and 1.0 pixel
.5 pixel were
r. The linear
y detectable
background
'cene-sample
1gth to width
ssed in next
width were
'und whereas
ontamination
| ability for
Three-pixel
as placed in
5m of 35%
y target was
eflectance.
e Gray Target
112 112::3112
115 112" 112
115 114 112
112 116. 112
115 114 114
Hs" 112 112
lare Soil Target
8 78 80
8 78 78
0 78 78
0 78 78
3 78 78
) 76 78
Soil background
) 77 77
) 82 77
! 91 80
1 93 80
86 80
80 77
77 77
—
—_
i
NM
et at two
IAPRS & SIS, Vol.34, Part 7, "Resource and Environmental Monitoring", Hyderabad, India,2002
The three- pixel by three-pixel size was selected to ensure that,
at least theoretically, one pixel corresponds to one detector
fully covered by the center image target.
The dump of the image, in 8-bit quantization, for respective
target is tabulated in Table 3. One pixel out of the three-pixel
by three-pixel black target generated a minimum of 83 count
when placed in a background of large pure gray target with an
average count of 114. The larger pure black target, of size 39
pixel by 39 pixel, shows an average count of 63 (Table 3). This
increase in black target count from 63 to 83 is because of the
contamination from the relatively bright background in which
the 3pixel x 3pixel black target is placed. The Table 4 shows
the strength of adjacency on four dates for both the target
configurations.
direction. The SWR thus obtained was normalized for the
ground-measured reflectance of the targets. Fig5b shows SWR
at various configurations for 17Th April 2002. Table 6 shows
the SWR obtained for all dates and all combinations.
Response of the sensor for Along Scan High contrast SWI target ot various frequenc ies
ZO
“ow ow wou ow
Line Number
Along and Across scan SWR at Two Centrasts for Four frequencies over Chharodi Calibration Site
0.45
1
75m
/
/
325
o0 0.13 0.17 0.25 0.50
Frequency (Cyc/Mix)
Fig: 5a,b SWR target response and SWR at various frequencies
Square Wave Response
Date Gray in |Contribution| Black [Contribution] View
Bare Soil | Of Bare Soil | in Gray [Of Gray Angle
11-Dec-01| 4 -0.16 20 0.175 7.889
Clear
7-Apr-02 23 -0.23 20 0.145 10.234
Cloudy >
17-Apr-02 17 -0.14 32 0.204 0.14
Clear
24-Apr-
24-Apr-02 18 -0.14 34 0.205 0.128
Clear
Table 4: Multi-temporal Adjacency Effect Analysis
The dump of 3 pixel by 3 pixel gray target in bare soil
background shows a decrease in pure Gray target count of 114
to a count of 101 (Table 3). The relatively dark background of
bare soil with average count of 80 reduces the Gray target
count by 13. In terms of reflectance, TES PAN sensor of 7bit
radiometric quantization with Imtr GIFOV has increased pure
target reflectance of 4% to 15% from background reflectance
of 35%. Similarly pure target of 35% reflectance has decreased
to 25% from background reflectance of 17%.
Considering that one detector was fully covered by center
target, contribution from the neighboring background was
computed. Except for the 7Th April data, the relative
contribution from the Gray background has increased whereas
that from darker background has decreased with increasing sun
elevation angle. The sky was cloudy on 7Th April with
shadows casting over the site intermittently. The adjacency
effect was studied previously on natural target Kurukshetra
tank for IRS sensor and similar trend was observed at edges
between water and Pavement. The controlled experiment on
SAC Recreation Club Ground was conducted with artificial
targets made of Black and White cloth. The systematic
experiment at Chharodi has quantified the behavior. This
experiment has scope to extend observations for size of pure
target as a function of a) radiometric quantization b) Target to
background contrast ratio.
5.3 Square Wave Response:
The square wave response (SWR) targets were deployed with
sixteen configurations in all. Two and a half cycles of one,
two, three and four pixel widths each for along and across
track direction were designed at two contrasts.
The Fig5a shows sensor response to the SWR pattern for 17Th
April 2002. The count variations for high and low contrasts at
four frequencies can be clearly seen.
The SWR was computed using the following equation:
SWR - (DNmax-DNmin) / (DNmax + DNmin)
Due to the heading angle of the spacecraft, the DNmax and
DNmin appeared in separate scan lines or pixel columns, the
DNmax and DNmin were selected from same scan line or pixel
At High Contrast At Low Contrast
Date Along Scan|Across Scan| Along Scan | Across Scan
0.13 Cyc/Pix
11-Dec-01 0.30 0.21 0.22 0.14
07-Apr-02 0.28 0.37 0.29 0.38
17-Apr-02 0.38 0.41 0.41 0.39
24-Apr-02 0.30 0.31 0.31 0.32
0.17 Cyc/Pix
11-Dec-01 0.26 0.20 0.19 0.11
07-Apr-02 027 — 0.35 0.25 0.30
17-Apr-02 0.37 0.39 0.32 0.35
24-Apr-02 0.28 0.31 0.23 0.30
0.25 Cyc/Pix
11-Dec-01 0.15 0.16 0.14 0.09
07-Apr-02 0.19 0.33 0.2 0.28
17-Apr-02 0.24 0.32 0.22 0.29
24-Apr-02 0.16 0.23 0.15 0.25
0.5 Cyc/Pix
11-Dec-01 0.08 0.11 0.04 0.05
07-Apr-02 0.14 0.23 0.13 0.25
17-Apr-02 0.14 0.27 0.12 0.25
24-Apr-02 0.11 0.19 0.10 0.19 -
Table:6 Multi-temporal Square Wave Response
5.4 Reconstructed Point Spread Function
One arrangement of trying to get a pure pixel as described in
earlier section was to place a 3x3-pixel target. A 5x5-pixel
matrix of 1 pixel size target with an incremental quarter pixel
displacement was made (Fig8). Square window of 5x5 pixels
are selected around individual PSFs, which are re-arranged in
One-pixel co-ordinate (Fig 10).
5.5 Modulation Transfer function: The 2-dimensional PSF
thus obtained was Fast Fourier Transformed to get the MTF
(Table7). The Count of reconstructed point and the one obtained
by the 3x3 matrix can be compared (see Table9,Fig 10).
Date Along Scan Across Scan
11-Dec-01 14.2 14.8
07-Apr-02 11.1 14.3
17-Apr-02 14.28 16.6
24-Apr-02 PSF could not be re-constructed
