`
`1
`
`Digitalcoding’of'Color Video Signals—A Revie
`
`J
`
`w.
`
`JOHNO. LlMB, savior; igEMBER,1EEE,CHARl.ES B. RUBlNSTElN,-AND JOHN E. THOMPSON
`,
`
`PMCAPL02442670
`
`ponents of the signal. To enable the reader to fully appre-
`ciate the various factors that bear on the encoding of color
`signals, we have gone into background material in some de-
`-tatl. Section II‘ provides basrc infomiation in the area of
`colorimetry,_laying a foundation for the colorimetricpro-
`perties of color
`television. Section III describes the format
`of the three major color television systems in use in the world
`today, and some of the considerations that led to these stand-
`ards. Section IV covers aspects of the work in color vision which
`bear on thc'problem of efficiently encoding color video signals._
`Final preparation for the coding sections is given in Section
`V where the statistical nature of color signals is described.
`Readers with familiarity in these background areas may wish
`to jump straight to Section VI although there is some work on
`color vision in Section IV that is perhaps not widely known
`to workers in the coding field
`p
`A
`We. assume that the reader has some familiarity with basic
`waveform encoding techniques (see {2}). Coding of the lumi-
`nance component per se will not be discussed in any detail.
`Of course, in codingthe composite signal it is not feasible to
`divorce the coding of the luminance and chrominance compo-
`nents. Reviews of work on luminance encoding are available
`fgr readers who would like more background. Reference [3]:
`covers the proceedings of a special conference on efficient
`picture coding; covering all- aspects. References [4], [5] and
`n‘
`[6] are special issues on signal processing which contain large
`- sections describing video coding techniques.
`From the first experiments in color encoding two somewhat
`sepa‘rate approaches have been exploited. The first is to operate
`directly on the composite television signal, whereas the second
`divides the signal into three components, codes each compo-
`nent separately and then. after transmission, combines them
`again to form a composite Signal. These two different ap-
`proaches are examined in some detail in Sections VII and VI,
`respectively. Finally, in Section VIII we compare the implica—
`tions of the composite and component encoding methods and
`conclude that, at least in the short~ to medium-term future,
`both coding strategies will find important application.
`We express a word of caution concerning the assessment
`of the performance of different encoding techniques. At the
`very minimum, such assessment
`rcqulres the measurement
`of the bit—rate for a given picture‘quality. It “is primarily the
`assessment of picture quality that is so variable. Picture qual—
`ity depends on many factors. for example lighting conditions
`and monitor adjustments. the range of picture material
`that
`is presented, whether a single stored frame (or photograle
`is being viewed, the amount and type ol‘rnovcment contained
`in the Iiccnc and the experience and expectations of the
`viewers.
`
`(Invited Paper)
`
`-
`
`L r,
`
`.
`
`Abstract~fllis paper reviews the field of the efficieht coding of
`color television signals. Because this paper is perhaps the first review on
`this topic, some background is given in the areas of colorimetryfiisual
`perception of color and color television systems. .We assume that the
`reader has some familiarity with luminance encoding techniques. a
`_
`Coding techniques themselves are divided into two broad groups:
`component coding method: in which each Eomponent (usually three)
`is coded separately, and composite coding methods in which the
`composite television signal with its ‘Tcolar” modulated suburrier is
`processed as a single entity. Both approaches are covered in detail.
`Theflekl is still growing pushed primarily by the desire in the
`television area to find digital coding standards accepted by both broad-
`casters and carriers and suitable for use with NTSC, PAL and SECAM
`television systems. We discuss this aspect by comparing composite
`and component coding methods.
`
`l/\
`
`I. INTRODUCTION
`
`HE digital coding of color video signals has reach/ed con-
`siderably less attention than the coding of monochrome
`video signals. However, given the current widespread prolif-
`eration of color television systems and the general preference .
`for; color pictures versus monochrome, it is obVious that the
`efficient coding of color picture signals is of prime impor-
`tance. BrOadcast color television systems make highly eff-
`icient use of the analog bandwidth to accommodate the in- ,
`creased informhtion content of color signals.
`If the same rel-
`ative efficiency is to be achieved in the encoding» of color
`signals as has been achieved for monochrome, it is "going to
`require a great amount of ingenuity.
`A
`The first attempt in digitally encoding a color signal most
`probably started in I960
`the with of R.’L. Carbrey [l]
`on applying PCM to a broadcast celor television signal. A
`few papers were published on the subject during the fol-I
`lowing 11 years until 1971 when there was a marked increase
`which has persisted to the present and gives every indica-
`tion of continuing. lyherefore seems especially appropriate
`to review this field now while many new techniques are still
`being explored.
`~
`’
`In laying the foundations of the present-day color tele-
`‘vision standards in the late t940’s, much study went
`into
`various background topics swat as colorimetry and visual
`perception so as to match the rcéulting signal
`to the color
`fidelity requirements of the it man observer. Further, addi-
`tional studies have been made-in the area of threshold color-
`difference perception and on the interaction between the
`“brightness” (luminance) and “color” (chrominancc) com—
`
`"‘
`
`Vamrrcript received December 7. [976; revised May In, 1977A
`I, ()r Limb and I", B, Rubinstein arc with Bell Laboratories,Holmrlr-L
`NJ 07711.
