`STANDARD
`
`ISOIIEC
`11172-2
`
`First edition
`1993-08..() 1
`
`lnfonnation technology - Coding of
`moving pictures and associated audio for
`digital storage media at up to about
`1,5 Mbit/s-
`Part 2:
`Video
`
`TechnoJ~g.es da rinforf!Jation - Codage de /'image anfmtje et du son
`associ~ pour /es supports de s tcckagB numenque j usqu •a environ
`1.5 Mbitls-
`Partie 2; Video
`
`Reference m.m ber
`ISO/JEC 111 72-2 1 S93!E)
`
`Page 1 of 124
`
`ZTE Exhibit 1 004
`
`
`
`ISO/IEC 11 172-2: 1993 (EI
`
`Con1ents
`
`Page
`
`Fo.e~n! .••... ··-·· .... . .•..... .... ... ... •.......•.. . ··-········ · .......•.............. ___ ....... .iii
`
`lntro<iuctlo11 ......... ................ ....... ............................. , ••.... : .... ................... Jv
`
`SectiQIJ 1: Ge~enl .............. ...... .. .. ................................................ ............... !
`
`Ll
`
`S::ope .................. . , . ... ................. .......................... ........ ................. !
`
`1.4 Normalivc rden:n~es .......... : .. ..................................... ....... .................. !
`
`Section 2 : Techllioal elements ............... ................................ ........................... 3
`
`2 . 1 Dcf'10iliou .............................. .. ........... ............. ........ ...................... 3
`
`2 .1
`
`Sy'Xlbols ·and ~bbraviatioos ........................................................ ...... 11
`
`2 .3 Me tllod of d~c1ibing bilsu-eiUXI synl8JL ..................................... . ...... 13
`
`2 .'
`
`Req11ireou:ots .. ................. ..................... .................................... ..... 15
`
`Aonex es
`
`A
`
`8 by 8 loYcrse di$crcle rosine IJ'Qnsfom .......... ............... ., ..................... :3 9
`
`8
`
`C
`
`Vari11b~ len,th oo:le table.-; ......................................... .. ......... ........... .. . < 0
`
`Video •otrerillgveriliec ......... ........ ................. .............................. ....... 49
`
`D
`
`Guide 10 €oo:>diog video ..................................................................... j t
`
`E
`
`F
`
`Biblio&rapby ................................................... ......... ................... 108
`
`List d pate nt holders .......... ............................................................... l 09
`
`e ISOIIEC 1~3
`Ali r:i~hto; rc.se:vm. No put o f thi1 p11blieatioa Dtay be reprodu:ed or uli.li.ud in M! fonn or b!
`aoy mea1s, elw<rotic or mecbao~l, iucloding J?hatocopyiog a.od micrortlm, "'iltout
`pennissim in writiJg from the publislur.
`
`ISO,{EC Copyrl~bl Office • C ase Pastale 56 • CH 1211 Genev~ 20 • Swit:zerllllld
`
`Prin!Jld in Switudmd.
`
`Pa~~ 2 of 124
`
`
`
`CO ISO'IEC
`
`ISO'IEC 11172~2: Hl93 (E)
`
`Foreword
`
`ISO rthe International O·ganiza1ion t or St3ndardization) and IEC (the lnter(cid:173)
`natiooa· Electrotechncal Commission) filrm the specialized system for
`worldwde stand-lrdizat lon. National bodies thai are members of ISO or
`I EC participate h the development of I n:emational Standards through
`technical commntees established. by the respe::i ive organizatiOn 'to deal
`with partbiar fields oi techn cal activhy. ISO and IEC technical com(cid:173)
`mittees collaborate in fields of mutual interest. 01her international organ(cid:173)
`izations. governmental ~d non-gcvernmental, in liaison vvith ISO ard IEC,
`also take part 'n the Vlork.
`·
`·
`
`In the field of informa'i01 'technology, ISO a1d IEC have established a pint
`technical committee, ·so/I::.C JTC 1. Doft l ntematon~l Standards adopted
`by the joi:lt t e:::hnical ccmmi:t·ee are circulate:l1o national bodies for vot(cid:173)
`ing. Publication as an lntem~tional Standard requires· app"'val by at least
`75 % d tl\e national bodies castirg a vote.
`
`ln1ema1ional Standard ISQIIEC 11172-2 was prepared by Joint Teeh nic::~l
`Convnittee iSOJI::C JTC 1, Informacion cecmology, Sub-CoJTYnittee SC 29.
`Coded represent3tiori·of a'.lclo, pcture, mt.itimedia and hvpermedia infcr(cid:173)
`matiOn.
`·
`
`·
`
`ISO/ EC 11 172 consists ol the following parts, urder the general title ln(cid:173)
`forrr.ati:Jn technology - · Coding of movng pictures 9nd associated auao
`for ciglt<!l storage media at t.p to stout 1.5 Mbir/s:
`
`- Part 1: Systems
`
`- Part. 2: ' Video
`
`- Part 3: Audio
`
`- Part 4 : Comp/i<nce testing
`
`Annexes A. 'Band C fo•m an integral part of this part of ISOftEC 11172.
`A nnexes D, E ard F are fer information only.
