TABLE OF CONTENTS Image noise in digital and video images is similar to grain in film. Low-light surveillance video will frequently be noisy, and this noise is often referred to as snow. The overall image degradation created by image noise may reduce the visibility of fine details.
Noise is caused by several factors in a digital environment. The most common contributors to noisy images are poor-quality components, long exposures, underexposure, and high ISO settings. Underexposure doesn't actually increase noise, but it reduces the signal-to-noise ratio, making the noise a larger percentage of the image data.
There are two general methods for reducing noise: processes that work on a single image and processes that work on a group of images (such as multiple frames of a video). This chapter will explore methods that work on individual images. The method for working with multiple images is a frame averaging technique, which will be covered in Chapter 23.
Of the methods available for reducing noise in individual images, some can reduce both the luminance and color noise in the image, and some can be applied specifically to the color noise in an image. If noise is primarily color noise (it affects one or more color channels rather than the brightness values in the image), the noise can be greatly reduced while retaining most of the detail in the image. If the noise affects the overall brightness values in the image, most noise reduction methods will cause overall blurring and loss of fine detail...
