The Internet Encyclopedia (Volume 3)

P1: IML/FFX P2: IML/FFX QC: IML/FFX T1: IML

Video ̇Compression ̇OLE WL040/Bidgolio-Vol I WL040-Sample.cls September 14, 2003 18:10 Char Count= 0

DIGITALVIDEOSIGNALREPRESENTATION 539

Table 1Data Rate and Corresponding Compression Ratio Required for Several Well-Known Applications

Compression

Application Network Data Rate Video Parameters Ratio

Mobile customer iMode About 9.6 kbps About 320× 240 ×16 at 15 fps About 1920:1

service video

Video-conferencing, POTS About 56 kbps About 320× 240 ×16 at 15 fps About 329:1

Shopping ISDN, FOMA About 64 Kbps About 320× 240 ×16 at 15 fps About 288:1

(initial)

GPRS About 128 Kbps About 320× 240 ×16 at 15 fps About 144:1

FOMA (planned) About 384 kbps About 320× 240 ×16 at 15 fps About 48:1

UMTS Up to 2 Mbps Up to 1024× 768 ×24 at 29.97 fps About 566:1

Digital Movie Broadband About 500 kbps–1.5 Mbps Up to 1024× 768 ×24 at 24 fps About 905:1

Digital Television Cable HFC About 1.5– 4 Mbps Up to 1080× 1920 ×24 at 60 Hz About 746:1

(29.97 fps)

Digital Versatile Disk DVD distribution About 6–10 Mbps Up to 720× 480 ×24 at 29.97 fps About 42:1

Digital Cinema Satellite About 60–80 Mbps over Up to 2560× 1080 ×30 at up to About 55:1

45 Mbps line 29.97 fps

Operations and Infrastructure

Supported by Digital Video

Video compression facilitates system functions compris-

ing storing, retrieving, transmitting, receiving, nonlinear

editing, indexing, manipulating, restoring, and denoising

digital video sequences.

Video compression technology provides a signifi-

cant component of infrastructure for familiar systems

including digital, interactive, or time-shifted television,

video-on-demand, home and in-flight entertainment or

shopping, interactive games and multimedia, streaming

media, video editing suites, camcorders, security cameras,

and medical scanners.

DIGITAL VIDEO SIGNAL

REPRESENTATION

Anatomy of a Digital Video Sequence

A motion picture scene is a subsequence of images

(frames) usually resulting from continuous camera ac-

tion. The images that define a digital video sequence com-

prise an ordered set of pixels. A pixel may be defined as the

smallest distinguishable and resolvable area in an image.

The number associated with an image pixel may refer to

any information of interest including sampled measure-

ments of light, temperature, image feature location, or

visual object trajectory. The pixels of multidimensional

images are labeled by at least two quantities referred to

as dimensions. A hyperspectral image records a collection

of subimages each of which is gathered from a distinct re-

gion of the electromagnetic spectrum. An inhomogeneous

image records a collection of diverse information. An en-

semble is a collection of images described in statistical

terms.

Sources of Digital Video

A digital video sequence may be captured or created di-

rectly in digital form or converted from an analog video

source. In view of the widespread use of existing ana-

log display devices, digital video sequences are frequently

converted to analog form and current digital television

standards are expressed in terms consistent with exist-

ing analog television systems. International Telecommu-

nication Union (ITU) Recommendation ITU-R BT.470–6

(11/98) records characteristics of conventional analog

television systems including the National Television

Standards Committee (NTSC), phase alternating line

(PAL), and System `e Electronique Couleur Avec Memoire

(SECAM) adopted by a number of countries worldwide.

Although discussion of analog video input types is out-

side the scope of this article, they may be classified as

composite, component analog, or S-Video. In contrast, a

serial digital interface provides connections for ITU-R BT.

601–5, composite digital video, and four channels of dig-

ital audio with a transfer rate of about 270 Mbps.

Interlaced Digital Video

According to ITU-R BT.470 –6, all the television systems

listed in Annex 1 of that recommendation employ an as-

pect ratio of the picture display (width/height) of 4/3, a

scanning sequence from left to right and from top to bot-

tom, and an interlace ratio of 2/1 resulting in a picture

(frame) frequency of half the field frequency. Odd and even

scan lines are traversed on successive scans.

Noninterlaced (Progressive) Digital Video

In contrast with the Recommendation ITU-R BT.470–6,

noninterlaced (progressive) video sequences result from

scanning all the lines of a picture frame once per field

resulting in a picture (frame) frequency equal to the field

frequency. In principle, video frames may be scanned in

any order such as a depth-first tree search widely used for

tiled images.

Color Models

The Commission Internationale de l’Eclairage (CIE)

defines standards for color illuminants and observers