You are on page 1of 7

Image Enhancement And Point Processing Operations

Image enhancement
Image enhancement means enhancing or modifying the image depending on
application. It is called a subjective process because the output of image enhancement
changes from person to person. It is also a cosmetic process because no extra
information is added during image enhancement.

Consider an image f(x,y) when we apply transformation T on the original image f(x,y) we
get a new image represented by g(x,y). This new image is called as modified image or
enhanced image.
Image enhancement is performed into domains
1. Spatial domain
2. Frequency domain

Image enhancement in spatial domain


Images can be enhanced either by using point processing or neighborhood processing.
In point processing to process a single point we can apply transformations directly over
it. Where as in neighborhood processing to process a single point we take help of its
neighbors.

Point processing or zero memory operations:


Identity transformation:

Horizontal axis represents the original gray-level denoted by r. The vertical Axis
represents modified grade level denoted by s. We are using r and S just for sake of

Digital Image Processing: Operations and Applications


Image Enhancement And Point Processing Operations
simplicity, but there is no rule that we must use r an s only you can use other variables
instead of r and s.
The solid line between r and s is the transformation T. the grey level 21 horizontal Axis
will become 20 after modification as shown in vertical axis. That is an any value in the
range 0 to 255 on horizontal Axis will remain same on vertical Axis after modification.
This is called identity transformation and the curve is called as identity transformation
curve.

1. Digital Negative

It means inverting grey level values that is black in original image becomes white
after modification and vice versa. Digital negative is used in digital photography.
It is given by expression s = (L-1)* r

If image is of 8 bits then, 2^8=256= L


Hence s = 255 - r

Case 1
r=0
s=255-0
s=255

Case 2
r=255
s=255-255

Digital Image Processing: Operations and Applications


Image Enhancement And Point Processing Operations
s=0

s= (L-1)-r
s=255

2. Contrast stretching
It means making dark pixels more darker and bright pixels more brighter by
assigning different slopes. We make bright pixels more brighter by assigning
slope m>1 and l and n<1.

s=l*r ……………………….when r is in the range of 0 to a


s=m(r-a)+v ………………………..when r is in the range of a to b
s=n(r-b)+w ………………………..when r is in the range of b to L-1

3. Thresholding
It means extreme contrast stretching. In this method we compare the pixel values
with threshold value and if it is less than or equal to threshold we make it black
and if it is greater than threshold we make it white as shown in figure.

Digital Image Processing: Operations and Applications


Image Enhancement And Point Processing Operations

s=0 ….r <= a


s=L-1 …..r > a

4. Grey level slicing


The output of thresholding is either Black or white but if we want some dresses of
grey color in an image, it is not possible using thresholding. Hence we use
technique called as grey level slicing. It comes into two forms

Without background

s = L-1 ….When image is in the range of a to b


s=0 ….otherwise

With background

Digital Image Processing: Operations and Applications


Image Enhancement And Point Processing Operations

s = L-1 ….When image is in the range of a to b


s=r ….otherwise

5. Bit plane slicing


It is a concept which is used in steganography. Steganography is the technique
of hiding secret data within an ordinary, non-secret, file or message in order to
avoid detection; the secret data is then extracted at its destination. The use
of steganography can be combined with encryption as an extra step for hiding
or protecting data. Here we take an image and convert it into its equivalent
binary for example consider

1 6 1

5 2 0

7 4 2

Convert it into equivalent binary

001 110 001

101 010 000

111 100 010

Digital Image Processing: Operations and Applications


Image Enhancement And Point Processing Operations
For simplicity we present image using 3 bits of binary, but in reality images represented
using 8 bits. So in the above example since there are three bits of binary we have three
planes, LSB plane, MID plane, MSB plane M

MSB
0 1 0

1 0 0

1 1 0

MID
0 1 0

0 1 0

1 0 1

LSB
1 0 1

1 0 0

1 0 0

The visually significant information is present in MSB plane, whereas visually least
significant information is present in LSB plane

6. Dynamic range compression or log transformation


We cannot see stars in the sky during the daytime because of very high intensity
of sun. Similarly in many images if we want to see the low intensity pixel values
which were not seen because of some higher intensity pixels, then the solution is
to compress high intensity pixel values so that we are able to see low intensity
pixels this is called dynamic range compression.

Digital Image Processing: Operations and Applications


Image Enhancement And Point Processing Operations

s = c * log(1+r)

7. Power law transform / gamma(γ) correction


As their birth defects in TV monitors of olden days, this technique was used to
rectify these defects by applying different values of Gamma. Defects but termed
as Gamma linearity’s or noise.

s = c. r γ

Digital Image Processing: Operations and Applications

You might also like