Reducing Noise and Sharpening

Noise Patterns
Noise reduction and sharpening are opposite sides of the same coin. Here we define noise as any unwanted non-image data, usually present in the form of patterns or fabrics that have been added to an image. The process of noise reduction attempts to mitigate or mute unwanted noise, which is defined by pattern elements that have edges. Sharpening does the complete opposite—it enhances any patterns, artifacts, or edges present in an image. Therefore, it is a good idea to eliminate any unwanted noise prior to sharpening an image.

There are numerous sources of noise. Three of the most common are as follows:
- Noise related to image capture (true noise)
- Compression-related noise (especially JPEG artifacts)
- Halftone-pattern (captured print screen artifacts) noise

Although these three common image patterns all have different origins, the process of removing them is similar. Unwanted noise can reduce the quality of an image by creating distracting patterns in the image. Nearly all unwanted image patterns can be removed, or at least significantly mitigated, by the application of smoothing or softening techniques that mask or disperse the unwanted patterns. Sharpening, by contrast, performs the exact opposite function: it serves to enhance rather than diminish any edges in an image, regardless of their source. In addition, the process of halftoning an image during printing can exacerbate patterns that already exist, so any unwanted patterns need to be effectively removed prior to sharpening or printing.