INTERNATIONAl
`STANDARD
`
`ISO/IEC
`11172-2
`
`Rrst edition
`19~1
`
`Information technology - Coding of
`moving pictures and associated audio for
`digital storage media at up to about
`1,5 Mbit/s -
`Part 2:
`Video
`
`Technologies di!J finforrration - Codage de /'image animeCJ et duson
`associe pour les supports de stcckag~ numenque jusqu'A environ
`!,5 Mbit/s-
`Partie 2: Video
`
`Page 1 of 124
`
`Refererce mmber
`ISO/IEC 11172-21S93{E)
`
`ZTE EXHIBIT 1004
`
`

`
`:
`i'
`j
`1
`
`I
`
`ISO/IEC 11172-2: 1993 (E)
`
`Contents
`
`Page
`
`Fo.e~M>rcl .............................. .... ... ....... . ....... .. .... .... ... . .... ....... .. ........ - ........ iii
`
`lnttoduction ........... .......... ..... .... ... . ... ... ............ .. ........ , . . ..... : .... ........... . ... .... lv
`
`Seetlcn I: Gencnl ............. ............................. .. .. .... ....... .... ........... ................ l
`
`1.1
`
`S:ope .. ................ .. ............ .. .. .. ........... .................. ........... ............... .. l
`
`l .l Normalivc references .......... : . ..... .. ........... ..... ...... .... . ............. : ................ 1
`
`Section 2 : Tec:hllical elemeols ..................................... ....................... .............. 3
`
`2. I DerUlitiou ........................ .... ....... .... ..... .. ........... ........ ....... ................ 3
`
`2 .1
`
`Symbols ·a.nd a~b~viatioos ......... .. ..... ........ ... .... ....... .... ................. ....... I 1
`
`2 .3 Method of desctibing bitstream syntax. .................................................. 13
`
`2.L Require(lleats ........... ....... .. ..................... ..... . ........... ....... ........... ....... 15
`
`An n exes
`
`A
`
`B
`
`C
`
`8 by Slm·erse discrete ~osine tnnsform ......................... ........ ................. 3 9
`
`Variabe Jentth co:le tables ......... ......... .... ..... .............. .. .. , ......... .... ....... 40
`
`Video ) ufferiug verifier ............ .... . ........... .. .... ... ........ ..... ........... .. ....... . 4 9
`
`D
`
`Guide IQ cn<X>di.og video ..... ............................. ................. .................. . S 1
`
`E
`
`F
`
`Biblio ,raphy .... ....... .... ............. .... ............... ........ .. ........ .... .. ... ........ 108
`
`List C!f patent bolders ........... ....... ......... .................. .... ........ ................ l 09
`
`e ISOIIEC 1!»3
`All ri~bts ~se:ved . N o part of tbi' public.alioa may be reprodu::ed o r ulili.zed in M! form or b~
`aDy means, el.bctrotic or mecbanical, including protocopyiog and microfilm, wiltoul
`penn.ission in writi.ag from the publisher.
`
`I SO..tEC Copyright O ffice • C.se Postale 56 • CH 1211 Geneve 20 • Switurl8nd
`
`Printed in Switurla~d.
`
`Pa~~ 2 of 124
`
`

`
`@ ISO'IEC
`
`ISO'IEC 11172-2: 1 g93 (E)
`
`!
`
`Foreword
`
`ISO !the International O·ganization for St3ndardization) and IEC (the lnta-(cid:173)
`natiooa. Electrotechnca Commission) f orm the specialized system for
`werldv.ide standardization. National bodies that are members of ISO or
`IEC participate h the development of lmemational Standards through
`technical committees established' bV the respedlve organizatiOn 'to deal
`with partb~ar fields of techncal activity. ISO and IEC technical com(cid:173)
`mittees collaborate in fields of mutual interest. Other international organ(cid:173)
`izations. governmental and non-gcvernmental, in liaiso(J Vvith ISO ar.d IEC,
`also take part ·n the work.
`·
`·
`
`It~ the field of informa!i0'1 technology, ISO a'1d IEC have established a pint
`technical committee, 'SO/I::C JTC 1. D·aft lnternation~l Standa!'Ps adopted
`by the joint te:hnical ccmmi:tee are drculated to national OOdies for vot(cid:173)
`ing. Publication as an International Standard requires·app·oval by at least
`75 o/o o: the national bodies casting a vote.
`
`International Standard ISO/IEC 11172-2 was prepared by Joint Technic:~l
`Committee lS0/1 ::c JTC 1. Information technology, Sub-Con'Ynittee SC 29.
`Codisd represena tiof'i· of ar.xio, p eture, mlitimsdia 'and hypermedia infer-
`matron.
`·
`·
`·
`·
`
`ISO! EC .11172 consists of the fdlowing parts, urder the·general title In(cid:173)
`formaton technology - Coding of movhg p'ctures 3nd associated avao
`fqr digiti!/ storage media ~t 4' !o sbout 1,5 Mbir/s:
`·-"- Part 1: Systems
`
`- Part 2: 'Video
`
`- Part 3: Audio
`
`- Part 4: Compliance teSting
`
`Annexes A. B and C fo·m an integral part of this part of ISO.t1EC 11172.
`Annexes D, E ard Fare fer information only.
