CA2003136A1 - High definition b-mac television signal transmission system - Google Patents

Abstract

HIGH DEFINITION B-MAC
TELEVISION SIGNAL TRANSMISSION SYSTEM

ABSTRACT OF THE DISCLOSURE
Apparatus for encoding a high definition multiplexed analog components television signal by decimating predetermined samples for transmission comprises separate processing paths (Fig. 17 and 18) far luminance and chrominance component signals respectively. The separately processed luminance and chrominance components signals are multiplexed into a composite signal for transmission via a skew-symmetric filter portion (1710) shared by both processing paths. A
complimentary skew-symmetric filter portion (1901) to the filter portion of the encoder apparatus is associated with a decoder coupled to a high definition television receiver, Predetermined samples are diagonally filtered and horizontal resolution information folded into the signal. Vertical resolution is improved through scan conversion at the receiver.

Description

~03~36 ~IGH l~lNITION B-MAC
TELEVISION SIGNAL TRANSMISSION SYSTEM
BACKGROUN~ OF THE INV~NTION
. Teohnical Field The invention relates to the field of televlsion signal transmis-slon systems an~, in particular, tO a tele~rision s~gnal transm~sion system ior transmitting R signal providin~ a higher resolution image than ~ transmitted under standard r~solutlon National Television Sub-committee (NTSC) or European formats.

2. D~cription of the Relevant Art There is a growlng interest in the transmission of television signals whi~h increase picture definltlon ln both the horizontal and vertical dlmensions. In the vertical dimenston, such a signal may have as many as twice the number o~ lines in comparison with exist-ing standards while irl the horizontal dimension, t~e number of picture elements per line is ll~e~ se ~n~reas~d. As a result of provldlng Stan-dard horizontal and vertical resolution, there are adverse ~ffects from provldlng wide screen displays of a transmltted signal. A viewer of a standard si~nal may complain of the fuzzy or unclear quality of the imag~ 1~ vlewed from a relatively olose pro~lmity. Th~se advewe efreot~ ~ ov~raome by a higher resolution lmage ~ut exlstent trans~
~nisBion systesns are not read~ly ad~ptable to transmlttlng a hlgh reso-lutlon lma~e.
on~ solution to the problem o~ pro~ldlng a high r~olutlon lmag~ wlthout increasin~ the require~ bandw~dth for signal transmi~-sion is descr~ in U.S. patent application Serial No. Og2,305 ~ilRd September a, 1937. A high re~olution signal is diagonally filtered and ~lternate samples deeimated on alternate lines according to dlglt~l sampllng teehniques to leave a figure-o~-~lve or qulncunx pattern or - 2 - Z00:~36 plcture samples. Odd line samples are added to even llne samples form~ng a llne summation signal. ~y m~ans of skew-symn etric filter-lng techniques, high resolution horizontal inIormation reglons nI the slgnal which normally carry diagonal in~ormation of marglnal value.
The loss of diagonal ~nformation from a transmitted sign~l does not cauSe percep~ible impairment to the original high resolution image input at the transmLtter. Only video line stores are re~quired at the transmitter. No field stores are r~quired. Furthermore, the digital filterlng requ~red at a receiver is relat~vely inexpenslve ln comparison with prior art interpolation technlques.
Another solution to the problem os transmittlng a hi~h r~solu-~ion image is to transmit a standard televi~ion signal and to ~reate and tran~mit a so-called augmentation channel. In accordance wlth the first solution des~ribed above, a new receiver ~s required for pro-cessing the received signal of standard bandwidth. In accordance with this solution, no change in receiver cir~uitry is required ~or receiving and displaying a standard resolution imaget However, sepa-rate adapter cireuitry is required for recelvlng the augmentation ¢hannel containing high resolution data and tor reinstitutl~g the high re~olution data into the standard resolution image to provlde a hlgh resolution image.
A metho~ and apparatu~ for increasin~ the deSlnition of an NTSC video signal using an augmentation ~hannel is descrlbed in U.S.
patent appllcation Serial No. 228,274 filed August 4, 1g88. According to the approach talcen in that applicatlon both a line sumsnasion signal and a line difference signal are formed irom a high resolution televi-slon sienal. B~ retrerse alternate sampling o~ the llne summation sig~
nal at mid video signal ~andwldth, high resolutlon horizontal informa-tion may be translated to ~aseband an~ added to hlgh resolutton vertl-cal In~ormation ~rom the line dif~erence slgnal and transmltted togother as the re~ulred augmentation chans~el~
Neither o~ these approaçhes, however, proYideS A solution to the problem of transmltt~ng a high defin~t~on multiplexed analo~ com~
ponen~s (MAC) television signal with Incr~ased horizontal and v~rtlca~
resolutlon as compared with ~tandard resolution television ~i~nals but ., . .

. . ~.
.

