How can an image be "noisy?"
Noise, in an imaging sense, means that there are pixels which vary widely
from the surrounding pixels. This creates a grainy look in an image which
is generally undesirable.
Is noise ever desirable?
Yes, sometimes you would want noise in an image. For one thing, noise can
help an image to look more realistic. This is especially true in the case
of computer generated images, which too often have a surreal look because
of the mathematical perfection that they represent. A slight amount of
noise in such a picture can make a profound difference.
Most of the time, however, you want to remove noise. This is accomplished
by looking at all of the pixels in a given rectangle, determining which ones
are most different than the ones surrounding them, and blending them in. It
is very similar to blurring using a Gaussian convolution
The image in figure 1.1 is an image without noise, and the image in
figure 1.2 is an image that has had noise added through the use of our
noise applet. The ripples in figure 1.2 look more real, because they are
less perfect. Please note that the difference may be hard to see if your
browser dithers either image.
How do you add noise?
The algorithm to add noise is the most simple of all the effects which we covered.
Random red, green, and blue values are chosen, and then simply added to the existing
values. For more noise, simply increase the variance (multiply the random values by
a higher number).
But don't take our word for it, try it for yourself:
Return to main page.