`
`Page 3 of 124
`
`ill
`
`

`
`ISOIIEC 11 172-2: 1993 (E)
`
`©ISOJIEC
`
`Introduction
`
`Note -· Readers in~ in an Ol'oview of the MPEG Video uyer sboald read this lntrodactioo and then
`pro<:ea1 ro 3l~X D. before rewrning to clauses I md 2
`
`0.1 Purpose
`
`Thi1 partofJSO/IEC 11172 was de\el.oped io respo~se to the growing -.ecd for a common foonat ior
`repesenting compressed video ou various digital storage media sucll as CDs., DA Th, Windester disks and
`q>tical drives. This part of JSO/ffiC 11171 5pecifies a ooded represenlalion that cat be used for
`COD!pressin: video reCJ!ences 10 biuates around 1,5 MbitJs. The use of this part of ISOIIEC 1117?. means
`that motion videoc:n be manipulated as a fonnofcomputtr dau andean '00 tr3nSmitted and received over
`eK.i!itiog and fU.tue uenvorlcs. The coded represenration em be • sed wilh bOll 625-line and 515~ine
`television 3lld provides .fleXJl>ility fCI" use wilb worbtaliOD and persooal computer di1pbys.
`
`This part ofiSO/lEC 11172 W1lS de.;eloped to operate principally from ru>raBe media offering a continuous
`uamrer rate of about 1.5 MbitA. Nevertheless it can be used xrore widely thlUilbis lxaule tbeapproacb
`tH::eo is generic.
`o. 1.1 Coding parameters
`
`The ittelltiw io.4e~lopilg this part or ISO.'IB::: 1.1171 bas been to define as011rce codin: algorithm with a
`large degree of flexibiityt.bat ca.lJ be med in many differentapplicatioos. To acbeve this gml, a aumbec of
`~ pasamettrs defining the cbaraderisl:ics of co:led bitstremn.s and ~ aJe Q)ntained in.Lt)e bitstream
`itself, This alloVIS fur exflmple, the algorithm to be uSed for picnres witb a variety of siw: md aspect
`ratios and oa channels or devices ()pernting ~a wiJe mage of bitrates.
`
`Because of !be large range of ae characteristics or bitstreaau that ca1 be represented by this pan of ISO!IEC
`11172 a sub-set of these coding pmmetets known as tbe "Corutnined ParamettrsM bas bttn defined. 1he
`aim in defming the consllained pmmeten is to offer guidance a>out a widely useful range of pll'3llleten.
`Conforming to tllis set of constraints is oot a requirement of tbi.s part of JSOIIEC Ill i2. A flag in the
`llitstteam irdicates wb~ or DO( it is a Coastraiaed Parameters bitltfeam.
`
`Summary of tha Constrained Parameters:
`
`Horizontal 11icuue size
`Vertical picture size
`PiciUre area
`Pel rate
`Picture rate
`Molion vec10r mllge
`
`Input buffer size (in VB V DlOdel)
`Bitmte
`
`Less thanoreooalto 768 pels
`LeSs thanor.eaualto 5761ines
`~ than or C(J.Jal to 396 ma:roblock.s
`Less thmoreoual to 396x25 macroblocbls
`l..es5 tmn or eaual to 30 liz
`Less ttM -64 to -+63,5 pels (using balf-pd veciOn)
`• [t'<IU<.wcU u_ f code am forward f cocJt <= -4 (see tableD.nl
`Less ttM or equal w 327 680 bi~
`Less tbm or equal to I 856 000 bits/s (co•stant btnte)
`
`0. 2 Overview of t he algorithm
`
`The coded representaoon defined in this put or ISOIIEC 11172 achieves a b.igb ca:npression ratio while
`p-eserving &ood picrure quality. The agorillln is not k>ssle$5 as tleexact pel "aloes are not preserved
`during ceding. Thecboi ce of tile techniques is based on tlle need 10 balance a bigb prureqt.alily wd
`compressioa ratio witb tte requiremenl to make random occess to tbe coded bitstrean. Oblaini~ go<Xl
`picll.re quaity al tbe ntntes of interest demanrts a~ bigb compression raio. whic.h is notacllievable
`witi iAtrafrune coding <Wne . . De need for ra.r.dow access, however, is best "'tis fled witb p1.1e inoaframe
`ooding. This reqtircs a careful baance betweea it~ and iotrframecoding and bernecn recursive and 1101r
`reCIIJ"Sive tell}X)£"41 redundancy redlxtion.
`
`Page!v4 of 124
`
`

`
`0 ISQIIEC
`
`ISQIIEC 11172·2: 1993 (E)
`
`A I1IIJI ba' of lecbn:ques are used to achieve a bigJl comprtSSioll ratio. The first. whlcb is almmt
`independent {rom tnis p111 ofiSOAEC 11172. is to select an ar.propriale ~ nsolulio• for the signaL
`The aJgori!lun !lEn us:s blod~tB.scd !ll()ti(lo comp;:malion to rt:dll.':e lbe te~ reduadmcy. MC>Oon
`ooopens.uion is used f01 causal prediction of the cuneot picture from a prevl>us pcnue, fur oon-cansal
`predid.ioo ol k CUTellt picture from a furue picture, or fcc- interpolm 'lie predictioo from past ood ftnJJe
`pi.cwes. Motion vee mrs rue defined for eocb 16-pel by 16-line regioo of lbe picture. The difference signal,
`the predictionenor. is fmrercanpressed asbg the ciscrete cosine tr.uufocm (DCf) 10removespU.ial
`mnelation before it is quarti.led in an irre\ef'UD!e process that discards t.be less iaportalt infunnatico.