`
`Page 3 of 124
`
`
`
`IS()(IEC 11172·2 : 1993 (E)
`
`© ISOIIEC
`
`Introduction
`
`No~ -· Readers inleleSteC in an overview oftbe MPEG Video ltya sboald read this lntrodacnoo a»d tben
`JroCC(X1 ro a1rexD. before rewming ro cl.mses 1 md 2
`
`0. 1 Purpose
`'lhil pnofJSOI!EC 11172 was de\eloped in resJI(JIIse co the growing eero for a common fermat for
`repesentiDg oompressed video on varioos digital &ora~ media sucll as CDs. DA 1\, Wind~ dis~ and
`<lllical drives. This part of ISOIIEC 111 n specifies a coded represeutalioo lltar ca• be 11sed for
`COOJpressin! video ~1ences to biuates around 1.5 MbitJs. 'l1:e use of this part or ISOIIEC 1 11 72. means
`llat motion video em be lll.'\1\ipUlal.ed as a form of compucer data and can re transmitted and rccei ved over
`eKisliog and fi:nue re~()(kS. Tile coded repreg!llrallon can lle tsOO wil:h boll 625-lire 11nd 525-line
`television aad provides flellibility fa- use wilb wor!Qtatiou and personal coJnpuret dhpbys.
`
`lbis part ofTSO/lEC 11172 was developed to operate prillcipally from stora1:e media offerin~ a continuow
`IJall;rer rate or allout J .5 Mbit!s. Nevenleless it can be used 100re widely th!Ul Lbis l:ccauae lhe approach
`tak:eo is gelli:ric.
`
`0 .1.1 Cod ing parameters
`
`1be i.!xelltic:m m kYclopilg tills pan or ISO.'~ i 1171 bas been to 4efinc a sott<X! axlin! algorithm with a
`lu~ degree of ~xibility tbat ca11 be mQ! iD many different applicauoos. To achieve Ill is gml, a aumber of
`~ pammeursddiniog tlle dwacterist&':s of C0\1e(l bits~reaDU and ilecodm. we a>ntained i11-!be bitstream
`itself. This allo"s for exrunple.lbe algorithm tO be a sed for picnres witb a variety of ~izes and a.~
`ratiOIS ·and oo cbalulds or devices Dperntiog at" a wile rallge of bitrales.
`
`Bea.~ of lbe large range of 0e clw'aeteristia of bitstreams that call be rep.estniW by this paJt of ISO!IEC
`11172 a su~set of lhese cocling panmeters lmoW!l as t1le "Comtr.l.ined Pru'ametttSM 1las been defined. The
`aim indcftniJ1g die oonstrained p<Jameters is to offer gtJichnco .-,OOuu widely useful range ofp..-:un~.
`Conforming tCI this set oC coostrainiS i~ not a requiruocnl or thls part of ISOIIEC lll i2. A fbg in the
`bii'Sirolm iruklms wb~or not fr is a Coa.~traiaed Paramctcts biWeam.
`
`Sum~nary of tha Constrained ParametC!rs:
`
`HoriZ<IIlb.l picwre size
`VerticalpiCIUre size
`Picture area
`PeJ mte
`Picruro rate
`Motion vec10r mnge
`
`Input buffer size [iu VBV ID;ldel)
`Bilmte
`
`Lass tl1an or eQUal to 768 oels
`Lass than or. equal to 576 lines
`L".& than orecJJalto 396 ma:roblocks
`L"'Sli thin or eQUal to 396x25 macrobloclcr/s
`Le.ss tban or eaual to 30 liz
`IJ)l
`Les! tmn-M tD -+63,5 pels {uSinl! half-pel vectors)
`I fbadcward f codearrl forward f code<=HzeetableD
`Less tlml or e<lualto 327 6&1 bits
`Less tbtn or equal to 1 856 000 bits/s (coast:\nl blr.te)
`
`0. 2 Overview of the algorithm
`
`The coded .eprerentation defined in thi; put of lSOJIEC 11172 achieves 2 htth conpressioo ratio while
`JXe$Crving 1ood piaure quality. Ibe agorilbn is not .bssle$5 as dleexact ptl "aloes rue not prescrved
`clumg coding. Thecboire of tile techniques is based on tile need to balance a higb pmreqwlity and
`oo111pressioa ntio witb tre requirement ro mak~ random aoctss ro th.e coded bitstrean. Obt\ini~ go<xi
`rkure quaity at the bilntes of i11terest dwJan1ts a very tigb compression roc.io. which is DOt achievable
`with iltrafrme axling abrlc . . Tie need for mroo10 aaess, h.o\1/ever, is besl ~tisled wilb pure intraframe
`ooding. This reqaires a careful balaroee between ntta- alld inErframe coding 2lld betv.een recursive and noo(cid:173)
`recursive tel'lportol redundancy reductioo.
`
`Pag~v4 of 124
`
`
`
`CISCVIEC
`
`ISO/IEC 11172·2: 19SD (E)
`
`A IUiba of h"flniques arc used 14) a:bieYe ~ b1gb compnssio• ratio. The firS.. whlcb is allnast
`il\dependelll iron tbis pu1 oOSOAEC 11 J7~ i5 to s.!lcct an awropnale ~ nscluliOI for !he sigm.l.
`'Ibe aJgolit/un t.!En uses block-based motion cmnpem.ation to rcdu::e tbe It~ reduadm<y. Mocioo
`wmpensation is used for Qll~ prediction oftbe cuom p1t."'.u1C froal a pn:vC>u; lictare, fur non-causal
`p-ediaion of lbe ctn'ttlt picrure fran a fullln! picu.e, or fer- intcrpol<ri w (ttlliction from pa& and !tWJe
`pc.llRS. MOOOII VCC!lm ate deli:ntd forea:b l(;.pel by l&line regiOil of lbepicnu-e. Thedilfet'CIIcesignal,
`lhe pn:dictioo cnoc. is fll'tllucanpressed asbg the dsae1e cosine tnnsform (DCf) to renovespatiaJ
`c:oneluian bel ore it is quantilcd in an irre\enible process that discards lh~ less ~t infunnalim.
