from multitemporal acquired data. Point spread function is
reconstructed using (5x5) matrix data.
4.3) Modulation Transfer Function:
Modulation Transfer function of imaging system is obtained as
the modules of the discrete Fourier Transform of the PSF. This
paper has reported MTF computed from reconstructed PSF
using (5x5) matrix data arranged for sample-scene-phase
analysis. The edge function is derived from uniform High/Low
contrast target by successive pixel Grey count difference or
derivative of edge function gives Line Spread Function (LSF).
The normalized Fourier transform of LSF also gives MTF of
imaging system, which depends on choice of edge function.
4.4) Square Wave Response:
The sensor's spectral/spatial response is measured in laboratory
from series of equally spaced bright-dark bars at various
Nyquist frequencies. Fhe width of bar is an integral multiple of
GIFOV of the sensor. The spectral response of the sensor is an
average constant over an effective spectral band and spatial
response is an average constant over an effective GIFOV. The
total system Square wave response characterizes radiance-
recording capability for given target contrast & size.
4.5) Radiance at Sensor:
Radiance at sensor level is a sensor response function, which is
a combined effect of weighted integrated effect of spectral &
spatial response. A linear relationship between sampled &
quantized Digital number and at sensor radiance exists. The
inversion technique to obtain band radiance values from image
DN is known as “Sensor Calibration”. This sensor calibration is
further evaluated by removing atmospheric effect, which
converts radiance to reflectance as a “Scene Calibration”. The
deployed Artificial Target covers 80% of available sensor
dynamic range. Simultaneously measured scene target radiance
& digital numbers are compared.
4.6) Ground Sampling Distance:
For push broom scanning the cross track Ground Sampling
Interval does not vary across the linear detector elements, it is
only a function of height & focal length of sensor. The along
track Ground Sampling Interval in Scan line directional is a
equivalent to spatial movement during the integration time of
detector array and platform velocity. The variation of
Integration time & platform velocity gives combined effect as
Image motion blurs which effects shape & size of target. Along
and across Track ground sampling distance is computed on
Test site. Four Corners of site are well identifiable on image,
which is precisely measured on ground using Differential GPS
mode data.
The other important imaging parameters like
Detectability & Adjacency effect is a function of type of target,
reflectivity and size with respect to surrounding background
reflectance.
5.0 ANALYSIS AND DISCUSSION OF RESULTS
This section describes the procedure of computing and
analyzing results of various parameters listed in preceding
section. :
5.1 Detectability:
It characterizes the sensor’s ability to distinguish the presence
of target from recorded signal. The experiment has been
conducted to detect the object at sub-pixel level with target
reflectance of 35% and background soil with 17% reflectance.
IAPRS & SIS, Vol.34, Part 7, “Resource and Environmental Monitoring”, Hyderabad, India,2002
Three Gray targets of size 0.5 pixel, 0.75 pixel and 1.0 pixel
and One Gray linear target of *T' shape of width 0.5 pixel were
deployed to check the Detectability of the sensor. The linear
“T’ target with width 0.5 of pixel was clearly detectable
although the remaining three merged with background
(Ref.Figla,1b). This can be attributed to 1) the scene-sample
phase 2) the aspect ratio, which is the ratio of length to width
of the sample and 3) the adjacency effect discussed in next
section.
The single target with sub-pixel and one-pixel width were
averaged out from all the four sides by the background whereas
for the “T” target with only 0.5-pixel width, the contamination
due to background was limited only to two sides.
5.2 Adjacency Effect:
The study of adjacency effect signifies sensor's ability for
target recognization from surrounding background. Three-pixel
by three-pixel black target with 4% reflectance was placed in
the center of a large Gray target of size 35m x 35m of 35%
reflectance. Another three-pixel by three-pixel gray target was
also placed in the Bare Soil background with 17% reflectance.
Typical Dump of Pure Black Target Typical Dump of Pure Gray Target
61 64 62 63 62 60 113 113 114 112 112: 112
63 64 65 63 62 60 115 115 114 115 112 112
63 64 62 60 64 63 115 113 114 115 114 112
63 62 62 60 62 63 115 113 114 112 116.112
63 62 62 60 62 65 115 113 116 115 114 114
63 64 62 63 64 63 113 113 114 115 112" 112
Typical Dump of Pure Gray Target Typical Dump of Pure Bare Soil Target
113 113 114 112 112 112 78 79 76 78 78 80
115 115 114 115 112 112 80 79 ." 79 78 78 78
115 113 114 115 114 112 80 77 79 80 78 78
115 113 114 112 116 112 75 77 76 80 78 78
115 113 116 115 114 114 80 79 76 78 78 78
113 113 114 115 112 112 75 77 76 80 76 78
3 x 3 Black target in Gray background 3 x 3 Gray target in Bare Soil background
115 116 115 113 114 116| 79 79 78 79 77 77
115 107 102 101 107. 114) 79 86 89 88 82 77
113 103 86 86 98 112] 79 90 100 99 91 80
115 103 83 86 98 — 112] 81 90 100 101 93 80
117 112 106 101 105 114] 79 88 96 96 86 80
117 116 117 117 114 116| 77 79 80 83 80 77
117 118 121 117 120 118] 77 77 78 79 77 77
Scan
Table:3 Effect of ad jacency on a 3 by 3 pixel Target at two
Different Contrast Ratios
The three
at least
fully cov
The dun
target is
by three
when plz
average :
pixel by
increase
contamir
the 3pix:
the strer
configur;
Date
| L1-Dec-0
Clear
Cloudy
17-Apr-0
Clear
Clear
Table 4:
The dun
backgrot
to a cour
bare soil
count by
radiomet
target re
of 35%.
to 2596 f
Consider
target, c
compute
contribui
that fron
elevatior
shadows
effect w
tank for
between
SAC Re
targets |
experime
experime
target as
backgrot
5.3 Squz
The squ@
sixteen «
two, thr
track dir
The Fig?
April 20
four freq
The SW]
SWR = (
Due to |
DNmin
DNmax |