`Fitully, lhe motion vectors are rom biDed will tk ocr infol'tJI.3DcD. aro coded usinz variab:e lengu CJ:Xks.
`
`Cl. 2.1 Temporal processing
`
`Because of tbe con11icting requisemerts of randan att.e$ and highly cff1cient compression, three main
`pcturl! cypes ~ dermed. htra-coded pictures (1-Pico.res) are coded ~it.tlo ... reference to other pictares.
`They provide aca:.ss pom to the coded~ wberedecocling canbegm, but are coded wit.b only a
`moderate compression nfiQ Predictire coded picbire> (P· Pictures) ate mded more efficient) 'S using notion
`compensated )rediaion from a past inn or predictive ro.1ed pictue and are generally ustd 1S a reference fcc
`futlerpredlctioo. Bldi.rccOOmlly-predictive coded pi cures (B-Pictures) provide the higllest degree of
`compression b.Jt require both pan and fi.tUie reference pct~res for motion compensa1i01. Bidi.I:Wiio-.ally(cid:173)
`Pfe(liclive cOOed pictwes are riever llseO as refumces Cor prediCtion The organisation of the three piaure
`types in a seqoooceis very flexible. 'lbccboice is left to the encoder and willdependonlbe lleqliremeatsof
`tbe application. Figure 1 illu.<Uaes tbe sebtioosbip between tbe three <lfferentpicnre types.
`
`Bi-directional
`Prediction
`
`Fi,ure 1 -- Example Dt temporal plc:twe structure
`
`Prediction
`
`The fotrth piciW'e type defmed ill this part ofiSOIIEC 11172, ue D-picture, is provided to aUow a 5imple,
`b•t limited quAliLy, fast-forward playback lllO<b.
`·
`
`0 . 2 . 2 Motion representation - nacroblocks
`
`The choice of 16 by 16 macrobla::Jcs for the IOO<:i:ln·<:Oflt>ellsatioo uuit is a result c:L the Lrnde-off between
`increasitl2 tlle coding efficiency provided by using motion infommlion and the ovCihead needed to store it
`Eac1J tmaobloc:kcan be one ofammberofdifferent typeS. For exarnrle. intra-coded, fcrward-predictive(cid:173)
`ccded, backwaJd-pmlicthe oodxl. aad bidi.ra::ti>oall!-pediai vc-axJed m.acrobbcks are permitted in
`bidireaionaJly-predCtive ocxJed piCIUJ'eS. Depending oo the type of the m.'lcrc:block, molio1 ~k>r
`information an4 «lbtt ~ide illfcm:.ation are~ with the mmpressed prediction enor s~nal in each
`ma~ock. The motion vectors are encoded differemiall y will respect 10 Ill e. last ~ motion vectDr,
`usin1 variable--length end~. Themaxim!Dlllengtb oftbe veders that may be represented caD be
`programmed, oo a pitn~re-by-picture basis, so that the most de:rmnding applications can be met without
`canpromising tbe p::ri>rmance of the system in more ooroal siiUOOalS.
`
`It is t.be n:sJX>n>iljlity of tbe eJ.coder to calculate a:ppJOpriale motion veclOI'S. Thh p3Jt of ISOIIEC l l t 71
`does net specify how thisslx>t.idbedooe.
`
`Page 5 of 124
`
`v
`
`

`
`ISO/IEC 11172-2: 1993 {E)
`
`~ 1$0/IEC
`
`o. 2. 3 Spatial redundancy reduction
`
`Bolh ori' iDal pictwes airl predicti<m enor signals have high spatial re(hcdancy. This part of !SO/lEC
`11172 U$eSa .JOCk· based DCf melbOO witb visually t~eigllted quantization <Uld run-JtnglbCOOing. £acl8
`by 8 Hock of lbe origioal piaure for in Ira-Coded ma::roblocts or of lhe prediction error for preclictive-<:oded
`nacrd>locls is tmnsformed into lhe OCT domain where it is scaled before being quantized. Aflfr
`qtmtiz.alion nany of the ccefficiens are zero in value and so lwo-.dmeruional run-Jengtb and variable
`length coding is ustd 10 enc:ode he remaining coeffici::nts_ efficiently.