which does not requlre modification of ex~stlng MAC receiver/decod~
ers ~or receiving a standard resolution MA~ composite signal.
During the l9~û~s sub-nyquist sampling techniques were applied in the art of dlgital tQlevision by the present inv~ntor and calleagues at the Indepen~ent Broadcast Authority o~ the United Kingdom. Digi-tal techniques for the elimination of alias~ng are described ~n an arti-cle entitled IlAn Introduction to Sub-Nyqu~st Sampling" by K. H.
Barratt and the presen~ inventor appearing In the I.B.A. Technical Review at pages S-15. In a companion ar~ticle at pages 21-26, entitle~
"Dlgital Sub-Nyquist Filters~ ~y J. H. Taylor, c~mb ~llters ar~
described for down conversion and u~conversion o~ vldeo data appli-cable ln a PAL televislon signal en~ironment ~or optimizing signal sampllng. These anti-allasing digital sampllng techntques provid~ a developmental basis for Improving horizontal resolutiont however, there still remains a requirement in the art for a method and appara-tW for lmproving both horizontal and vertical resolution of a trans-mitte~ multlplexed analog components television signal, SUMMARY OF THE IN~tENTION
The present invention relate~ to a method and apparatus for transmltting and r3cei~1ing a high definition multiplexed analog com-ponents (MAC) television s~gnal. Accordlng to the B-type MAC trans-miS3ion format for compo~ite signal transm~sslon, the ~rideo slgnal ~s carried within an acti~e line pe~iod whlle all other signals comprising at least audiat control data, utility data and teletext are transmitted durlng a line blanking period or a longer ~teld blanl~ing perioa. Sepa-rate lwninance and chrominance ~i~nals are dlgltally sampled, com~
pre~ed and transmltted durlng separate por~ions of a vld~o llne si~
nal. In ac~ordance wlth the well known ~-MA~ ~ormat, lumin~nce sample~ ar~ compressed tor transmlssion at a ratlo o~ 3:2 wh~
chrominance is comp~essed at a ratlo of 3:1. Chrominanc~ lnlorma-tion is translated into U and V ~omponents, each component ~elr,g tran~mitted e-tery other llne.
To accompli~h a tran~m~sion o~ hi~h resolution vld~o lnforma-tlon within the ~undaries o~ multiplexed anal~ component formats ~enerally and In accordan~e with the method of the pr~nt invention .~
. . ~ , - :
- .. ,, ., ~

2003~36 a folding o~ high horizontal resolutlorl information is accomplished in~o the high frequency diagonal ~omponents o~ the sampled video ~l~n~l At hasehan~ frPrlllPnniP"~ hPlnw ,S MHZ (~ M~Z when t~me-compressed 3:2 according to the ~ormat), the spectrum ls unmadifled.
Thus, for example, slnce standard B-MAC ~ecoders are typically equipped with 6.3 MHz passband lowpass filters at their input, the ~olded high resolution information d~es not affect recept1on. The additional transmltted information at high rr~qllene~ hnply ~locked an~ ignored.
In particular, acc~rding to the present transmisslon method, a high d~finitiotl analog television slgnal is ~lrst orthogonally sampled at 28 MHz ~a rate of eight times the color subcarrier o~ 3.5B M}~z or 8 Fsc). As a resul~ a two dlmensional sample spectrum Ig achieved which Is then passed through a diagonal digital filter which decreas~
the diagonal ~requency response but which decrease i~ pra~tlcally imperc~ptlble to a viewer.
The diagonally filtered data ~s then decimated by discardlng alternate samples on alternate llnes. As a result, a flOEure-of-five or quincunx pattern o~ samples remains. As a result of the decimation o~ alternate 3amples, the base~and spectrum remalns unchanged but high resolution repeat spectrums comprising horizontal and vertical resolution components exlst at half the sampling frequency and at th8 sampllng frequency. ~he repeat spectrums serve to ~old addltlonal re~olution into the baseband signal.
Tne samples of the ~ol~ed signal may then be converted to ana-log ~orm and passed through a low pass sk~w-symmetr1c filter o~n-te~d at seven megahertz or simllarly digltally f~ltered. Aooordlrlgly, h~h r~solutlon lntormatlon related to the horizontal ~mension ~9 rolded about a dlagonal axis at seven megahertz into the approxl-n~ately ~ive to se~en megahertz or hl~h ~requenc~ portion of the p~ssed b~eband signal. ~f~ectively, the high resolutlon lnîormatlon 18 traded for the diagonal information.
Accordlng to transmission apparatus oS the present ~nventlon, tne dlg~tal diagonal ~llter of the encoder may comprise separable horl-zontal ana vertlcal ~llters. The vertlcal filter at the transmltter may be very ~imple provi~ed the horizontal filter ~s suffic~ntly complex to achieve a 40d~ re5ec~ion in the stop band. For example, the hori-zontal fil~er at the transmitter (permi~ting a 0-5 MHz passband) may be at a complexity on th~ level of sixteen coef~icients. Besldes per-mitting a much s~mpler vertical ~ilter at the transmitter to achieve a dia~onal filter, il has also been found that as a result of employing a complex horizontal filter at the tran~mitter, a much less expensive and simpler diagonal filterlng arrangement may ~e employed at the receiver. In pa~ticuIar, a 5-MHz low-pass filter at the receiver need comprise only eight coefflcients. Wh~le the inv~ntion is described ln terms ot improvlng luminanoe horizontal detail, the t~chnlque and apparatus may be adap~ed f or improvlng chrominance horizontal detail.
An embodiment far improving chrominance horizontal detall applles slmilar principles to those applied for improving lumlnance hortzontal detail. However, an intentional lncrease in horizontal c~romlnarlce detail has been ~ound to be unnecessary for a ~-M~C
signal, allowing a simpler technique for chromlnance transmlsslon.
A vertical filter interpolator receives at its input a 525 line, L:1 non-interlaced signal, a lOS~ line, 2:1 interlaced ~ignal or an 1125 lin~ 2:1 Interlaced ~ignal. The slgnal is approprlately filtered ln a vert~cal dlrection and an output provided to a 4:~ llne decimatlon cir-au~t. The output of the line dectmation circuit ~s ~iltered ln a hori-zontal dimenslon about a center ~requency of 2.5 ~eg~hertz and the hlgh frequency output samples provided to a multiplexer. The multl-plexer comblnes the chrominance, luminance and any data~audlo sig-ns~S for transm~ssion. Howe~er, prior to transmission, the comblne~
hlgh det~tlon ~-MAC~ signal may be passed through a skaw-symmet-rlc ~lltor portlon c~ntered at 10.~ megahertz whlch in comblnation wlth a ¢omplimentary ~ilter portlon at a de~o~er eliminates aliadn~
in a d~cod~ slgnal. Con6~quently, tne sam~ skew~ym~etrlc ~ilter portlon o( the encoder may be shared ~or both luminance and ch~omlnanoe proce~slng.