`Finally. me motion vectors are 0001bioed wiu lle OCT infonn.ali<Xl. and tOded usio& variable leugtk cod¢S.
`
`0. 2. 1 hm~oral proc6lssing
`
`Becau~e or Lbe oon.flicting req!Ai~im~eru of talld<Jn ao:ess and highly effiCient compression. tJirtt main
`pcture rwes ate defined. Ia~n-codtd pkbues (1-PicO.Cl) we coded v.ithol!. refErence to other pictlres.
`They provire access points to the coded ~ where da:oding can beg a , but are axled wilh only a
`moderue CQmpressioo talin Pm:litth·e coded picw~ (P-PictrJaS) are wc1:<J more efficientlJ "'~ molioo
`~to:l prediction fi'oiD a pastinh.orpndictlvecoded p!cttreard ~regrncrally used 15 a reference fer
`fliJtierpredlctioo. Bidi.ra:OO!L'llly-pml icti ve a:ldM piwre.~ (B-Pi ctures) JlfD vide l!le taghcst tlegJee of
`compres.sion but require bolh pan and fu.ule .efereace (icwres tor motion OOillJ:emaiiOL Blditu;liQially(cid:173)
`prodiclive coded pictures are never used as teft.rences for prediction. The orgmhation ot t1x: lhree picw~e
`t~ in a sequ:a;cis very Cleliblc. Theclloice is left to tre enaxter and will depend on the te<p1iremeatsof
`llle application. Figure 1 illu.~ the .ebtioustlip bctwun l:be threG dlfferattpicnre tytleS.
`
`Bi-di.rectionaJ
`Predjction
`
`F,,ure 1 ·- E:umple of temporal picture strudurr
`
`PredictioD
`
`The Coertb piclure typt d t tined i n dlis part oflSOIIEC 11172, t•e D-picture, is providetllo ~low a rimple,
`bn limlteAJ ~alit}', fast-forward playback~-
`·
`
`0 .2.2 Motion representation - hlacrobloeks
`
`The choire of 16 by 16 mACroblodcs for the mot:ion-CQil'{l(:aSat:i<>n unit 4 a resuh d r)le lJlde.o(fbetwet.o
`inCIU\SIIlg tlle coding efficimcy provided byuting moticn infonnallon and !he overhead nt~edcd Lo st«e iL
`EaciJ rmc:roblodc can re uue of a mmber of <lifferc.nt typeS. For example. il11n.-a>dtxl. fCl'Wdfd-predictive(cid:173)
`ccded. baclc'lllaJd--predi<Lhe axb:1 ud bi<linlcfun:tlly-pe4iaive<OOed n'UlCI"Oblocks are penniu.ed in
`billirectioDally-prcdi:ovec:ttled piC!Ul""..s. repending oo the typeofthe rMcrcbloclt, motioa \-t!CIOr
`inromutioo ancl ahcr ~ide itfCIUiatioo are stored wa.ll the rompres.'Cd predJCtion eaor sigml in c:acb
`lllilet~ock. The motion vectorS are encoded ditTerentia!ly will ~t to tbe last coded motion vectt>t,
`usi.n, Vl\riable-lecgth mdes. 1lle maxim~tm lmgtb of tbe veClDf's th.'ll may be represented coo~
`programmed. on a pictl.tre-by-pic:twe baSs, so tb:u the most delmndiJ:lg appticat.ons tUJJ be met W1tbout
`ccmpmnisiag lhe peB>nnance of the system in more oomal sitW'Lioos
`
`It is the resp:>lb'ililif>' uf the e1codtr 10 c.nJcuhrr apptoJriare motion vecl.Of'S. lhis part ofiSOIIEC l ll7l
`does n« specify how this strud be dooc.
`
`Page 5 of 124
`
`v
`
`
`
`ISO/IEC 11172·2: 1993 {E)
`
`© ISOitEC
`
`o .2 . ~ Spatial redundancy raducti()n
`
`Boih origiml j)ictures am predicti<JU error $igna1s have lligb spalial re(itL,dmcy. This part of [S011EC
`l l112 uses a block-based DCT methoo· witll vimilly ,.,eiglited quantization and run-Jengtb coiing. Ead 8
`by 8 Hock of lbeorigiilal pQu.refor in!Fcreoded ma::roblocts or of lhep~iotionerror for pre<llCiive<oded
`nacr<blocls is transformed into the OCT domain where it is scaled before being qt.mltit.Cd. Afwr(cid:173)
`qUllltizalion many of the c<efficietts uezero iD value a11d so LWI)-cimen>iooa1 run-~ta and variable
`.lengtb coding is us«110 eni:Ode be remaining coeff.icients_ etflcicntly.
`
`0.3 Encoding
`
`This pa,rt of ISO!IEC 11172 doe.> not spec;ify an encoding pr~. It specifies the SYJl~ and sem1lllics of
`the biJStJeam md tle signal processing in the decoder. As tt res ult, many op!ioiiS are left open 10 enm<b"s
`k:> trade-off ccst and speed 1gainst picture qoaliLy a11d coding ellicieocy. This cla.Jse i~ a breftbcription of
`the functiolls l.bat reed ro-be penotmed ~Y an m:ode1. Fi!ute 2 shows the Illil.ii1 funaicoal blocks.
