MPEG: A Video Compression Standard for Multimedia Applications

OutlineIntroductionMPEG GoalsMPEG DetailsPerformance and SuchSummary

Introduction1980s technology made possible full-motion video over networksNeeded a standardOften trigger needed volume productionAla facsimile (fax)Avoid de facto standard by industry1988, Established the Motion Picture Experts Group (MPEG)Worked towards MPEG-1

The Need for Video CompressionHigh-Definition Television (HDTV)1920x1080 30 frames per second (full motion)8 bits for each three primary colorsTotal 1.5 Gb/sec!Each cable channel is 6 MHzMax data rate of 19.2 Mb/secReduced to 18 Mb/sec w/audio + control Compression rate must be 83:1!

Compatibility GoalsCD-ROM and DAT key storage devices1-2 Mbits/sec for 1x CD-ROMTwo types of application videos:Asymmetric (encoded once, decoded many)Video games, Video on DemandSymmetric (encoded once, decoded once)Video phone, video mail (How do you think the two types might influence design?)Video at about 1.5 Mbits/secAudio at about 64-192 kbits/channel

MPEG CompressionCompression throughSpatialTemporal

Spatial RedundancyTake advantage of similarity among most neighboring pixels

Spatial Redundancy ReductionRGB to YUV less information required, same visuallyMacro BlocksTake groups of pixelsDCTRepresent pixels in blocks with fewer numbersQuantizationReduce data required for co-efficientsEntropy codingCompress

Spatial Redundancy ReductionZig-Zag Scan,Run-length codingIntra-FrameEncoded

GroupworkWhen may spatial redundancy reduction be ineffective? What kinds of images/movies?

GroupworkWhen may spatial redundancy reduction be ineffective?High-resolution images and displaysMay appear coarseA varied image or busy sceneMany colors, few adjacent

Loss of ResolutionOriginal (63 kb)Low (7kb)Very Low (4 kb)

Temporal RedundancyTake advantage of similarity between successive frames950951952(Simpsons News Clip Here)

Talking HeadTemporal Activity

Temporal Redundancy Reduction

Temporal Redundancy Reduction

Temporal Redundancy ReductionI frames are independently encodedP frames are based on previous I, P framesB frames are based on previous and following I and P framesIn case something is uncovered

Group of Pictures (GOP)Starts with an I-frameEnds with frame right before next I-frameOpen ends in B-frame, Closed in P-frame(What is the difference?)MPEG Encoding parameter, but typical:I B B P B B P B B II B B P B B P B B P B B IWhy not have all P and B frames?

GroupworkWhen may temporal redundancy reduction be ineffective?Many scene changesHigh motion

Non-Temporal Redundancy Many scene changes

(Mixbag clip here)

Non-Temporal Redundancy Sometimes high motion(Simpons News Clip here)(Hockey Clip here)

Sequence LayerGroup of Pictures Layer

MPEG Layers

MPEG TodayMPEG video compression widely useddigital television set-top boxesHDTV decodersDVD playersvideo conferencingInternet video...

MPEG TodayMPEG-2Super-set of MPEG-1Rates up to 10 Mbps (720x486)Can do HDTV (no MPEG-3)MPEG-4Around Objects, not FramesLower bandwidthMPEG-7Not (yet) a standardAllows content-description (ease of searching)MP3For audioMPEG Layer-3