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
|
.help imstatistics Feb01 images.imutil
.ih
NAME
imstatistics -- compute and print image pixel statistics
.ih
USAGE
imstatistics images
.ih
PARAMETERS
.ls images
The input images or image sections for which pixel statistics are to be
computed.
.le
.ls fields = "image,npix,mean,stddev,min,max"
The statistical quantities to be computed and printed.
.le
.ls lower = INDEF
The minimum good data limit. All pixels are above the default value of INDEF.
.le
.ls upper = INDEF
The maximum good data limit. All pixels are above the default value of INDEF.
.le
.ls nclip = 0
The maximum number of iterative clipping cycles. By default no clipping is
performed.
.le
.ls lsigma = 3.0
The low side clipping factor in sigma.
.le
.ls usigma = 3.0
The high side clipping factor in sigma.
.le
.ls binwidth = 0.1
The width of the histogram bins used for computing the midpoint (estimate
of the median) and the mode.
The units are in sigma.
.le
.ls format = yes
Label the output columns and print the result in fixed format. If format
is "no" no column labels are printed and the output is in free format.
.le
.ls cache = no
Cache the image data in memory ? This can increase the efficiency of the
task if nclip > 0 or either of the midpt and mode statistics are computed.
.le
.ih
DESCRIPTION
The statistical quantities specified by the parameter \fIfields\fR are
computed and printed for each image in the list specified by \fIimages\fR.
The results are printed in tabular form with the fields listed in the order
they are specified in the fields parameter. The available fields are the
following.
.nf
image - the image name
npix - the number of pixels used to do the statistics
mean - the mean of the pixel distribution
midpt - estimate of the median of the pixel distribution
mode - the mode of the pixel distribution
stddev - the standard deviation of the pixel distribution
skew - the skew of the pixel distribution
kurtosis - the kurtosis of the pixel distribution
min - the minimum pixel value
max - the maximum pixel value
.fi
The mean, standard deviation, skew, kurtosis, min and max are computed in a
single pass through the image using the expressions listed below.
Only the quantities selected by the fields parameter are actually computed.
.nf
mean = sum (x1,...,xN) / N
y = x - mean
variance = sum (y1 ** 2,...,yN ** 2) / (N-1)
stddev = sqrt (variance)
skew = sum ((y1 / stddev) ** 3,...,(yN / stddev) ** 3) / (N-1)
kurtosis = sum ((y1 / stddev) ** 4,...,(yN / stddev) ** 4) / (N-1) - 3
.fi
The midpoint and mode are computed in two passes through the image. In the
first pass the standard deviation of the pixels is calculated and used
with the \fIbinwidth\fR parameter to compute the resolution of the data
histogram. The midpoint is estimated by integrating the histogram and
computing by interpolation the data value at which exactly half the
pixels are below that data value and half are above it. The mode is
computed by locating the maximum of the data histogram and fitting the
peak by parabolic interpolation.
.ih
EXAMPLES
1. To find the number of pixels, mean, standard deviation and the minimum
and maximum pixel value of a bias region in an image.
.nf
cl> imstat flat*[*,1]
# IMAGE NPIX MEAN STDDEV MIN MAX
flat1[*,1] 800 999.5 14.09 941. 1062.
flat2[*,1] 800 999.4 28.87 918. 1413.
.fi
The string "flat*" uses a wildcard to select all images beginning with the
word flat. The string "[*,1]" is an image section selecting row 1.
2. Compute the mean, midpoint, mode and standard deviation of a pixel
distribution.
.nf
cl> imstat m51 fields="image,mean,midpt,mode,stddev"
# IMAGE PIXELS MEAN MIDPT MODE STDDEV
M51 262144 108.3 88.75 49.4 131.3
.fi
.ih
BUGS
When using a very large number of pixels the accumulation of the sums
of the pixel values to the various powers may
encounter roundoff error. This is significant when the true standard
deviation is small compared to the mean.
.ih
SEE ALSO
.endhelp
|