`
`0.3 Encoding
`
`This part of IS<Y.IF£ ll l 72 does rot specify an encoding process. It specifies tbe S)'llta:t and semantics of
`lbe bitStJeam and tke signal processing in tbe decoder. As a result, many optioi1S are left open 10 tn<X>ders
`k:l trade-off cost and speed against picuu-e quality and cod~ efficiency. This claJse i& a bref description of
`lbe functions that need 10 -be perfonned by an encoder. Fi~ure 2 sbows lhe main functicnal blocks
`
`Picmre
`lloc/
`P~dor
`
`OCT is discrete cosine troosform
`ocr-tis inverse ili~~ire 113DSform
`Q is quantization
`Q" 1 is dequootizalion
`VLC is variable length co:liag
`
`Figure 2 - Shnplifie4 video encoder block diagram
`The i.Jplt video si~al ·must re digitized and represened as a kul.imnce and twO colour difference signals
`(Y, Cb. Cr). This may be followed by preprOCC$Siag and fonn.at conversim to &:lectan appropriate
`window. resolutioll ud input format. This~ of !SO/lEC 111 n requm that tbe colour diffe~eree
`;ignals (Cb and Cr> are suJ.sampled with ~t to tre lumina.n<:e l>y 2:1 i.J botll vertical andhoruomal
`diteclioos aml are refomatted, it necessay, as a 110*-interlarecl sigml.
`
`1be encoder nust choose 11bicb picn!re type 10 use for each picture. Havmg defined the pict:ue type:1, the
`:.DCO<lttestimates m<ltion rectOI'S for each 16b)' 16 macrd>lock in tbe picture. In P-Picwres onevectoc is
`needed Cor each non-intra n.aaoblcdcan::l in B-Hd!Jres one or two ve«ors are needed.
`
`It B-Pictures are used. some reoJdering (f !be pitture !e<}ueace is n~ before encodilg. Because B(cid:173)
`Pictures are coded usilg biilim:lioaal motion conpemated prediction, they can only be decoded aftu tle
`subsequent referellCe picnre (aD I oc J>-Picw.re) bas been decoded . . Therefo~e the pictures are reo«tered by tle
`
`vi.
`Page 6 of 124
`
`

`
`CISOIEC
`
`ISO!IEC 11172-2: 1993 (E)
`
`enroderso thai me pictUJeS arrive at rre dtcodfr ii tt.e ordU fordeooding. Jbe conea dispay on:ler is
`rec:ovetal by fledecoder.
`
`The basic uoil of coding wkbm a pictue is lhe macroblodL Within each piclllR, macrd>lodcs ;we encoded
`in sequence, left to righ~ top to bolton Eadl truemblock coruists of ~x 8 ~Y 8 blocks: f001 blocks of
`luninance.o• e blockofCb chrcmilaxe, :ndooe bloclc of Cr c.bromir~See figu-e3 . Notetlut die
`picture area covered by lhe fou blocks of 1 tminaAoe is the sune as tJ:e area coyaed by eacb of the
`cbromillance blocks. This is due to smsampliag of the cllrominalcc in(onnatioo 10 lllaldl ~ semitivity of
`the human vi$!Jal syste01.
`
`lTI ITJ
`
`y
`
`Cb
`
`Cr
`
`Figure 3 -·
`
`'Macroblock struct11re
`
`Fustly, for a given macroblodc, the codi11g mode is chOsen. It depends on the picture l)'I>C, lie
`.
`eff&tiveness of motion compensated predictim ., that loc.1J region, and tbe nature of the sigaal wihin the
`block.. Sea>odly, depending on ~cOOing RIOOe, a moo on oompensated prediclio• or rne CODI.ellS of the
`bloclc based 01 past aadlor ruture refe~ence pictures is formed . This predictic:n is subtracted from the actaal
`dala .il tlecurnnt mac:robJoct to forn aa error signal. Thirdly, tlis e nor si~nalls separated ink> 8 by 8
`b locks (4 JIIOlinmce aoc12 ch!oorinance blocks in each macrob1ock) aud a dis::.n:te CO$ine b'ap~form is
`perfCll'm:d on~ block.. Each reslllting 8 by 8 block of DCT ooetllcients is quantized and the two(cid:173)
`diJ\ensOnal bloclc is seamed in a zig-lag order to convert it ink) a Olle-di.mellsional suing (}f qumti:z.c:l ocr
`coefficienu. Frurthly, the sidc-nfomation for lie macrd>lock (mode, motion vectocs etc) and lhe
`qwnci.u:d coefficient data are encoded. For mat.imum efficieu cy, a nunbel ol variable len8$11 code lablesare
`defined rorthe dffereot tat\ elemel!ts. Run·lagdl OOdtOg is US«1 fortbe qoantm:d coefficientdaa..
`
`A OOJSCqoence of using different pi:wre types and variable len&lh <XXJiAg is tnattbe overall data rae is
`variable. In ~cations that ill \'Clive a fJJte4-ntecbaltlleL a FIFO bttfer ruay be Uied ro matdl theeocoder
`OdpUt to tle chumel. The Statw or tlis buffer may be monitored 10 OQnttOl the DJmber a bits geoeraJtd
`by the encoder. Conro1li~ tbequanlizatioo ~s is ~e ~direct way of ooatrolling·the titnte. 1b6
`part oflSQIIEC 11172 specuesanab>tnctmodel <t'~e )U!felin& SJSUm (tlle VideoBuffiriig Verifier) in
`order to cons !rain lbe maximum variability iu the uamber o f bits lhal ase oS«l fora gi veo picture. This
`ensures tha a birsllea!n can be decOOed with a buffer of koo'IVD si2e.