. ~.. . .
, . .
.

~00313fi ~RIEF DESCRIPTION OP THE DRAWINGS
Figure l is a graphical dep~ctlon of verti~al versus horizontal definition for a high definition televlslon signal.
Figure 2 is a representation o~ an orthogonal sampllng grid for sampling the signaJ of ~igure l at 28.6 m~gahert~ (8 Fsc after prerilterlng at 9,~ megahertz).
Flgure 3 is graphical depiction of vertical versus horizontal definltlon as a r~sult of application of the orthogonal sampllng rld o2 Flgure 2 such that a baseband spectrum results as well as a repeat spectrum centered at the sampling ~requency.
Flgure 4 is ~ graphical depiction of vertical versus horizontal deflnition with ~iagonal information, a block of data, for example, between five an~ nlne megahertz, having been fllter~d from the ortho~on~lly ~mplod ~p4ntrllm~ c~ Fi~llrP ~
~ igure 5 is a representation of a sampllng grid at 14 MHz gen-erat~d by discarding alternate samples on alternat~ line~ to achieve a f~gure-o~-rive or quincunx sample pattern.
Flgure ~ i5 a ~raphical depLctlon of the result o~ the decimation Of alternate samples where, besides the repeat speCtrum at 28 MHz, two repeat spectrums at 14 ~Hz, half the initial samplinE rate are lntroduced.
Flgure la is a graphical dep~ction of trertical ver~u~ horlzontal resolutlon for showing the proces~ oi ~ilter~ng about a c~nter ~r~
quency o~ approximately 7 MHz, the filter having a skew-symmetric low-pass re~ponse and Figure ~b the characteristi~ amplltude vers~s frequency re~ponse.
Flgure 8 ~s a graphical ~eplctlon o~ a first step o~ proaesses accompil~h~ at a recelver, 3y app~ying a r~sampllng a~ fourteen megahertz uslng the ~igUre~ ive pattern shown in Flgure 5, the ~olded hlgh resolution horlzontal ~mens~on lnformat~on is returned to hlgh frequency. Figu~e 8a represents a flrst graphical deplction ot vertlcal versus horizontal de~inition and Figure 8b repre6ents a s 3cond graphtoal deplctlon o~ amplitude versus frequency showlng how ali~sing 18 elin~inated and lnrormation close to 7 MHz ls regenerated ~y the slcew~ymmetrlc characterlstlc o~ the filter, :

;, , 2003~3~i Figure 9 ls a graphical depiotlon o~ vertical versUs hori~ontal definltion showing the result of upconverting from 14 megahertz to 28 megahertz. upon d~agon~l filtering, a nuI] ror high dia~onal frequen-cles in the range of 5-~ megahertz is produced leaving a signal having high horizontal resolution but an irnperceptible sacrifice in diagonal inîormation.
Figure lo is a schematic diagram o~ apparatus of a transmitter ~or enco~ing a high definltion B-MAC television signal.
Figure 11 is a schematic diagram oi apparatus of a receiver for decoding a high clefinition B-MAC television slgnal.
Flgure 12 is a graphical representation oî the characteristic response oi the sixteen coe~ficient horizontal low-pass filter, inclu~-ing coetficient data, shown ln Figure 10.
Ftgure 13 is a graph'cal representation o~ the charact~ristic respon~ of the eight coefllcient horlzontal low-pass filter, includlng coefflcient data, shown in Figure 11.
Flgure 14 ~s a graph~cal represen~ation of vertical resolution in llnes per plcture height versus horizontal resolution in l~nes per plc-ture width showlng characteristics of the appllcation o~ t~o present invention in combination wtth conventional but proprietary scan con-vewion line cloub~ng technlques in B-MAC versu~ resu~ts of more expensive multlple field store technlques used in a 1125 Itne MUSE
slgnal transmlssion system.
~ i~ure 15 is a graphical representatlon of amplitu~ versus ~requency ànd hor~zontal versus vertlcal resolution for the three ill~
t~s applied in the present technique: diag~nal filterin~, prefilterlng and skew~ymn~ctrio ~iltering, Fl~ure 15a belng o~ amplltude versus ~quency and Flgure lSb belng of vertical ver8us horizontal rewlution.
Figure 16 is a graphical depiction o~ the two di~enslonal ¢on-tour response o~ the transmitted signal ~omp~able to Flgure 14, horl-zontal re~olution being traded for diagonal resolution ln a system accor~lng to the pr~sent ~nv~ntion.

. .. .
.,, .
, , , ,. . .
'"

' ':

Figure 17 is a l~lock schematic cliagram of luminance processing circuits at the location o~ a high defln~tlon B-M AC encoderin a~r-dance with the present invention.
FigUl'e 1.8 55 ~ block schematlc diagram of chrominance pro~
cesslng clr~ults at the location of a hl~h definition B-MAC encoderin accordance with the present invention.
Figure lg is a block schematic diagram Or processing at the location of a high def~nition B-MAC decoder in accordanoe wlth the presen~ lnventlon.
Figure 20 ~s a block schematlc diagram of a hi~h definition televlsion receiver for processing the output of the high definltion B-M~C decoder of Figur~ 19.
E`lgure 21 is a b~ock schematic dlagram o~ a high de~inition television receiver having multiple applications In ~ hlgh de~lnltion or standard resolutlon televLsion environment.

