TABLE OF CONTENTS The Discrete Cosine Transform, or DCT, is used throughout the MPEG standards as a way to provide video compression. This mathematical tool allows images to be represented with amazing visual accuracy (at least within the limits of human perception) with far fewer bits than those needed to create the original image.
The DCT used in MPEG operates on 8 pixel by 8 pixel blocks of image data. Luminance and the two color difference signals (see Chapter 3) are compressed separately by MPEG, but they use the same DCT principles. Figure A-1 shows a simple before-and-after picture of the DCT process. At the input to the DCT, an 8x8 block of pixel data is present. At the output from the DCT, an 8x8 block of data coefficients is present. The data contained in the output is a very accurate representation of the input; however, the transformation has made it easier to get rid of redundant information.
Let's take a closer look at the output of the DCT. First, it is important to understand that the DCT output is no longer a picture in any form recognizable by a human. In order to create a recognizable picture, the DCT has to be run in reverse to create a block of pixels (this is part of what an MPEG decoder does). What the DCT output contains is a set of numbers that each corresponds to one aspect of the 64 input pixels. For example,...