`
`At this stage, the coded .rep-esentation of the picture has reeo ge.nemted. The final step in the encaler is to
`te&eoence 1-PiciJre..c; and P·Pictures by dec<Xling lhe da!a !'0 lha4 trey can be used as refereu:e pidme.s for(cid:173)
`subso:Juentenca:ling. ne quantized coeftidens aredt:quwlizcd and an invetse g by 8 OCT i~ pc:rformed on
`each block. 1be predicthn enor sipali produced is then added back 10 the p~diction signal and limited to
`the required range to give a <leco4hl reb'ence plctare.
`
`0 .4 Decoding
`
`DecOOiDg is the intet>e of tJle encooillg operatioo. It is cotuiderably simples than euoodi~ as there is ro
`ne:XIto perform motion estinu.tioo and the~ rue maDy fe\l'eroJtions. The decoding p-<>Ce$ is defiled by
`this part of JS01IEC 11172. The descriJtiat that follows is a \'ery bieC overview of 0t1e possible way of
`decoding a bitstream. Olha recooers with di.1Jerent anhilectur~ are possible. Fi~ure 4 slx>ws tbe main
`functioml blocks.
`
`Page 7 of 124
`
`vii
`
`

`
`1SO/IEC 1 1 172·2: 1 993 (E)
`
`@ISOIIEC
`
`+
`
`Reoomll\lcltd
`olllput pictwes
`
`Motion V ectoJS
`
`OCT' 1 is inverse dilaete cosine transfCJlll
`o-1
`is dequanli.zmoo
`MUX-I' is demtlt.iplex.ing
`VlD
`is variable lenglh deaxling
`
`Figure 4 -- Basic video d eco der block diagram
`
`FOf" fhed-rate applications, tbe channel fills a FIFO buffer at a coostmt rate with the cooed bitstream. The
`decOdc:t reads Ibis baiier and decOOes tile data el.enellts in the bitSt~ean according ID tle defult'.d syttaJL
`As the decodet reads tbe bitstream, it identif.es the s1at1 or a code<l picture am tbea tile type of the picture.
`It decodes each maaoblock in the Jictnre in nm. Thenu.croblodc rrpe and the motion vectors, if present,
`a-e used 1o construct a prediction of tbe ctJtTeOtmacrobbck lased on past aud future refereace pictul"C$ that
`· .bave 1x!eD seored in lledeco:Jet. The crefficient data are deoOOed axJ deqlWitized. E:!cb 8 by 8 bJodc of
`roefficieDt da.IJl is t:rarufoon..'"<l by an inverse ocr (specified in ana ex A), and lh~ result is crtred to lbe
`prediction signal and J.mited to tbe defiled lliD~e.
`
`Aftu all tbe maaolllock! in the {iCOJre have been proces..<ed, tbe pic lUte has been re<:onstruC1ed. If it is m l(cid:173)
`pielure a a P-piawe i1 is a reference picture f<r sutnequent pictures and is siOted., repl.acq lbe oldest s!Ored
`ref ere-ace picture. Before the pic:Wres aJe displayed tlley may need 10 be re-oolered from lhe cOOed order co
`tlleir nall!ral display Older. Aflerreordrnng, die picllreS are availabC. in di9ital form, for post-processing
`aud display in any manner thai the application chooses.
`
`0 _ 5 Structure of t he coded video bitstream
`
`This part of ISOJIEC 11172 specifics a synlaX for a coded video bit.strean. 'Ibis synta~ contai.lls six layers,
`each of which either supports a signal processing or a system function:
`
`Layers or tb e syntax
`Se:Juen<e b.yer
`Group of pictures layet
`Picture layer
`Slice layer
`Macroblock layer-
`Blockla)er
`
`Function
`Raaoom access unit: conte>~t
`Randou access unit: video
`PriJllary coding UDil
`Re..~nch.rooiution u.nie.
`Motion compen;alioa wit
`ocr uou
`
`0.6 Features supported by the algorithm
`
`AppticaliOilS usiag conrre-~ video on digit2l Stonge media need D k able to ~dorm l namber of
`opetatiOIIS n addition to normal fortvard playback of rre seqaeoce. The coded biiSb'erult ~ bem designed
`10 support a numbet cf Ulese operalioas.
`
`riii
`Page 8 of 124
`
`

`
`C>ISOIIEC
`
`ISO/IEC 11172·2: 1993 (E)
`
`0 . 6 . 1 Random access
`
`R.ruldool ac::oeu is an ess:nial feamre fa "ideo on a ~e nedian. Rlmdcm access requiJes lint any
`picture can be de<xxJOO in a li:n1ed amomt of time. It implies the eli~ ol access points i:A tbe
`bilSltCam · Lba l> segnenu of ir.fonmr:ion that areidtnti.lable wd can be decoded wilhollt reference to Olher
`segtnents of dua. A ~g of two nndom axess poiits (btra-Pictures) per sectlld em be cdieved
`witboot signif~CSI~t loss of picnre qaility.
`
`0 . 6 .2 Fast search
`
`Depending oo tte storage mediiiD, it is possible to scaa the access pOintS in a COded bitSttea.m (with the
`help dan <4)plicatioo-speci.fic <irecwry or oflet knowledxe be :rood the S<lOpe of his part ofiSOIIEC
`11172) to ol>tain a rast-forwam and f~-re,·ene playtnct effect
`
`0. 6. 3 Reverse playback
`
`Some applicaliollS may Jec:juire lbe video siglla1to be pkyed in re\'ecse order. 1llis can be achieved in a
`deaxler by using meoocy 10 store mtire gro~s ofpicttrCS aft&. they have been decoded before being
`displayed ia reverse order. An encoder can make this reatureC$Ier by re<llcing tre k:ngth of groups of
`pic lUres.