`r----1 fteg.JlaiOr t - - ----,
`
`Piclun:
`ao~ I
`14-_...--t 1'n::ftctor
`
`OCl' is discrete cosine I.Ialsfom.t
`OCT"" 1 is inven;e. rli:,;crece ca;ire transfOilll
`Q is quantization
`Q" 1 is dequootir.alioll
`VLC is variable length co:liag
`
`Fig\lrt l - Si10plified viclec encoder block diagram
`
`The ap11t 'ideo sigPal must ~ digitized aod tep~esened as a bminanre lllld two colour difference si gna1s
`(Y, Cb, Ct}. This way be followed by preproces.siDg and fo.rmat converni(n to re.lectanappropri.ate
`window, resolution ud i.nput fOI'lDlU. This part of lSO/lEC 11172 requi"~ that the cotout dilfeten:e
`sigoals (Cb and Cr) are su~sampled with rtS}ttt to UJe Juninance by 2: l iJl botb vertical and ho:ri2ot1al.
`dirtctiocs and cue refoJDatted, if nere;swy, as a no1-inredared signa].
`
`The eocootr mmt eboo~ "bich pictllre type to ose fot" each picrure. H.avmg defined tbe pictlll'e typet, lbe
`eDCOderestimatts motion vectors for each 16 bJ 16 matroblock in ·tte picwre. In P·Pictures one v~tci is
`Beeded !or each ooo-intra naaohlcckam in B-Ikturts one ot two veGtOTS are needed.
`(f:B-Pictures mu..<ed. son:ereotdering a llle piaure ~ueace. isnece~sary before enc:OOng. Becau>e B(cid:173)
`Picrures are cOOed usilg biditoctiollal motion COIDpeumtt.d prediction, they can only be decoded aftet be
`subsequent reference pittu-e (lUI I oc P.-~cture) bas been decoded. ~o~ the pictures are reordered by de
`
`vi
`Page 6 of 124
`
`
`
`©IS-Q/EC
`
`ISO/IEC 11172-2: 1993 (E)
`
`eoccxler so thal the pictures arrive at the dec.olh ill llle an:Jer i'or decoding. Jte mnect display order is
`rerovend ey fledeooder.
`
`Tbe basic unit o f coding within a picture is lbe macro block.. Witbin each ~tic run:, llll\CI'd>locl!:s are enc:o<ied
`in sequence. 1eft to righl. top to bottom. Eacb rnacroblock consisiS of si;o( 8 by 8 l>lO«lks: fow blocks of
`lmniDallce,oDe blockofCb cbromiumce, 31d ore block of Cr-cbromi11ance.See figllfe3. Notetlw Cbe
`picture~ covered by tbe four blocks oflumina~~ce is the same & !k: area. coyered by eacb r:f tbe
`cbrommance blocks. This is due to subsampling of th.e chromina.are irloTillatioo 10 mlll.C'.h the sensitivity of
`tbe blUrlan vi&~.tl systern.
`
`[UTI
`[ITIJ
`y
`
`[I] ITJ
`
`Cb
`
`Cr
`
`Figu.,e 3 - M•c:robloek structure
`
`Fitslly, fora given t~cmblod<:, the cod.iug rucde is chosen. It depel.lds Oll t1te picture l~, the
`.
`effectiveness of motion compemattd predictioo n thai local region, and the lla£UR ol the 9pal w1,hil the
`block. Sea:mdly, depending en b e corliDg 1node, a majon rompmsated p-edictioa of the contents of the
`block based oo past a11dlor ruu.u-e n:fecence pivlures is famed. This precliuim is subiJacted fmu tlJe actaa1
`da!a io dlecumnt ma<:roblod: to form aa error sign'\1. Thirdly. Ois enoc si~-nal is sq>arated into 8 by 8
`blocks (4 luminance and 2 \;brommance blocks in eath m&croblock) an<1 a diocrete cosine tJamfonn is
`pe!formed on eacb block. Eacb resulting 8 by 8 block of DCT cotfr.cien~ is qtJ.!U'Iti7..ed and IJte lwo(cid:173)
`dillension~ block ls scanned in a zig~~g Ofder to conven it inlo a Ot1e...iJ.imensiOll~ Slrillg 0, quantizetl DC'f
`coefficients. F'rurtllly. the side-infQnnaOOn for l:le ll:laCn:block (mode, motion vectoiS ec:c) aDd the
`quantiud coefficie~~t dala are en cod~. For mailinum effiCiency, a nunbe4 of variable length code cables are
`defined (or the dffereot data elements. RJ.m,le-gtb axling is use,d for the quantized coefficient d<n.
`
`A co.asequcnce of using r1 iffer< picture types and variable Jen,.r;tb cofuig is lllal tbe overall data ntt is
`variable ln '1>Jll.iettions rmt inwh-e a fi~tei-ratech.umel. a FIFO buffer IU3-Y ~ used to matdllbeencodet
`o~ul to Ue channel. The statu. of tkis buffer may be monitore<l to control tlle ntl1Jlbe;r d bits genetal.ed
`by tlle encoder. Controlling t:hequanlization pr~s is Ue ntost direct way of ootUrol.lin.g' the titnte. This
`pa!1 ofl')O/JEC 1 Jl?Z sp«if1es anabstnctmod~l of Ue lmtfaint system(tlle VicJeoB~r~rillg Verifier) ill
`order to cmslrnin lbe maximum variability in the number of bits tbl'll. are IJS(ld fora giv~n picture. This
`ensures tha a bi!sl.lealll can be dc.co:led -with a buffer of leo own si2e.
