Introduction

Compression reduces the number of bits used to represent each pixel in the image. The algorithms used to compress the images introduce, to a lesser or greater extent, visible artifacts. In general, the more complex algorithms introduce fewer of these artifacts. The processing power available to us increases with time, so more sophisticated algorithms always are being introduced into the codec products. This has to be done in a way that uses realizable computation, and has a short processing delay. The processing should not distort the image so that visible artifacts are generated.

As the level of artifacts decrease, so does our acceptance of the distortions. Once we were happy to watch monochrome television; now it must be color. Families were thrilled with the flickering home movies shot on 8 mm film; now a DV camcorder produces almost broadcast-quality results. Even so, the acceptable level of artifacts will depend on the application. If you paid to go to a movie theater to see the 70 mm print with Dolby Digital and they showed you a projection from a VHS tape, you would consider the picture unacceptable.

Compression systems exploit the mechanisms of human visual perception to remove redundant information, but still produce a compelling viewing experience. A typical example of the process is the persistence of vision, where an instantaneous view of a scene (for example, through a shutter) fades over about one-tenth of a second. This allows us to portray the continuum of time by a series of...