`
`0 . 6 . 4 Error robustness
`
`Most ~gital stonge n ediaaiXI oomm.tricatioo dwloels are DOtetror-free. Appropriate channel rodint
`s:bemesshl>l1d beus.!darxt arebeyood tbe s::opeoftbls part of iSOIIEC 11172. Nevtttlleles$ the
`COOipesiioo scheme defired in this pan of ISO'IEC-111 n 6 robust to residual errors. The slice strlx:twe
`allow~ a cJoeodel to ro::over after a data ~ and ao resyochronize its decodi~. Therefore, bit tTI'OfS in the
`a>Dipressed data will C311St eliOIS in lbedfcojed p.aures k> k limited i'l area. Decalers may be able 10 use
`roooeililent strntepe> to dis!uise lheseerrors.
`
`0 . 6 .5 Ed iting
`
`There is a ronflia between t.be reqUmre:u for ligtl rodint dficieocy aod ~yedtilg. The ceding stnJCtul"!
`aod S)nt.ax bave notl:een designed with the primary aim Of ilinplifyin£ editing at any pi awe. NevellbeJess
`a n•mbet of featores bave hem included tbal eoab1t editing of coded dara..
`
`Page 9 of 124
`
`

`
`This pa;Je intentional! ~ left blank
`
`Page 10 of 124
`
`

`
`NTERNATIONAL Sf.t.NDARD C ISOIIEC
`
`1$0/IEC 11172-2: 1993 (E)
`
`Information technology - Coding of moving
`pictures and associated audio for digital storage
`media at up to about 115 Mbit/s
`
`Part 2:
`Video
`
`Section 1: General
`1 .1 Scope
`
`This pan of ISCYJEC 11172 speci&s the coded ~boo of video for dip tal storage media and
`specifies Ire <!eaxling process. The representttioo supports normal S}leed forvrud playbact, as wei as
`spcciaJ functions such~ J3l<i>m aa:ess, fast focward play met. fasl reve~Se phybad:, normal speed revesse
`playblct, pause aDd still pictures. This part ofiSOIIEC 11172 is cc.mpali>le 'lll.itl! slandald 525- md 625-
`line teJe•1isioo fom.ats. md it provides fi~ibility for use ~ person.u computer a1d worksution diij>b.ys.
`
`lSOIIEC 11172 is primwy applicable 10 digital s10rage me<ia wpport.itg a eoDtnuous tnmfer rate up to
`about 1.5 Mbh/s, sucll as Com~t Disc. Digital Au<io T~ 31<1 magnetic lwd o sks. Nevertheless item
`be used mere wilely th<n dW bc:cause of the &eaeric approoc._ liken. T~ stOrn~;e 11edia may be diredly
`ooanooled lo the decoder, or via ccmnooications means such as btls.c;es, LANs, or teltCXImmunicatious
`links. Thi~ panofiSO/iEC 11172 is in !ended for OOII· il'ILerllK:ed video {()f[ruiiS baviag appmtinalely 288
`lines of 352 pels aad picture rateS arou• d 24 Ht to 30 Hz.
`
`1 .2 Normative references
`
`The followin& InLem:a.tional Standard; rontain f"''Iisions "'hicb, llrougb reference in this te)t. constitute
`provisions of this part cL ISO/IEC 11172. At the lime of publication. the editions indiatted were valid.
`All standaldsare s11bject to revisioo, md parties toa~cemeniS based oo lhis p.'Ut diSO/IEC 11172 are
`encouraged to investigate fle p<lSSibi6ty of applying lhe most receat editions of the sttnda«<s 'indicated
`below. Members of 1EC Md ISO maintain registers of curreatJy valid lnr.emational Standa'ds.
`
`lSOIIEC 11172-1: l993 ll!j)rmotion tedl1()f.ogy- Cot!inz lf mo\in: pictwres tJJU.f rusocitJJed audio [or digiJal
`rtorage mJdia 01 up 10 about 1,5 Mlit!J - Part I: System
`
`ISOIIEC 11172..3: 1993/l!j)nnation tecillclogy - Cot!irtllf TMling pictrres tJJU.f rusocitJJeJ a.u.tib [or t.hgiJaJ
`storaze mtdia 01 up 10 clx!ut 1,5 Mbllls - Part J Audo.
`
`CClR Recomnendalion 601-2 Encodin: par4melus ofdigilal televuzor.for JJudios.
`
`CCIR Report 624-4 Ch4ractt rirtics of systeJKS pr m.oMcl!r~ alld corJtu televisio•.
`
`CCJR Recommendation 648 Recording of aidio Jiglllls.
`
`CCIR Rep;>rt 955-Z Sound broadcutin~ by satelliJe jor porrcbll and 11Uibile receirerr, vtduding NUe.x fl'
`SUI7II'IIll)' kscription of Advanced Digital SJSttm If.
`
`CCITf Recommeadation 1.17 Prt-emphasisusedon SoUNi-Pt?grCUIJ'ne Ci~uils.