Referrlng to Figure 1 there is shown a high def~nition televl-slon signal graphic~lly depicted in terms of vertical versus horizontal resolutlon.
According to ~igure 1, the present method is a~sumed applica-ble to a hlgh definltlon 16:9 aspect-ratio picture scanned sequentially u~tng S25 llnes, the horizontal resolution being at least 945 lines at 9 megahertz. A sequential scan signa} o~ th~s type supports a vertic~l deflnitlon o~ ~80 line6.
Wlth th~s slgnal as an input, thls descriptlon refers ~y way of example to a new type of ~-MAC signal which carries lncreas~l re~
lutlon at the impercepti~le expense o~ diagonal resolutioni However, it may be llkewise applied in other multiplexe~ analog component vldeo 8ignal transm~sslon sgstems, The lncreased resolution ls ~old~d Into the hlgh vldeo frequencles. At ba~eband ~requencies ~elQw 5 M H~ (7 M Hz wh~n tlme-compressed in M A C) the spectrum Is unmodlfled. As known ~-M A C decoders have low-pass input ~ters having a pass band limited at 6.3 M Hz, doooder opera~on i~ unat-fected by the additlonal transmitted informatlon. on ~he other hand, a B-MAC~ ~ecoder in accordance wlth the present invention retrleves . .~ . ~ :' : .:. .

. :

g and decoàes the folded horizontal detail information and causes a hlgh resolution image l:o be displayed by a receiver.
A standard resolution s2s-line 2:1 interla~e video slgnal ~on-sists o~ two rields each containing 240 a~tive lines. ~Ines o~ every other (odd) field are spatially offset relativ~ tq lines of even flelds so that all 481) active lin~s are regularly spaced on the display screen.
In prlnciple, this line structure can carry a verti~al reso~utlon equal to ~0 lines for stati~ pictur~s. However, a normal interlaced ~i~pl~y d~ v~ vi~lw. Tlle r~ason ror tl~ ln the fact that only 240 lines are dlsplayed ~n each ileld, and the human eye/brain is expeoted to sum the two fiel~s and perceive all 480 llnes.
It cannot do this perfectly. The intensity of the first field perceived ~y the e~e/~rain has decreased ~o 50% of its initial value ~y the tlme that the seoond field arri~tes (1/60 seconds later). This has two consequences:
~ i) Line structure be~omes vislble in the display.
(li) trertical frequencies exceeding 24û lines are partlally alia~ed in the display.
The net result is that the perceived vertlcal resolutlon of a standard resolutton 525 line ~nterlaced ~lsplay lies somewhere between 480 llnes and 240 llnes. Th~ r¢duction from ~80 lines lS con-ventlonally des~ribe~ by introducing the concept of a so-called "Kell Factorl: ~
Perceived Resolutlon = 480 x 0.66 (Kell Factor) = 320 Line~
Por statlo plctures, Kell Factor may be entirely ellmlnated and r~olutlon restored to 480 llnes ~y displaying ~ll 480 llnes (~rom both odd and even Slelds) ln eaoh l/60 sHcond Iield perlod. Thls technlque Is known as scan ~onverslon. Applioatl~n Or the technique involve6 use of a iield store memory storing a~ 240 a~tive lines to movo ln~o~
mation between ilelds, and a display at twlce the normal llne ~r~
quency. However, the method can be directly applied only to statlc parts of the pi~ture, slnce sign~ficant motion can oocur between field~. Con~equently, a motlon detector ls also requlred so that Inter~
tleld lnterpalation can be us~d rOr st~tionary obJects, whlle . ~ .
. - - - . ,;. ~ - . . .................. .

....
~ .

2003i36 line-interpolation ~s used for moving obJ~cts. Scan conrersion tech~
niques employing adaptive field-store line doubling techniques thu~
achieve 48~ lines of vertlcal resolutinn for stati~ pictures and approx-Lrnately 320 lines in moving dynamic areas of an image.
Field store line-doubling is gaining acceptan~e as a standard method tor increasing vertical definition by TV-set manufac~urers.
Its main advantage is that it eliminates line structure and signifl-cantly lmproves picture quality wlthout requiring any a~itional information transmitted. Several manuIacturers of T~-sets and pro-Jectors are using proprletary line-doubling technlques, i~cludln~
Philips, Hitachi, Sony, l~egami, etc. ItS obvious advantagesare:
(i) The technique applies equally to component slgnals (luminance, ~hrominance) or NTS¢ signa~ received from any sourc~ including S-VHS VCRs.
(li) The technique is applied in the telev~slon receiver and therefore permits retention o~ a 525 Interlace connec-tion tO th~ set (NTSC or wideban~ YlC).
(iii) The field store in the TV set image pro~e¢tor can be sed ~or other consumer ~eatureS such aS piCture-in-pi~-ture and nolse reductIon.
(lv) The techn~que requires no additional transmitted infor-mation and allows any high definition television (HPTY) format to concentrate on the problem oi lncre~slng hor-lzontal definition.
In U.S~ patent application Serial No. 255,328, fil~d October 11, 1988 of Chrlstopher Birch entitled "Metho~ and Apparatus ior Improv-lng V~tical ~:)e~lnltion of a Televlslon Signal by Scan Converslon", ther~ ~ de~crlbed a technlque ~or lmproving the vertlcal re~olutlon oi a televl~lon slgnal by deve~oplng a plur~ty o~ alternatlve lnterpolated valu~ tor dlsplay and seleetlvely cho~irlg a parti~ular vRlue in acc~r-dance wlth tests of the video signal for shadlng, moven~ent and vertl-cal edge transitions, In order to solve the problem o~ increuing horl~ontal resolu-~on, the present high definitjon MAC system employs su~Nyqulst ~ampling (spectrum folding) to tr~de dlagonal resolutlon for lncreased ,~ -:

2003~36 hor~zontal resolutlon. The process will be descrl~ed In connectlon with Figures 2-9 and the apparatus at a transmitter or receiver will be descrlbed in connection wlt~ Figures In-ll. All frequencles quoted (bandwiclths ~nd sampling freqtlencies) will be referred to the uncompressed luminance signal of a B-~AC signal~ Equlvalent band wldths and sample frequencies in the ti~se-compresse~ (MAC) domain must be increased by a factor of 1.5 (3:2).
The 525-line 2:1 interlaced luminance signal is first ban~limited to ~ MHz using a low-pass analog ~ilter. Reîerring brieily to F~g. 10, the ~llter is described as an 8.7 MHz pre-fllter. Thls bandwidth (BW) is sufflciently broad to achieve a horizontal resolution of 945 lines per picture width ~PW), calculated as ~ollows: Lnes/PW =~W x Ac~ive Line x __ 2 = 94s Llnes~W
Irotal Llne Line Freq This signal ls in~tially sampled at 28.6 MHz (8 ~sc) using an orthogonal sampling grid ~s shown in Figure 2. As a result of orthogo-nal sampllng and in accordance wlth Flgure 3, a baseband spectrum as well as a repeat spectrum centered ~t the sampling frequency results, The ~aseband spectrum comprises high horizontal resolution compo-nontS ~t as high as g MHz or 945 llnes as calculated above.
A dia~onal d~gital ~llter ~s then applle~ which de¢reases the dlagon~l fr~quency respon~e (Step 2~. Separable horizontal and vertl-cal filters are employed ~or simpliclty as will be further d~crl~d in conne¢tion with th~ discussion o~ Figures 1~ and 11. Referring to Flgure 4, it may be seen that blocl~s of dlagonal (ho~lzontal versus vertical) informatlon are removed at horizontal ~requencies between flve and nlne megahertz in the basoband spectrum a~ well a~ the repeat SpectrUm.
Now, alternate sample~ are dlscar~ed leaving a "q~ncunx"
(flgure~ot-fl~te~ sample pattern at approximately 14 MHz ~4 fsc) with-out causing aliasing~ Accordlng to Flgure 5, alternate sample~ on alternate lines are removed.
The result of Step 3 ~s a sequen¢e of digital sample~ which n~w have th~ ~ompacted 2-dimenslonal spectrum ~hown in Figure 6 .

- 12 ~

wherein repeat spectrums ex~st at the fourteen megahertz sampling frequen¢y of the alternate sample decimation step.
Before transmiss~on, the horizontal resolutlon improvement lnlormation is folde~ Into the signal for transmission. For example, the samples may ~e converted to analog form and passed through a transmission fllter with specifie characterLstlcs. ~he analog trans-mlssion filter may ha~e sl~ew-$ymmetric low-pass respo~se which ~s -BdB at 7MHz. The result is shown ~n Figure 7a where hlgh re~olution in~ormatlon at 7-9 megahertz is transla~ed to fill the ~oid formed from diagonal flltering. Alternatively, a digital non~recurslve filter ma~ be applled aS wil~ be further described herein alleviatln~ a requirement ror digital to analog con1Jers~on or upcon~erslon to 28 megahertz sampling~
Etfectively, horizontal resolution between 1 MHz and 9 MHz has been ~olded around 7 MHz and replaces diagonal resolution b~tween 5 MHz and 7 MHz. Accor~in~ to Flgure 7b, the filt~r charac-te~istic response is shown at ~dB attenuation at 7 MHz. Note that in multiplexed analog component (MAC) transmission, the transmiSsion ~llter will be skew-symmetrlc about 1.5 x.7 MHz due to the time com-pre~sion factor (3:2). The sl~nal Is now ready ~or transmisslon, which may be consi~ered step 5.
When the slgnal is received, it iS resampled at 14 MHz ~ sing the altern8te llne qulncunx (I'lgure-of-five) sampling pattern ~step 6). By resampling at fourteen megahertz, a translation of the hlgh frequency informatlon occur~ to 7-9 MHz according to Flgure 8a.
The re;ampllng process thus regenerates the horlzontal energy b~tween 7 MHz and 9 MHz. Referring to Figure ~b, los~e8 in the transml~slon ~llter around ~ MHz are also precis~ly componsated by the alla~ term A, provided th~ transmittlng filter ln oombinatlon wlth the decoder pracess ha8 a skew-~ymmetrlc response~ A~ ~ 1. In other words, the alias term introdu~d is necessarily can~elled no matter where the amplltude ~ ls measured in the overhpplng dashed l~ne, solld l~ne area proxlmate to ~I MHz.
The spectrum of Figure 8 applies to th~ 14 MHz sequence ~
qulncunx samples. What may remaln is to upconv~rt to 2~ MHz, ~: ~
:
.

, 2003i~6 introduce a 2-dimensional ~ilter to remove the remalninB~ energy on the diagonal, and to ~andlimit the signal to 9 MHz.
A dlagonal ~i~ter (which produ~es a null for high vertloal fre-querlcies in the range of S-~ MHz) cannot be implemented at the 14 MHz sample rste. Therefore, the up-conversion o~ sample rate to 2~
r~qHz occurs as a part of the digital filtering process. According to Figure 9, the upconversion and diagonal filtering re~ults in a null at dlagonal frequencies and improved horizonta~ resolution.
Step 7 results in a sequence of samples at 28 MHz with an orthoEonal sampling grid. They carry a spectrum with horlzontal res-o~ution of 9 MHz, with no aliasing. These samples are available for dlrect conversion to analog form. A~ter bandl~mitSng to 9 MHs, the analog signal may be displayed on a high def~nition recelver.
Alternatively, the samples may be paæsed directly to a ~nown but proprietary ~leld-store scan converter for llne-dou~llng to incr~ase the vertical de~initlon suoh as the apparat~ de~cribed b~
U.S. appllcation Serial No. 2S5,2~8, entitled "Metho~ an~ Apparatus ~or Improvlng Vertical De~inition of a Television Signal bg Scan Con-version" of Chrlstopher Blrch filed Oatober 11, 1988 and inoorporated hereln by re~erence. The result o~ the sCan ~onv~rslon is a 525 s~quential-scan signal samplec~ at 56 ~IHz (2 x 28 MHZ) and carrylng horizontal luminance resolution up to 18 M~z. (The line doubl~ng pro~
ce~;s halves the active line period and double~ ~oth the sampllng ~re-quency and the vldeo bandwidth).
The described process provide~ a very cost effectl~re lmple-mentatton Sn a multlplexed analog comp~nent signal transm~sslon sys-ton~ wi~l be described in connection with block schenlatlc dia-grams Plgure~ 10-11 of the encoder and the decoder r~specti~Joly of tho pr~ont apparatu~.
Transmltter apparatus accordln~ to FlEure 10 performs all st0ps Of the above descri~d process ~Ut for transmlsslon:
step 1, 28 MHz orthogonal sampling after appllc8t~0n OI an 8.7 MHz p~filter;
step ~, dlgital diagonal filtering;
~ tep 3, 28 MH2 - ~4 MH~ alternate sample decimat~on; and ~ ................ .
~!

