One problem in any video compression system is the large amount of data that must be handled. If every bit of data is sent exactly as captured from a video source, a huge bandwidth would be needed. Instead, data must be selectively discarded an important part of any lossy data compression system as part of the process of reducing the amount of data until it fits the available transmission bandwidth.

In this chapter we selectively review some basic aspects of the human visual system (HVS), in order to understand how to reduce the amount of data in moving picture sequences. Our goal for this chapter should be to describe how to get the best possible visual quality when a sequence is played back by a decoder. Unfortunately, we can't quite meet that goal. In practice, optimization of MPEG encoding and decoding systems is a very complex topic, and most MPEG implementors regard their optimization techniques as proprietary. What we can do, however, is describe some of the basic ground rules for this optimization.

Readers who have no familiarity with this area of technology will find the excellent introductory text by Cornsweet [Cor70] helpful, although somewhat dated. A recent book on vision by Wandell [Wan95] is also a good introduction to many aspects of MPEG. Poynton has written two very readable introductory monographs on color [Poy95a, Poy95b] that can be downloaded from a World Wide Web site on the Internet. The book edited by Benson [Ben85] contains a wealth of material relevant...