`
`Page 11 of 124
`
`1
`
`

`
`lSO/EC 11172-2: 19-s3 (E)
`
`Q ISO/IEC
`
`IEEE.rnati Slallda'd Pll80/D2 1990 Speci.ficcuion for lht iraplut.entatioll of 8r 8 i11verse discre~ oosiltf
`lTQ/JSjrJnlt ~.
`
`IEC publicati:>o 90S:1987 CD Ditital AuditJ Systent..
`
`2
`Page 12 of 124
`
`

`
`e ISOIIEC
`
`ISO/IEC 11172-2: 19m (E)
`
`SecUon 2: Technical elements
`2. 1 Definitions
`
`For the pwposes or ISOfiEC 11172, ibe following defmilior:s apply If spedfi: 10 a pan. tbis is DOled in
`square brackets.
`2.1.1 ac coemclent lvldeo ]: Any ocr coefficient fa wbicb tbe rrc:qutncy in one or b()(h dimensions
`is non·u:ro.
`
`2.1 .! ac~~ •nit (system}: In the care of compressed Rudffi an ~s uoit is an audio acc:e$ mit. In
`lle case or comp;cssed video an acre$ unit is tre roded represen!rooo or a pictllJ'e.
`
`2.1.3 adaptive segmentation (audio]: A sulldlvisi.on of lbe d igilld repnsenution o f aa audio si!Dal
`i1 variable ~gmen1S of time.
`
`2.1 .~ adapUve bit allocation (audio ): The assigtm ent o f bits to smbba:ads in a time and freqaeucy
`varying fashion a::o:>rding t:1 a psych(Ba>ustic mQ<Jel.
`·
`
`2.1.5 adaptive noise allocatioo (audio!: The assigmner11 of coding noise to freqaency bands in a
`time and frcqu:ocy vaying faslii>n aa::adilg tO a ps)ClJoacoustic mcdel.
`
`2.1.6 alias (udio}: Mirr<red sigaal component resulting rrom suiH'i)'QtJist sa.mpm g.
`
`2.1.7 aoal,-w Nterhu [andiol: Fik.elbaok in tbe ~ncoder that tnn!forms a broadband PCM aadio
`signal iniO a stt of subsanpled rutiY<md sanple5.
`
`2.1.8 audio access ~it [audio}: Foc-Layers 1 aod II an a.Jdio ace~ mit i> def111ed as the sm2llest
`partol die rocoded bitstmun wbicb cal be decoded b) it>elf, wheredecaled Illems ftfully recon&n.x:ffd
`sou•d•. Foe layec III an a1<1io aa:ess unit is part of tle btitream Out ~ decodable witb lbe usc of
`preYioosly 3CqlliJed mai.a iafcnmtion
`
`2.1.9 audio b11fkr [audio): A buffer in tbe sy>tan target decodu fo1 stonge of compressed audio da!a.
`
`2.1.10 au.dio sequence [audlol: A noo-intempted series of audio frames ~ whi:b tbe iollowillg
`parameterS are not cbanged:
`- ID
`- layer
`- Sampling nequcncy
`- For Layer I ru~ ll: Bitrare index
`
`2.1.11 backward motion ~ctor [vi•u): A motion vectOr Lhat is used for ~ol.ion ccmpensation
`f.lom a reference picture at a later time in d isplay order.
`
`2.1.1Z Baric (ru.di•): U n.it o{ critical band rate. The Baric ~ale is a non-linear mapping of the frequency
`sc:ab over tbe audio l"".,nge closely correspoodutg ~th Ill! freq uency selectivity of the human e3' across the
`1:8nd.
`
`2.1.13 llidinctionally predictive·coded pic ture; B-picf\u:e (video ]: A. pict.ure that is coded
`~ing m<ltim compensated pre<lli:tioo from a ~ Md'or ~we reference ric1w'e.
`
`2.1.14 bitnu: The rate <t vrbicb the compJeSSed bitstream isdeliveled from tbe ltora&e mediu.m to tle
`iApvt of a decoder.
`
`2.1.15 block cunpaoding [audio): Nonru.lizing of the digil:al repr:stntatioo or an :mdio sigoll
`will in a cettain 6me perio4.
`
`2.1.16 tioc:k {\'Ideo): An 8·row by 8-coltntl cctho&oaal blodc of pds.
`
`21.17 bound (aullo): Tie IOW~t mbband ill vrbicb imeDSity Stereo codin& it. uc;ed .
`
`Page 13 of 124
`
`

`
`ISO/IEC 11172-2: 1993 (E)
`
`© ISG1EC
`
`~.1.18 byte a.lipflJ: A bit in a coded bitstream is by~e-ili,ned if iu position is a mllltiple of 8-bits
`ftolll !be rust bit m tte rtream.
`
`U.H •yte: Sequence of 8-bits.
`
`2..1.2t dl:;mnel: A d1g,ital xred.ium. that stores or lmlsports an ISOIIB: lll72 sueam.
`
`t.1.2l (bannel ( audiol: ne left 3ld ri&hl cbaanels of a stereo sig.al
`
`2..1.21 cbromluoc:e (component) [Tideo): A matrix, bloct cr single pel represeating one of the
`two colour differerc.esi:nals related 10 he pciJ:nal) col run in the manner defined in COR Rec 601. The
`symbols u~ for tle colour dfference si~ rue Cr and Cb.