Z003i~6 step 4, digital to analog conversion and skew-symmetrlc trans-mlsslon flltering.
~ lthough it may ~ppear that the diagonal filter must be imple rnented at a sample rate of 28 r~lH~ the ract th~t samples are to be discarded at step 3 allows a simpl~fl~ation, Figure 10 shows a configu-ràtion In ~hich the main elements of the dlgital filter can be imple-mented at a 1~ MHz sample raCe.
The diagonal filter ls implemented as two separable (horlzontal and vertical) filters. It has been ~ound that the vertical ~ilter can be very s~mple i.e., a line store lOl and ad~er 102 prov~ded the horizontal filter 103 achieves approximately a 4~ dB rej¢ction ln th~ ~top ~and.
To accomplish such a rejectlon, the horizontal ~ilter 103 (5 MHz low psss) employs 16 coefficients ~t 28 M~z. ~he design of the horizontal filter 103 is a symmetrical non-recurslv~ filter which has ~aen optt-mized wSth 9-bit coefficient values. The response and the coeffi-~ients are presented in Figure 12.
According to Flg. lO, switch 100 sw1tches alternate samples of a pre-filtered 28 MHz sampling slgnal into two 14 MHz paths.
According to the upper path, the samples are vertlcally îiltered and horizontally flltered. The lower path ~ su~strated ~rom the upper p~th at adder 102 whll~ the upper path is added to the low~r path at adder 104. ~t the output of filter 103 is shown a low p~ comblng characterlstlc ~vith zero energy at zero ~requency while at the Output of adder 104 Is shown a low frequency ~ombing charactQristlo (solld llne) wlth energy at zero tre~uency extending to 9 MHz and ~ high frequen~y a~ased character~tic (dotted line) extend1ng ~rom 5 MHz up. T1~ reslllt at the output of adder ~05 ~s a signal for transmis~lon wtth ho~20ntal resolutlon ~olded lnto the S-7 ~ Hz regton. For ea&~
al, ~1 sampl~ palr, one sample romalns rrom quincunx samplln~ on alternate llnes~ separated by 1~ MHz a$ ~hown. Consequently, a 28 meg~ertz process is a¢compllshed at 14 M~lz because of the alter-nate sample dec~n~ation.
Flgur~ 11 shows the deco~er, In whlch lt i~ also pw~ e to lmplement the digltal tilter at 14 MHz. ~n thls caSe, the 6 MHz low-, -' ., . .
-. :

pass filter 111 contalns only 8 co~fflcient~, the oharacteriSttc r~sponse and coe~icient data ~re shown in FlOEure 13, According to Fig. ll, a l.4 MHz inpu~ signal is sampled accord-lng to the figure-of-~ive sampling pattern shown at 14 MH~. The re~eiver decoder ~urther co~prises adder 112 and line store 113 in an upper path. Tlle low~r path further comprises interpolation oireuit 114 lncluding single element delay D and averager (divide-by-two) circuits for restoring missing ~ata to the sample pattern. The upper and lo~er paths are switchably upconverted at 28 ~Hz and ad~ed at ~dder 1~5, Figure 14 shows the ideaiized two-dimensional frequency response achieve~ by the system ln comp~r~son with the known hlgh deflnitlon 1125-line ~IUSE system ~e~feloped by ~apan Broadcasting Corp. (NHK). The MUSE system islvolves a plurality Qf fleld stor3s and thus ls considerably more expensive to implement than the present invention including scan com~ersion apparatus inv~lving one i'leld store. Neverth~less, for dyaamic or statlc images, the hori~on-tal re~olutlon is either equivalent or clearly superior to the MUSE
System according to the present ln~ention and with sCan conversion iS
almost comparable ln vertical definltion.
Flgure }5 shows the actual response achieved using the ~ilters whlch have been descrlbed. According tO Flg. 15a, the po~i~on of tbe folded ener~y (hatched are~) ls also shown in one-dlmenslQn~ In ~ither Fig. 158 or 15b, A relates to the encoder diagonal ~ilter, B to the pre~fllter and C to the skew-symmetrl~ f~lter. The skew~symmetrlc tran~mls61on fllter may ~e implemented in analog form, accorcling to the a~ove~described meth~. According to the above~ ed app~-ratw, a symmetrical non-recurst~re rilter may be used alternately which produces an ideal llnear phase ~haracteri~tic. It a ~lgltal fllter Is us~d to oreate the slcew-symmetric responsel It is unneceS6ary to u~onvert to 2a MHz sanlp~ing. The required digital ftlter will hav~
z~ In all alternate ooe~ficientc ~x~ept ror the central term and so Ieads to a 14 MHz implementat~on. (A 28 MHz implementatlon is not r~qu~d wben alternate sa~nples are automatleally dropp~d wlth alternate ~ero coe~ficlents. The central term is dea~t wlth :, ,.- . , ~: :
. . . .
`' '~:

separately). Figure 15 assumes use of a 16 coefficient non-recurslv~
fllt~r.
~ i~ure 16 shows tl~e actual 2-dlmerLsional response achieved by the system. From Figure 16 in comparison with Fig. 1~ shows improved horizontal resolution out to 945 (gS0~ lines at the cost of dia~onal informatlon. In ac~ord~nce with the above-described embod-lment and method, lun~inance horizontal resolution is improved and with scan conversion vertical resolution improved as well. ~he present lnvention may be adapted to provide improved chrominance horizontal resolutlon recogniæin~ the 3:1 compression of U/V
chrominallce data ln alternate transmitted lines. Scan conversion has already been adapted for o~taining chrominan~e vertical rese~ution improvement. However, in B-MAC, to Intentionally improve chrominan¢e resolution may not be required.
While detailed schematic dla~ram of a hlgh deflnition B-MAC
transmitter and receiver has been described in reference to Flgur~
10 and 11 respectlvely, luminance and chrominance proce~sing will be ~urther de~cribed wlth reIerence to Figures 17-19.
~ eferrtng to ~igure 1~, there is shown a block diagram of ci~
cultry ror luminance processing at an encoder locatlon which works In concert wtth ¢irouitry ~or chrominance processing at the same encoder location. Accordlng to Figures 17 and 1~, the separate pro-c~ psths share the same skew-symmetrlc Eilter portion 1~10.
A 525 lino 1:1 non-sequentlal scan, a 1050 line 2:1 ~nterlac~d or a 1125 line 2:1 5nterlaced vi~co luminance signal is provided to a ver-tlcal filterlintorpolator clrcult 1701 in accordance wlth the Invention resulting In a 480 llne static vertical resolution. As already dcscrl~d, alternate llnes are de¢~nated at declmator 1702 to result in a 2:1 lnterlaced 62S line slgnal and provlded to a diapnal rilter 17Q8. The 28 megahertz sampllng output o~ the diagonal ~ilter is then provlded to a dectmatlon circult 1~0~ for decimat~ng alternate samples on alternate line~ leavlng a quincunx sample structure at a 14 megahertz samplin~ rate. The resultin~ sample~ are stored in memory and read out at a dlr~r~nt rate etfectuating a 3:2 sample compre6slon ln the time domain at time compression circuit 1705. The time compressed . . ~
. . , ~
. ~ ., .

~003~3~i - ~7 -samples sre ~hen mixed with chrominance and any data/audio signals ior transmisslon at multip~exer 1706. Skew-symmetrie filter portion 1710 then is shared by the processed luminance and chrominance signals.
Rsferring to Figure lg, the chrominance processlng is simllar to that for luminance processing but recog~izes that U1V component signals are tran~mitted every other l~ne, and chromin~nce is normally compressed, in accordance with the B-MAC transmission format, at a ratlo of 3:1. Consequently, a chrominance input oi the same num~er of lines in either interlace~ or non-interlaced format as the luminance slgnal ls provided to a vertical ~llter~nterpolator 1801 to achleve a 120 line dynamia or 180 line statlc chromlnance resolutlon in picture height. In the horizontal dimension, the achieved chrominance reso-lution can be up to ~00 lines of picture width. Because every other line is proce~sed, a line decimatlon circuit 1802 for U and V compo-nents is operated at a 4:1 rat~o. The output o~ the llne ~ecimation clrcuit ~s prnvi~ l tn a horl7.nntal_filtPr ~ Pn~tPr~d.~t ~.5 ~2~
hertz. Hlgh frequ~ncy color samples output ~rom the horlzontal filter are then time compressed at 3:1 a~ordlng to B-MAC transmlssion standards at time compresslon circuit 1804, the color s1gnalB are then mlx~d for transmlssion, that ~.c, U and V oomponents are mlxed wlth luminance slgnals at multiplexer ~7~6 and transmitted every other ~ne a ln stan~ard B-MAC with audio or data channels also lnput to multlplexer 1'~06, Again, however, it is indieated In Figure 18 that the Same ~kew~ymmetric Iilter portion 1710 may bè applled to the aombined slgnal prlor to transmi~ion.
Re~erring now to Flgure 19, one ~mbodil~nent for proc~sing o~
a hlgh definitlon B-M~C sl~nal comprlslng ~oth lumlnance and chrominan~e informatlon ls shown. In a~oordan~e wlth the present inventlon, a compllmentary skew-symmetric filter portion 1901 at 10.7 mogahertz receives the HM8-MAC signa~. ~onse~uently, alias~ng o~ the comp~slte signal is ellminate~. The output of ~he complilnen-tary sl~ew~ymmetrlc fllter por~ion 1~01 i~ sampled at a 21 megahertz rste and is provlded to a ~umlnance process~ng path and a chromln~nce processing path. The result is a flgure oï flt~e or ~ .
..
,, .

