6 2 8 8 7
1 0 0 0 .7 5 4 8 3 5
- 1 9 3 .5 7 6 4 5 9
2 2 .0 8 3 4 6 7
- 1 9 .7 0 1 7 5 9
3 1 4 .4 3 0 0 9 0
- 1 5 2 .7 4 4 9 4 5
- 2 3 1 .2 9 5 0 3 6
1 1 3 .9 5 1 1 2 3
- 5 7 1 .9 8 8 8 3 0
3 3 5 .3 0 2 4 5 4
1 5 8 .2 0 2 0 9 7
- 4 4 7 .4 7 7 2 6 2
11
2 1 3 .4 7 3 2 6 1
3 9 7 .9 1 9 0 3 0
12
- 3 0 8 .9 0 8 8 3 1
- 2 5 9 .8 3 4 3 2 6
13
7 3 4 .4 7 0 8 8 2
- 1 5 1 .2 9 0 1 3 4
10
1
2
3
4
5
6
7
8
9
10
11
12
13
- 8 9 .2 2 5 9 7
7 7 4 .7 4 0 1 3
- 4 6 .5 5 5 3 0 6
- 1 8 1 .1 3 8 9 1
1 0 0 0 .1 1 3 6 8
2 2 .1 2 9 9 8 5
3 1 4 .3 7 6 7 2 8
- 2 3 0 .5 4 8 7 3
- 5 7 1 .3 0 5 7 0 9
1 5 8 .1 9 8 3 9 7
2 1 3 .4 3 2 4 1 9
- 3 0 8 .4 6 8 7 8 3
7 3 3 .7 2 8 2 2 1
std . e rro r
in x
1 6 1 .6 1 3 8 5 1
0 .0 7 8 8 5 4
- 2 4 9 .9 7 4 2 0 4
0 .1 4 4 4
- 1 0 7 .1 6 9 8 1 1
0 .1 3 4 9 6 6
- 2 5 4 .1 2 2 3 1 1
0 .1 0 4 1 0 6
- 1 9 3 .2 4 5 0 4
0 .5 9 7 2 2 8
- 1 9 .7 4 4 2 2 2
0 .0 8 0 5 1 6
- 1 5 2 .7 2 9 3 2
0 .2 6 2 6 2 3
1 1 3 .8 9 4 4 8 3
0 .0 8 2 1 3 2
3 3 4 .9 1 6 2 4 1
0 .1 6 8 5 3 9
- 4 4 6 .8 1 1 0 7 5
0 .1 4 9 1 7 4
3 9 7 .5 6 1 9 4 9
0 .2 5 5 6 3
- 2 5 9 .4 9 2 9 3
0 .1 7 1 3 6 9
- 1 5 1 .2 9 2 1 4
0 .5 7 2 6 5 4
m e a n s td . e r r o r i n x : 0 .2 1 5 5 5 3
m e a n s td . e r r o r i n y : 0 .1 3 6 1 1 4
s td . e r r o r
in y
0 .1 0 0 7 4 6
0 .0 5 8 4 6 7
0 .0 7 0 1 9 3
0 .0 7 8 8 3 8
0 .0 7 8 4 6 5
0 .1 2 0 4 5 2
0 .1 1 1 3 5 7
0 .0 7 4 8 7 9
0 .2 3 7 8 4 9
0 .0 5 7 9 5 1
0 .4 6 7 4 5
0 .2 3 1 9 0 6
0 .0 8 0 9 3 7
Exp.time
in secs.
No. of
common
objects
CCDF100 11.11.96
CCDF201
12
CCDF212 23.02.97
12
CCDF302
23.02.97
CCDF303
23.02.97
12
10
10.02.97
CCDF116* 23.02.97
1.6
CCDF224* 23.02.97
2.4
CCDF225 23.02.97
2.5
CCDF228 23.02.97
2.8
0.25
0.20
0.15
1.08
1.04
1.04
1.02
1.05
2.94
1.02
0.18
0.13
0.32
0.29
0.12
0.10
0.16
0.13
0.19
0.10
0.44
0.43
0.89
0.35
0.84
0.34
0.37
0.33
B
C
0.40
0.35
0.30
1.03
1.04
B
C
0.40
Unit
mean standard errors
variance
(in pixels)
o2
x
y
0.35
0.30
0.25
0.20
0.15
0.10
1.0
1.5
2.0
2.5
3.0
1.0
1.5
2.0
2.5
3.0
Table 5: Precision of centroid determinations from moment analysis and PSF fitting metho
Time of
Exp.
observation time
(UTC)
(secs)
1
2
3
4
5
CCD101*
CCD120*
CCD150
CCD180*
CCD181
21:45:59
21:47:59
21:57:59
21:59:54
22:01:19
6
7
8
CCD200*
CCD201*
22:02:32
22:03:59
9
10
CCD220
CCD221
CCD250
11
1
1.2
1.5
1.8
1.8
0.99
1.23
1.03
1.04
1.01
0.47
0..56
0.27
0.37
0.24
0.46
0..53
0.29
0.40
0.24
1.03
1.09
1.10
1.04
1.01
0.32
0.40
0.09
0.34
0.21
0.39
0.37
0.15
0.24
0.18
2
2
1.27
1.48
0.61
0.45
0.60
0.43
1.04
1.09
0.43
0.17
0.42
0.21
22:09:51
22:10:58
22:13:34
2.2
2.2
2.5
0.98
0.99
1.10
0.21
0.23
0.19
0.11
0.23
0.23
1.05
1.02
0.25
0.19
0.14
CCD251
22:12:11
2.5
1.05
0.14
0.14
1.05
12
CCD280
22:13:11
2.8
1.05
0.17
0.10
13
14
15
CCD281
CCD301
22:16:19
22:18:33
2.8
3
1.04
1.02
0.23
0.18
0.15
0.13
1.01
1.04
0.20
0.36
0.16
0.29
CCD253
CCD103
CCD153
22:19:58
22:24:06
22:28:17
2.5
1
1.5
1.09
0.99
1.03
0.23
0.23
0.18
1.03
1.03
0.99
0.26
0.16
0.22
0.18
0.10
0.18
16
17
Conclusion
0.19
0.23
0.22
0.22
0.14
0.11
-
2.
3.
4.
5.
camera
system.
Proccediings
of
the
International Symposium on Navigation (KISS
94) Calgary Canada.
6.
11.
7.
8.
9.
10.
13.
14.
15.