`
`Al this stage, the axied Iep'~latiou of the pict11rebas teen generared. The final step io lheenccxJer is to
`regerente J-Pict1res and P-Picrures by de.codlng lhe data~ that trey can h: wed as cefereoce pictures fO!(cid:173)
`sub86JUent eoocdmg. lle qumtlzed COefficielts aredequanlized 2nd an inverse 8 by 8 OCf i~ perfi>rmed O.D
`each block. Tbepredlctbnenor!'ign24 produced is 1ben added back 10 !he prediction signal aDd limil.t'tl lO
`the required range ID give a dec<xb;l reretenc.e picllre.
`
`0 .4 Decoding
`
`Decoorug is the ime~Se of !he ereooi.ng operatioo. 1t is considr.rdbly simple• thw encodillt as mere is no
`need to petfoiDl motion l'St:inatioo and there are m:u~y fe.,.ec optiots. The decoding p-ocess is defiled by
`tbl.'l part oi IS01IEC 11172. The dtscrip:iat tbal fOllows is a "ery brief overview of one possible way of
`rleooding a bitslrealli. Other decoders with diJlerentarchilectures are possible. Fiture 4 slx>ws tbe main
`funcliora.l blodls.
`
`Page 7 of 124
`
`vli
`
`
`
`ISO/IEC 11172-2: 1993 (E)
`
`©_ISO/IEC
`
`+ R~
`
`oil pol pitrures
`
`Motion VertotS.
`
`OCT 1 is inversedis:retec:osine tmnsfCJlJl
`isdw.juanlizmoo
`Q- l
`MUX-1 is demtJtiplexing
`VI.D
`is variable lenglh decOOinB
`
`Figure 4 -- .Basic video dtCOcler blcc.k diagram
`
`For fi~ed-rntc appHcations, the cba.rmel fills a FIFO bulfer at a COOStult rate wilh lhe cooed bitstream.. The
`decoder reads Ibis bllfftr an<l dtcOOes tbe dal!\ elemoots in tbe bitstJeal'lliCO)rdi'Dg 10 tle deftned symu_
`As the d«XX!et ream tbe bitstream, it identifies the start or a oode<i piorure am the& U.e type. of the picture.
`It decodes eacb maaoblQck in tbe pctDTe ill tlnl- The mu:roblodc type aad the rnouoo vecrors. if present.
`are used 10 tonstruct a prediction of !be CU1Ult rotar>blodc lased on past lind fwuc refereace pictures thai
`tlave beeD siOred in lled6:ocler. The <:<efficient d.ataare deccded 31d dequanliud &ell & by 8 block of
`coefficieDL data u rramfortned by an inverse OCT (specified in SllleJt A), :mel dl~ resull is oored ro tbe
`pmdiction StgJW and tmiled 1.0 the defined ran&e-
`After all the macroblodc.:s in tbe pcture have been proces..«ed, t.be pick.lrt h\s been llXOnstructed If it is ln [(cid:173)
`picture or a P-picnue it is a refermce (t.Cllln! fa- sutnequent pictures .Uld is stored, replacing the oldest sr.otUI
`efereace picture_ Defore t.bo pi;tDres cue displayed tlley may Deed lO be te-Ordered from lhe ca:led order CJ>
`lbeir natt:ral display Older_ Afler reordui!g, tile pic lUteS are a"Yailab le. in di ~ita! Coma, for post-processing
`and disphy in aoy manner tbal t.be applicatil){l.cbooses.
`
`0 . 5 Structure of the coded video bitstream
`
`This part of JSOiiEC 11172 specifies a syntax ror a cod~ vidw bit:;trearn_ This iyntat contains six layers,
`ead1 of which either supports 1\ signal processing ora ~yst.em fuoctiol'!:
`
`L .aiVUS or tbe syntax
`Se~ooxe b.yer
`Group of pictwes layer
`Plct:utt layer
`Slice layer
`Maaoblocklaycr
`Block la}ef
`
`Ftmdioa
`Raadom actess unit: context
`Randon aecess unic vic:Jco
`Primary codin~ uail
`Reeyrchtooiution 11nit
`Motion rompeu;a.lioD mit
`ocr u11it
`
`o .6 Features supported by the algorithm
`Applicauoos usiag comJreSSe41 viclw on digi 121 stor:\ge media nec:d t1 k we 10 p!rfoorl a namber or
`operatiOIIS in OOdition to normal forward playback of ~ lieq8er.ce. The coded bitst.reaDl br:s beEn deslgoed
`10 support a number rL ~ese operalioas_
`
`.,iii
`Page 8 of 124
`
`
`
`<0 ISO/IEC
`
`ISO/IEC 11172·2: 1993 (E)
`
`0 . 6 . I Random access
`
`R.aadcm a~ is an ess:ntial feature fa "ideo on a ~a~e me:lian. Rarldom atrelS IC:Qllircs thai any
`picture can be <kcOOed Q a linied amruot cf lime. [t implies the etisteoce ot acctS.<; points ill the
`bi~ -!hat I& ~enu of ir.foorutiim tllat are identif"able md can bedeooded witllotJt reference to Olber
`segmett.s of <hUI. A ~ng of two nndom .&CC~:SS points (lrtra-Pictures) per se:cood c:,n be 3Clicvc1
`wilhoat signifJant toss of pi clUe qwlity _
`
`0 .6. ~ Fast $Oarch
`
`Depe11ding on tile storage medilllXI. il is p<lSSJble to scan tbe :u:.cess poiniS in a coded bllStream (wUh the
`help c:L an applicatioo-speci.fic cirectory m o b a knowled&e heyood the scope of Ibis part ofiSOIIEC
`11172) to olmln a fast-fonv:ud and fasl-fe\'~ playback effect
`
`0. 6 . 3 Reverse playback
`
`Some applicalioos may fe(]Uire the video signal tO be playe<l in rm·~ llrder. This um be achieved in a
`decoder by using memory 10 store en lire gro~s of pict~IeS aftet. they have btE.n decoded before bemg
`displayed in reverse Ofder. An wc:xtec can rnake this feature ea;.ier hy red.iang tre k.ngt.b Qf groups o(
`picrurcs.
