Image Processing Reference

In-Depth Information

Fig. 11.8
Random inversions

11.1.5 Randomness

It can often be interesting to add some randomness in order to create a visual effect.

This can be done in many different ways and only the imagination of the designer

sets the limit. The output will be different from time to time even though the input

in the same. This adds a nice uniqueness to visual effects involving randomness.

Below we describe one concrete example, namely an algorithm denoted
random

inversions
.

The algorithm is based on the idea of generating a random binary pattern and

then using this to apply image inversion locally. First the output image is divided

into squares of equal size
S

S
. For each square we draw a random number between

zero and one. If this number is above 0.5 then the square is set to white; otherwise

to black. This will result in an intermediate output like the one in Fig.
11.8
(a). In the

next step the intermediate output in blurred by a mean filter in order to obtain softer

shapes, see Fig.
11.8
(b). The kernel size is equal to
S/
2. The blurred squares are

now thresholded, using a threshold value of 128, to ensure the edges of the shapes

are sharp, see Fig.
11.8
(c). For each white pixel we now invert the corresponding

RGB pixel in the input and place that in the output. For each black pixel we simply

×