Page 242 - Digital Analysis of Remotely Sensed Imagery

206 Cha pte r S i x

255
f

DN
out

θ
0
0 255
DN
in
FIGURE 6.2 Relationship between a pixel’s value in the input image DN and
in
its output value in the enhanced image DN after it has been modifi ed via
out
the function f.

contrast than the original image. If tanq < 1, then the contrast of the
output image is suppressed. This enhancement function is mathe-
matically expressed as

DN − DN
1
DN = in min × (2 n − ) (6.2)
out −
DN DN
max min
where DN = the largest DN in the input image
max
DN = the smallest DN in the initial image
min
n = quantization level of the stretched image; it usually
has a value of 8, even though a larger value is
possible with the recent generation of satellite data
The degree of stretching depends on the DN range (DN −
max
DN ) of the initial image. It is measured by the stretching ratio that
min
is defined as
2 − 1
n
Stretching ratio = (6.3)
DN − DN
max min
Apparently, an input image with a narrow range of pixel values
has a larger stretching ratio than that of a broader range.
In linear stretching, the disparity between any two adjacent gray
levels is enlarged proportionally irrespective of their actual value. In
the output image, the disparity between any two adjacent gray levels
is always constant (Fig. 6.3c).

Example What output value should a pixel receive if it has a value of 112 in an
input image in which DN ranges from 48 to 132 (assume the output image is
recorded at 8 bits)?