60
2: Raghuveer M. Rao, Manoj K. Arora
15
10
5
o
50
100
150
200
250
a
b
15
10
5
o
50
100
150
200
250
c
d
Fig. 2.4. a Original image. b Histogram. c Histogram -equalized image. d Equalized histogram
a cumulative distribution function Fx{x). The random variable can be transformed to a new random variable y that has a uniform PDF over 0 to 1 through
the transformation
y = FAx) .
(2.5)
Thus, histogram equalization is a linearization in some sense of the cumulative distribution function. Because digital images are not continuous-valued,
only an approximation to this process is implemented. Consequently, the resulting histogram is rarely perfectly uniform. Histogram equalization is most
effective when the histogram of the input image is not dominated by one
prominent gray level to the exclusion of others.
2: Raghuveer M. Rao, Manoj K. Arora
15
10
5
o
50
100
150
200
250
a
b
15
10
5
o
50
100
150
200
250
c
d
Fig. 2.4. a Original image. b Histogram. c Histogram -equalized image. d Equalized histogram
a cumulative distribution function Fx{x). The random variable can be transformed to a new random variable y that has a uniform PDF over 0 to 1 through
the transformation
y = FAx) .
(2.5)
Thus, histogram equalization is a linearization in some sense of the cumulative distribution function. Because digital images are not continuous-valued,
only an approximation to this process is implemented. Consequently, the resulting histogram is rarely perfectly uniform. Histogram equalization is most
effective when the histogram of the input image is not dominated by one
prominent gray level to the exclusion of others.