ABSTRACT
This chapter reviews some types of noise and the methods by which the noise can be reduced. One simple method of reducing noise is to smooth the image. This works because the noise of one pixel is independent of the noise in the neighboring pixels. Thus, the average of the noise over a region should be close to a constant. Salt and pepper noise describes the noise in which some pixels are completely white or completely black. Early couple-charged device cameras had issues that infused a lot of noise into the images. Decades of improvements though have eliminated many issues. Consecutive erosion and dilation operators are very useful for removing salt and pepper noise, but as this example demonstrates, this process is less effective on removing random noise. The Wiener Filter attempts to minimize the noise in the Fourier space. Shadows also provide a source of noise. A vehicle driving along a tree-lined lane will have several shadow patterns.