~Qo~l 36 quincunx sample struclure. Luminance samples are time-expanded 3:2 at time expansion circuit 1902. The output is then provided to a diagonaI ilter/lnterpolator stage 1903 which provides a digital OUtpllt with missing altern~te samples replac~d in alternate lines.
Accordin~ to the chrominance path, chrominance samples are time expanded by 3:1 at ti~ne expansion circuit 1904 and the time-expand~d samples at 7 megahert~ are provided to a vertical filter/
interpolator 1905 for providing useable R-Y and B-Y chrominance outputs. These may he modulated at modulator lgO6 ~nd provided to a share~ digital to analo~ converter 1907 for providing luminance and chrominance outputs Y and C or separately output to high definition televi~ion receiver apparatus.
A hlgh de~inition televisien rec~iver LS shown in block diagram form according to Figure 20. Luminance samples 4 output irom th~
decoder of Fi~ure 19 are input to the high de~inition receiver along with separate R-Y and B-Y inputs. According to the block dia~ram, there i~ provided a scan conversion apparatus 2001 comprising ~n combinatlon a field store 201, a motion detector 20û3 and an interp~
lating algorithm processor 2004. The ~ntent is to provide line doubllng ln th~ vertical dimension as e~pedltiously as possible. S¢an conver-slon apparatus may be employed as di~closed in appllc~tion Serial No. 255,238, filed Octaber 11, lg88 and incorporated herein by r~fer-ence. A8 a result o~ the interpolation process whlch requires a fleld 9tore and associated delay, another luminance sl~nal pa~h oomprise6 delay and compression circuits 2005, 2006 for ~:1 compression which ~s ~lxed ~rith the scan con~rerter ~utput via ¢ompresslon circult 200 ~t multiplexer 2008 ~or display.
Chrominance ln~ormation ~s interpolated In the Yerti¢al dinlen-slon at vertlcal interpolator 2010 compresset 4:1 at oompresslnn cir-cult 2011 and mlx~ with the chrominance input signal for d~play at multiploxer 2012.
Re~errlng to F5gure 21, an o~er~l end user location is shown in pictorlal ~orm. A high cleflnitlon ~-MAC signal 2101 may be recoived and displayed altern~tlvely to 8 video cassette recoder output 2102 or a land~llne ¢able input 2103. A high definition t~levls~on recei~rer is ..
-: . :

,,, , , , ,~
,, ,-:
-,.~ , .. .
., , . ,.
:: : ~ , .. .. .
.

- 19 2003~36 shown which, responsive to a user-sele~ted input entere~ via selector 2104, app~ or example. the already described adaptive scan con-ver~ion methods for line doubling at s~n ~onverter 210S an~ so ~
plays a high definition 16:9 aspect r~tio image on the receiver. Also, an off-the-air broadcast television signal 21û8 may ~e received by antenna, tuned at tuner 2109, deco~ed ir ne¢essary at NTSC decoder 2110 and provided via the same scan conversion apparatus 2105 for display. ~ terrestrial hlgh definition decoder 2111 is required ~f the broadcast signal 2108 is hi~h defln~t~on. Consequently, a user equipped with a high deflnition television receiver is able to improve resolutlon of a displayed image regardless of whether the recelved signal is high definition, standard NTSC low resolutlon, or eT~coded In hlgh or ~ow resolution multiplexed analog components format.

.. . ,~ . . .

: '. ' ............ , `: ~ .. ;,. , ~ .
, .

Claims (10)

1. Apparatus for encoding a high definition multiplexed analog component television signal for transmission characterized by luminance processing means for processing a television luminance component signal by decimating predetermined samples for transmission, chrominance processing means for processing a televi-sion chrominance component signal by decimating predetermined samples for transmission, multiplexing means for combining the processed lumi-nance and chrominance component signals for transmission, and a skew-symmetric filter portion, responsive to the mul-tiplexing means and coupled to a complimentary filter portion, for filtering the multiplexed signal for transmission.

2. Apparatus for encoding a high definition television sig-nal for transmission according to claim 1, the luminance processing means comprising a vertical filter interpolation circuit for filtering an incoming video signal, a line decimation circuit, responsive to the vertical lit-ter interpolation circuit, for decimating alternate lines, a diagonal filter, responsive to the line decimation cir-cuit, for providing a quincunx video sample structure, and a time compression circuit, responsive to the diagonal filter, for compressing video signal samples for transmission.

3. Apparatus for encoding a high definition television sig-nal for transmission according to claim 1, the chrominance processing means comprising a vertical filter interpolation circuit for filtering an incoming video signal, a line decimation circuit, responsive to the vertical fil-ter interpolation circuit, for decimating each of U and V components every other line, a horizontal filter, responsive to the line decimation circuit, for filtering at a color center frequency, and a time compression circuit, responsive to the horizontal filter for compressing high frequency color samples at a ratio of 3:1 for transmission.

4. Apparatus for encoding a high definition multiplexed analog components television signal for transmission characterized by luminance processing means for processing a television luminance component signal by decimating predetermined samples for transmission, chrominance processing means for processing a televi-sion chrominance component signal by decimating predetermined samples for transmission, and a skew-symmetric filter portion, responsive to the lumi-nance processing means and the chrominance processing means and coupled to a complimentary skew-symmetric filter portion, for filter-ing the high definition muitiplexed analog component signal for transmission.

5. Apparatus for decoding a high definition multiplexed analog component television signal upon reception, predetermined television signal samples having been decimated prior to transmission, the apparatus characterized by:
a skew-symmetric filter portion, coupled to a compli-mentary skew-symmetric filter portion, for filtering the received television signal, sampling means, responsive to the skew-symmetric ter portion, for sampling the filtered signal, luminance processing means for processing the sampled signal by interpolating decimated luminance samples and chrominance processing means for processing the sam-pled signal by interpolating missing chrominance samples.

6. Apparatus according to claim 5, the luminance process-ing means comprising:
a time expansion circuit and a diagonal filter and interpolator for restoring deci-mated luminance samples.

7. Apparatus according to claim 5, the chrominance pro-cessing means comprising:
a time expansion circuit and a vertical filter and interpolator for resorting decimated chrominance samples.

8. Apparatus Coupled to a high definition multiplexed ana-log component television signal decoder for processing luminance and chrominance component signals for display, the apparatus character-ized by:
a luminance processing path comprising a motion detec-tion and interpolation circuit for interpolating luminance samples determined from motion between first and second fields and a chrominance processing path comprising a vertical interpolation circuit for interpolating chrominance samples.

9. Processing apparatus according to claim 8, the motion detection and interpolation circuit comprising:
a field store for sorting a first video field and an interpolator responsive to the motion detector for interpolating samples from the stored video field and the second video field.

10. Processing apparatus according to claim 8 further com-prising first and second compression circuits for compressing a delayed luminance input signal and the output of the motion detection and interpolation circuit respectively and a multiplexer for multiplexing respective outputs of the first and second compression circuits.