1. The processing of a signal (by a simple electric circuit or by some more complicated device) in such a way that the behaviour of the signal is affected in either the time domain or in a transform domain.
In time-domain filtering each element of the original signal is replaced by a sequence of elements, proportional in amplitude to the original signal but spaced in time; the sum (assuming linear filtering) of these sequences forms the new signal. In transform-domain filtering the elements of the original signal are not those of its amplitude but rather of its components under, for example, Fourier analysis or Walsh analysis; they are then spaced not in time but in frequency or sequency respectively. Many other transforms are also used.
Filtering, both in the time domain and in various transform domains, is of great importance in multiplexing. A simple but very common example of filtering in the frequency (Fourier) domain is the use of passive resonant circuits, RC circuits, or active filters to effect low-pass, band-pass, high-pass, and band-stop functions; these are much used, e.g. in data transmission lines and modems.
2. A technique for anti-aliasing. Aliasing occurs in an image when the sampling rate is not high enough to capture the changes in the image. Filtering applied to a scene spreads the influence of a pixel across the scene. Thus every object makes some contribution to each of the final-image pixel intensities. The value of the pixel in the anti-aliased image is computed as the weighted sum of its immediate neighbours with the weight inversely related to distance. The effect is to reduce effects such as ‘stair case’ changes in edges at an angle to the axis of the pixels at the expense of some loss of resolution.