Another very popular image processing task in which blurring is used, is removing noise from images. Images can be distorted because of various reasons such as, for example, from their scanning process, where the film grain adds unwanted noise, but the scanner could also introduce noise, or the photograph to be scanned might have aesthetic marks on it (such as scratches). Furthermore, even digital photographs may have noise in them, for example, due to their CCD detectors. Transmitting images over electronic mediums may also corrupt them, leading to a noisy result. Many types of additive noise have been implemented in the Image Processing Toolbox of MATLAB and they can be used to simulate some of the aforementioned image corruptions. The function that is used for adding noise to an image is called imnoise. Its usage can be explored using help. Let's see the first lines of the result:

>> help imnoise

The output of the preceding command is as follows:

imnoise Add noise to image.
    J = imnoise(I,TYPE,...) Add noise of a given TYPE to the intensity image
    I. TYPE is a string that can have one of these values:
       'gaussian' Gaussian white noise with constant mean and variance
       'localvar' Zero-mean Gaussian white noise with an intensity-dependent variance
       'poisson'  Poisson noise
       'salt & pepper'  "On and Off" pixels
       'speckle'  Multiplicative noise
    Depending on TYPE, you can specify additional parameters to imnoise. All
    numerical parameters are normalized; they correspond to operations with
    images with intensities ranging from 0 to 1.

This means that we can add five different types of noise to an image. Each type corresponds to some physical source of noise and this should be taken under consideration in your chosen course of action for removing noise from an image.

In this section, we will try to get a rule of thumb on which filter should be used for each type of noise.