`
`0 . 6. 4 Error robustness
`
`Most 4igit.al sroage media am commJrication cbant:eb are not error-Ucc. Appropriate chrumcl codnt
`9:be.mes sho~d be. u~d and ue beyood d:Je ~pe of this part of lSO'lEC 11172. Nevcttllel!!sS the
`compes,ioo scbexne definw in tbis pan c:K ISO'IEC-1) J 12 ii robusiiO residual errors. The slice siTUcture
`allows a decoder coro::ova afietadala enor and to rcsyochrcnizc its decodi~. Therefore, bit errors intbe
`compresled dal3 will cause eaors n thedecOOed p ct.Jres b be limited n uea.. Decoders may lie able _, 11se
`concealmeut strarepes to dis!Uisetheseerroxs.
`
`0 . 6 . 5 Editi ng
`
`There is a mnllict between lbe req1liru:lrnfoc l.igb mding eErlCieocy aodcasyeditiflg. lhccodingstru~
`and S)nWC have not teen designed witb tbe primary !lim of ~lifying editing at MY picture. Nev~less
`a nambef of featllres bave beEn inclured tbat ernlble ed~tiJI~; of Coded rum.
`
`Page 9 of 124
`
`
`
`This pa~e intenfionally left blank
`
`Page 10 of 124
`
`
`
`INTERNATIONAL STANDARD <0 50/IEC
`
`JSDIIEC 11172·2: 1993 (E)
`
`Information technology -Coding of moving
`pictures and associated audio for digital storage
`media at up to about 1,5 Mbit/s
`
`Part 2:
`Video
`
`Section 1: General
`1.1 Scope
`
`This pan ofiSorJEC llln specifus lbe coded reuereotatio11 or video for lligilal s10rage merliaand
`$J)(cili~ the <lecodins wocess. The representation supports normal speed forward playback, as wed as
`speci21 fJnctioos such M r.n<bfJ' aocess, fast forward playlllct, fasl reve~re playbad: , normal speed reverse
`playb1ct, pause. and still pictures. This plirt ofiSOIIEC 11172 ls COOJpati>le v.i tb stanilard 525- and 625-
`tine tcleYi.sioo foruats, and it provides fl~ibility for use wilb personal computes- md worksution di!jplay~.
`
`lSOIIEC 11172 is prim<srily apJ!licable 10 digital s10nge meda supporting a contiauous cnmfer rare up to
`about 1,5 Mbit/s, such as COlllplCIDisc. Digital Audo T~ aad ~n:~gnetic llanl dsb. N~venhe~ ileal
`be used more wnely 1h<ll dill bccalse of t..be geaeric apprmcl lalc"en. Tile n.orat;e nedi!i may be diredly
`coanccltd 10 the dccaJer, or via carununications means such as busses, LANs, or telecarommicalious
`Jinks. Thi~ pan oflSDIIEC 11172 is in !ended for non-mterllX:e4 video formats having appTOi'ti.t.nately 288
`Jill~ of 352 pels aDd picture r<~tes arou•d 24 H t to 30 llz.
`
`1 .2 Normative references
`
`The followin& Intermtional Standanli contain fiO'Iisions which. ~rougtJ refl!rence in Ltlis text, constitute
`provisiros of this part r1 ISOIIEC 111 TL AI ti.Je lime of publicatloo, tl~,e editioos indicated were valid.
`All scandatds are subject to rnvisjoo, and par1ies to ap-er.lnents b.'\Sed on this p.'Vl c:f ISO'IEC 11171 are
`etlcourn:!,ed to invcstigrue b e.possibifity of applying lhe mosr receat editions of the st:mllards "indicated
`beJoyr_ M~mbers of JEC and ISO mai.nlai.D reghters of cun-e11ll)· valid lar.ernaticnal S~mdad$.
`
`ISOnEC 11172-1:1993/njJni'WliOtttecluwtogy - Codin: ~~w~irtz picl11rt.r an.d aJsocitJied audio{ordigilfll
`storage mulia at up 10 aboml ,5 MbiJIJ - Part r .- System.
`rso/IEC 11172·3: 19'J3 fnjmnntion reclllology- Coding t:f fr.Dl'ing p«:urts an.d associare.d autio fo r dlgiJal
`rroraze nudia 01 up 10 aboZJt 1.5 Mbdls- Part J Audio.
`
`CCIR Recomnmdatioo 601-2 Encodinz parill7le1Usofdigilolttlevl.SI.Ot:[ar swdi.os.
`
`CCIR Rcpi>n 614-4 Chttrccurirrics of system.s j:Jr nwnoc/lr~ cud colOur televiJio• .
`
`CCIR R.ecormnendal:ion C48 Recording oj auiio .sigwls.
`CCIR ~n 955-2 SowuJ broadciJstiTtg by satelliJe }or pon ablt aut mob11e reuirers, 111dwling Alli!Lt N
`Sumraary kscriptioll of AdviU!Ced U gitalSJsum If.
