Page 112 - C:\Users\msi\OneDrive\Documents\Flip PDF Corporate Edition\E-Commerce\

The Sequence of events for MPEG

First an image is converted to the YUV color space. The pixel data is then fed
into a DCT, which creates a scalar quantization of the pixel data. Following
quantization, a number of compression algorithms are applied, including run-
length and Huffman encoding. For full-motion video, several more levels of
motion compensation compression and coding are applied.

MPEG -2

It is defined to include current television broadcasting compression and
decompression needs, and

attempts to include hooks for HDTV broadcasting.
The MPEG-2 Standard Supports:
1. 1.Video Coding:
 MPEG-2 profiles and levels.

2. 2.Audio Coding:
 MPEG-l audio standard for backward compatibility.

 Layer-2 audio definitions for MPEG-2 and stereo sound.
 Multichannel sound.
3. Multiplexing:

 MPEG-2 definitions

MPEG-2, "The Grand Alliance"
It consists of following companies AT&T, MIT, Philips, Sarnoff Labs, GI
Thomson, and Zenith.
The MPEG-2committee and FCC formed this alliance. These companies together
have defined the advanced digital television system that include the US and
European HDTV systems. The outline of the advanced digital television system
is as follows:
1. Format: 1080/2: 1160 or 720/1.1160
2. Video coding: MPEG-2 main profile and high level
3. Audio coding: Dolby AC3
4. Multiplexor: As defined in MPEG-2
Modulation: 8- VSB for terrestrial and 64-QAM for cable.

Vector Quantization Vector quantization provides a multidimensional
representation of information stored in look-up tables, vector quantization is an
efficient pattern-matching algorithm in which an image is decomposed into two
or more vectors, each representing particular features of the image that are
matched to a code book of vectors. These are coded to indicate the best fit.

In image compression, source samples such as pixels are blocked into vectors so
that each vector describes a small segment or sub block of the original image.

The image is then encoded by quantizing each vector separately