`
`2.1.23 Cloded audio bitstream {.•udio]: A coded representation o f a<\ audio signal as ' pecifierl in
`ISO/IEC 11172-3.
`
`2..1.24 cooed Yldto bitstre.am [videol: A. c:odee representation of a serie' of one or more. pictures as
`specified .in this part of rso/IEC 1_1112._
`2..1.25 a>ded ordtr (video]: The or4er in wbicb the pictm:s are stored and decoded. This order u not
`necessarily tbe same as the display order:
`
`:Z:.J-26 (Odec! represerC.ation: A data element as represc.nted in its encoded form.
`
`2.1.21 coding paramele" (ville.]: 1be set of user-definable parameters Wit cl!ancterize a coded video
`bictrtam. Bastreans are cbaracterised by coding paron~ Decoders are dlaractet'Sed by tte bitstreams
`tba1 t11ey are capable of de(o(Hng.
`
`2.1.21 colilpooent [vicleo ]: A matrix, block or ringle peL from one of the thr~ mrurices (lllminance
`and two cbomina.ct) tllattmke ll' a pcture.
`
`2.1.2~ compression: Redll<tioo in !he naoi>er rf bits used to represent 8D item of dati.
`
`2.1.30 constant bltrate coded video [video): A comp~ video bitstttalll with a roasunt
`a~bi!rale.
`
`2.1.3J c:oiiSbnt bitrate: Operatioo ~here tlle bitnte is constADt frcm start to finisb of be compressed
`biiStream.
`
`2.1.32 coastrained p arameters [video): Ibe values ct. the set of coding parameters defined in
`2 .... 3.2.
`
`2.1.33 corutraiaetl systean parameter stream (CSPS) [sysfem): An ISO/ffiC 11172.
`mal~lcxed stream for wblc• the coasttaints defined in 2.4.6 o f ISO.'IEC 11172,- 1 apply.
`
`2.1.34 CRC: Cyclic red•ndancy code.
`
`2.1.35 critical band rate lawiN): Psychoacoustic function of frequency. At a lhen aldible
`frequency it is prq>ort:ioml to lhenumberofcriti<al ~d~ below that fiequency. The unjts oflbecritital
`bood ra1e scale are Barb.
`
`2.1.36 erkical band [a udio): Psycll<>icoustic measwe in the spectral doma.il wbid ronespoods to the
`frequency selectivity of the human eu. llis selectvity is expressed n .Bask.
`
`2.13'1 data dement An item of dala as rep-esetted before enc:OOing aud after decoding,
`
`2.1.38 dc-c:oemdent [video): The OCT coefficient for wbic:h lht f~eiiC! is zero ill both
`dimmsions.
`
`4
`Page 14 of 124
`
`

`
`C ISOIIEC
`
`ISO/IEC 11 H 2-2: 1993 (E)
`
`2 .1..39 de-coded picture; D-pkture [video)" A pictur6 that is cooed nsing ooly infon:ilation from
`itselC. Of the OCT coefficietts in dleccded represenw.ioo, ocly the do-cnefficieats are presenl
`
`2.1.40 OCT coefficleat: 1\e aJr4>litude of a specific cosille balis f\llClion.
`
`2.l .41 4ecoded stream: 1be d:coded reronstruclioa of a CQillpressed biutream.
`
`2.1.42 duo4u iAput buffEr [video): The f"mt-in flnt~ut (FIFO) buffer specified in t.be video
`buffering ,crifltt.
`
`:U.43 tecoder irt>ut rate (vidoo): 1re datA rate specified in the vickx> ~uCferin& ~er llld encoded
`in t.be coded video bitstream.
`
`2.1.44 liewder: An ernb<ldim.eot Of a <lecodin& (IOCCSS.
`
`2.1.4S •eeodlng (proc::ess): The proc~s defined in ISO/JEC 11172 !hit reads an input coded bitstream
`ani }Toducesdeco<led pictures cr audio samples.
`
`2.1.46 cleco<ling ti~ne.stamp; DTS lsysteml: A field !.hm may be prcsenL In a packet header that
`indicates the time flat an access w"it is decoded in 1hesystem tar&etdecoder.
`
`2 . l .47 de-empllasis (audiol: Filleting applied to an C'Aldio sign.'ll after stomge or transmission to undo
`a linear distortion due to emphasis.
`2.1.48 de,uaotlzatmo ["ideo): Tbe process a r:staling tre qumlized DCf coefficierm after tbeir
`represeotalioo in tfle biutream b<lS befn deooded and Wore they are presented to the invCISCDCT.
`
`2.1.49 difita.J st orage media; DSJrf: A digi&al s torage 0 1 t:tcuumission revice or syslelll.
`;u.so ci L~crd~ cosin f hansful"lll; OCT lvik.): Eilher lhe f()r.\.aroi discrete cosire t.ralSfonn or the
`inverse discrete cosir.e tr.usform. The ocr is an ilYert!ble, discrete omogollal tnmsfonrutioo. The
`invezse DCris defined in ua.e~ A
`
`2 . l.Sl display order {video}: The <rd er