`
`CCITr RecommeaW.liQn f.1 7 Pu-emphasis u.!ed on & uMI-PngrOJnnle Circuils.
`
`Page 11 of 124
`
`1
`
`
`
`ISO/EC 11172.-2: 1993 (E)
`
`Q &SO/IEC
`
`IEEE Draft StallCWd Pll801D21fYJOSpecijicalionfor lht if'4/)len.entalio11 of8r 8 invem discree cosine
`lTQ/ISjJ f7ll ~.
`
`lEC publication 90S:1987 CV Digital Audic Sys~en~.
`
`2
`Page 12 of 124
`
`
`
`«> ISO/IEC
`
`1SOI1EC 11 172-2: 1993 (E]
`
`SecOon 2: Technical eleme-nts
`2.1 Definitions
`
`For the JlWl)Oses of ISOfiEC 1 117'2, ibt following dt.finitious apply. (f spedfi::; to a put. ll!i~ is oo;ed in
`square bractets.
`
`2.1.1 ac coefflelent I video]: Any DCT coefficient fc:x whicb the frequency in one or both dimensions
`is non-zero.
`
`2.1 .~ acc,ess en it ls)Sl.em}: In the care of comp~ audw an access uoit is an audio acce&S an it. lo
`t11e case of compressed videoAAac:ce_'8 unit is lte roredreprt>Senl<tioo or a pictllre.
`
`2. t .J adaptive ie&mentation (audio): A mt:division of the digil.-tl repnse11tation of rua oodio si&na.l
`ia variable ~gments of 1in1e.
`
`:2.1 .~ adaptive bit aUoeation [audio!; The ~ignment of l:its to subballd~ in a lill\e and freque~~c~
`varying fa&bion acconling 1o a IJlY<:boamustic model.
`
`2 1.5 adaptive nolst 31Jocation (audio}: The assignment of coding noi ~o to freq11cncy bands in a
`line and freqoeocy v~ing fastiiln acco:di\g to a ps)'Cboacoostic mcdd.
`2.1.0 alias [a1.1.dio]: M.irrcced sig:aal component resultiJJg rccm sub--Nyquist sunpliog.
`
`2.1.7 1uuJ)'Sis filterloaak [audio!: Fiketbmlc in tre encoder lbat ~forms ~ broadballd PCM aqdlo
`signal into a set of subsampled subbaud samples.
`
`21.8 •udio access w it £audio]; Foc-Layers 1 and nan audio a~ ulli. ii dertOed as ltae smillest
`parto( \tle rocoded bi!Slnam wbicbC3l be decoded by it>elf, wheredeocded neMS "fuUy reconstnJCtcd
`souad·. For layer 01 an au <flO access llllit is part or be bitstream 11\at is decodable willllbe u~ of
`pe~iwsly acquiaed IJ:ICW i:Dfonmlion.
`
`2.1.~ audio buffer [ 2udio}: A buffer in the sy;tem tal get decoder f<» stonge of a>mpressee audio daLa..
`
`2.1.10 udi.o sequence fuudioJ: A non-inttmpleel series of nudic frame s i'l \llbicb lbe iollowillg
`pararne~ers uenot dlmged:
`- ID
`- l.ayCf'
`- Sampling ITequf:IKY
`- f"Qr Ulyer I run ll; B itr<.~te index
`
`2.1.11 baclc:"ward motion vector (vi.teof: A motion vec tor lbal i s used for 11:aotion c a.npensatioq
`from a refere~~ce picll.lre at llater lime in display order.
`
`2.LU Buk [1n.Jio): Unit of critical band m e. The Blltk scale is a non-linear mapping of tbe freq ueocy
`scale over the audio rnn~ closely cone.<.pondin1l with the irequency selt.elivify of the buman ea- acr<6S tile
`OOnd.
`
`2.1 .13 bidinctionally predictive-coded pic ture; B-piclure (vide.oj; A. pic;ture lhaJ. is caied
`using motim c:ompen~tl:d prulictioo from a~ <nd!oc: fitwe reference picture.
`
`2.1.14 b itntt: The mte It vrbich tbe. compressed bitslream is dtlive1ed from tl:e slClfllte meditun to tle
`iapct of a decoder.
`
`2.1.15 blbdc cmnpancfmg [audio): Noonalli.ing of U1e dig rat repr>..senru.ioo of aa audio signM
`vriuin a ccrtam lime period.
`2.1.16 hlock l~deol : All &-row by 8-cohm• «tbor,onal b lodc or pels.
`2.1.17 boW\d l ~u4io j1 Tbe lowest mbband ia wb.icb ioce usily stereo codtn& i4 u;ed .
`
`Page 13 of 124
`
`
`
`ISO/IEC 11172-~ 19e3 (E)
`
`~ ISOJlEC
`
`t .1.18 byt~ •ligntd: A bi iD ' coded biiSI.IUllll IS byec-ili,ncd if Its position ~ a m11ltiple of &~bits
`irom 1M fust bit bJ the unam.
`
`1.1.19 ~yte; Sequerce or 8-bits .
`
`.!.1.2t da:m11el: A d1gital medium that stoces or l:t'Msports an TSOiiOC 1l l72 streaD
`
`l.l-21 <hannel [uucliol: ne left andri&bl cba-nelso(a s~eo sgaal
`
`~.1.2~ cbtoh•iaat•ct (component) ['ri:deoJ: A mntrix, bl.x:l. or ringle pel rcpreseatiag ore of lbe
`two colour difference si!nals related 10 tie prin:w)· colwn i n tht: 11la'\ner defined in COR Rec 501 The
`symbols us:c I for tlc colOur tiffert:l1re s~ rue Cr m4 Cb.
`
`~.1.23 codefl audio bi13otrum !audlo]: A coded repte$eruation o i an audio sitn21 as specified in
`ISO!IEC 11172-3.
`
`l .l.:U melee! vlch :o biL"trum l•ideol: A coded representation ot a series of erne or more. pkturos a.~
`specifie4 in tb.is part of ISOIIEC l1172.
`
`Z.l.:ZS c:oded order !video): The order in wtlicb !be pictnC!I ere stored md decOOoo. 'This 01der b not
`aet:eSWily t.bc !WOO as lhe display or.Jea:
`
`2..1.20 coded represelltation: A daLa element as represented in irs eflc«fe.1 form.
`
`2 .1.27 <odlng parameten (vi• e.): Tbe set o f Ufoee·ddinable paralllCiel$ that cllan.cterize a eoded vidw
`bitstream. Dtstteansare <bataaelised \y coding IX'lram~ ~an darnoerised by dlebitstrea~na
`tba. they are capable of decodi.n&.
`
`2.1.23 wli:lpooenl [vid~): Amalrlx. bklck or W1g1e p!l frcm ore or the: thre: matrices (JilutiflUl(e
`aod two cbotninaact:) tAaJ Emk_e ~ & ptnue
`2.1.21» tOII'I(II'tiSinrc Re.ci}Ctioo in the nani>er a bits used to represent aJi iiCill of dal2.
`
`Z.1 .30 coi\Stant bltrate coded video [video]: A compressed vldoo hiutrtalll wilb a CODSI.IUl!
`ave:race biii'8Je.
`
`2.1.31 tOf\.'lt.nl bitrnte: Operation where the bit.rate is const.aot iretn sw1. to f'loisb cf lbe compre.~sed
`biiStream,
`2.] .32 CODStrained p aranaders ("Video): The values a tbt:. $Ot of C.odin& p:ll'aiDefCf'S defUJed lTt
`2.4.3.2.
`
`2.1.33 cowtrai•e• systean parannter s tream (CSPS) (sy~leml : Au ISOJtEC 111 72
`ma.J~le"lted stream (()!' whid the couttaillts defined in 2.~.6 o f ISOIJFC t 1172,- 1 apply.
`
`2.1..34 CRC: Cycli( red1ndrulcy codt.
`2.1..35 critical band rau lndit]: Psychoacouslic fimctian r frequency. Al a gjven aadible
`frequency it is pr-opo:tionalro tben&.mh:t of critical baad6 below !hat freque11cy. The units of !he critical
`baod ere scale 2re1Wb.
`
`2.1..36 crtial band [aucUu): PsycboaCAJustic trea.IUIC in we spulml dom~ :vbick conesponds to lbe
`freque~y selectivity of tre hunao e.u-. Ibis selectivity is expre.>sed Wi B~tic
`
`1.137 •ata element An item of data~ re~ before encodillg and ait.cr d::codl:lg.
`
`2.1.38 dC·COellidel!t [ virl~:nJ• The OCT coetfideot ror wh.i('b the fJ"eqtJCUC! i; 7,UO in botll
`d immsi<lns.
`
`4
`Page 14 of 124
`
`
`
`~ ISOIIEC
`
`ISO/IEC 1 1 172·2~ 1993 (E)
`
`2.1.39 cJc:.cccJcd p icture; D-pic:ture [~ideo ); A pi cwrt. l.bat is eodt'tl osiog only infomu.tion from
`itself. Of !he DCT coefficiens in dle cxx1ed representation, only lbe dc-mcfficieats are prescnl
`
`2. L40 OCT c:oefficle•t: Tie ampl:itude of a specific cosioe basis finctioa.
`
`2.1.41 i ecoded stream: lbe dx<;lded rcconstrnclioa of a compressed bit.u.rcam.
`
`2-1.4Z decoier i.Dpa.t bufft"r fvideoj : 1)e first-in fi.rst-OUI (FIFO) buffer specified in lhe video
`buffering Tcrifia.
`
`2. 1.43 l ecode.r lrtl\lt rs1te (v id~): The dau n.le specified in lhe Vldeo bulferin& ~rifia md ellccdcd
`in~ coded vid.oo bilStteam.
`
`2.1.44 llec.oc:ler. Ao embc.Jdimmt of a deoodint )l'ocess.
`
`2. 1.45 decoding (process): The l)tOCe>S defiued in ISO/lEC 11 172 tint reads an i11put cored bitstrec:m
`arx:l producesdecodcrl pictures or au<lio samples.
`
`2.1.46 deco<ling ti1ne-starnp; DTS !system!: A field tbal Jllity b.: present lu a patlcet beailer that
`indicates the time tlat an access unit is decoded in tbesyslem t.arjetdecodet.
`
`2.1.47 de-emp hasis J ;~ u d iol~ Filtering applitXI to AA audio signal lifter stomgl! or transmission to undo
`a linear dhtortion \iu~: to emphasis.
`
`2.1.48 de41aantization lvidw ]: The process c:L cscaling tbe qu31ltlze.1 DCI coe(f'lcient> after their
`representation in lbe biutream bas bew decoded and before tbey a.re presented 10 lbe inve~SCDCT.
`
`2. 1.49 d~ib~ storage media; DSM: A dig iial storage 0 1 uanunission revice or system.
`
`l.l.SO d~cr