`I E Thompson it with tho Post Office Research Center, Ipswich.
`lingland.
`‘
`
`PMC3683177
`
`PMC Exhibit 2039
`Apple v. PMC
`IPR2016-00755
`Page 1
`
`
`
`1350 ‘
`
`‘ f
`..
`a
`/
`IEEE TRANSACTIONS ON COMMUNICATIONS. VOL. COM»25,N1D. 11, NOVEMBER 1977
`
`I
`
`1
`
`units of color (P2), and 7 units of colqir (P3). The colors (P1),
`(P2) and (P3) are conventionally called/primaries.
`2. The luminaan of alcolor mixture is equal to the sum
`of the luminances of the componentsin the mixture
`
`v,
`L=L1+L2+L3.
`
`.
`
`'
`
`.
`
`(2)
`
`.
`
`’
`
`In the work on component coding described'in Section VI,
`rates in the range of 1-2 {bits/pel (picture element) have been
`found to give “good’.’ quality: pictures. But, in many cases,
`quality is based on single—frame simulations .or on a small
`range of' picture-materiah‘not representative of that-handled
`by broadcast
`television._ The situation-will; only be eaSed
`when all workers adhere to a common testing'procedure
`(such as that recommended by the CClR [7]) and use com—
`.morrpicture materialf”
`'
`5
`W
`This difficulty in comparing picture qdalityjin trim, com-
`plicates the task of the authors in comparing different coding
`schemes. Even side-by-side comparison of picture quality
`would not provide definitive ratings since other factors such ,
`as error perfonnanpe, cost, complexity and compatibility will
`all affect the final decision on the type of coder most suitable
`for a given application. Finally, the field of color coding is the
`subject of a great deal of current activity. and much of the
`research required to reach firm conclusions on the relative
`1
`i
`worth ofvanous Schemes has yet to be completed
`
`These rules have been extensively tested experimentally, and
`some discrepancies have been found (see Guth [12] for a re-
`view).
`_
`’
`7
`Wintringham (13} discusses the foundations of calorimetry
`in terms of a somewhat idealized set of experiments designed
`to determine what mixture of three primary colors would ap-
`pear exactly lilte a spectral color. In these experiments an ad.-
`justable mixture of three well-chosen monochromatic lights
`act as the primaries. Subjects adjust the strengths of the three a
`primaries to match each color in the series of spectral colors
`after first matching a white that has equal energy in all parts of »
`the visible spectrum. The match to white represents one unit
`of each primary.
`‘
`‘
`'
`Note that we have not mentioned the absolute intensity or
`' radiance of the spectral colors that are matched. This was. in-
`tentional because the relative proportions of each of the three _
`primaries is, independent of
`we radiance'ove? a wide range.
`instead of referring directlyt
`the number of units required to
`make the match, colorimetrists use normalized quantities
`called chromaticity coordinates expressed by the relations:
`
`PMCAPL02442671
`
`lI.’REPRESENTATlON OP COLOR
`4
`
`A. Trichrarndcy
`
`‘_
`
`There is a long history associated with the science of color.
`and much of this history has been recently reported by
`MacAdam [8}. Those interested in the early work will also
`find an interesting collection of papers in a book by MacAdarn
`[9] that spans a period of approximately 2000 years.’The credit
`'for the great advance in the science of color concerning the
`trichromacy of vision generally goes to Thomas Young for his
`advancement of the concept
`that the retina is composed of
`three sets of sensitive mechanisms~one for each. of three
`principal colors [10] . Today it‘is generally accepted that there
`are three types of cones in the retina which mediate color vi—
`sion at light levels greater than approximately 10 cd/mz.
`The natural counterpart
`to‘ the trichrOmacy of vision is
`’the trichromacy of color mixture. Maxwell studied color mix-
`tures and demonstrated the first
`threecolor projection in,
`1855, a description of which appears in the reprint of his 1857.
`paper {9] .Maxwell’s work formed the basis for'colorimetry—
`the‘technique oflhe measurement of color.
`' 5'
`
`B. Calorimetry
`
`Calorimetry is based on the premig that a relationship can
`be found between the physical stimuli and the visual‘sensation
`that arises from them. At the foundation of3-color colorime—
`try lie a. series of rules generally attributed'to Grassman [11].
`Two of the more important‘rules can be expressed as follows:
`1. Any color stimulus can be matched (in appearance) by
`the additive mixture of three matching stimuli provided that
`no one of th; three matching stimuli can be matched by the
`remaining two. This can be expressed as
`
`{C}
`
`all”: ) + {ff/’2) 4' 7W3)
`
`‘
`
`(l)
`
`whorls ((7): is a color-Ilia! is matched by arunits of color (1’, ),[l
`
`r=R/(R+G+B)
`
`g=G/(R+G+B)
`
`b=B/(R+G+B)
`
`"
`
`‘
`
`'
`
`‘
`
`r
`
`(3)
`
`‘
`
`where we have changed the notation such that the units 01,3
`and 7, which are called tristimulps values,'have been replaced
`by the symbols R. G and' B, re: ectively, to agree with the
`common usage of Red, Green an?Blue primaries in these ex-
`periments. The results obtained by Guild [14]
`in his fundai
`mental measurements of color mixture for 7 subjects are
`shown in Fig. l.
`The normalizing process we have just carried out to obtain
`the chromaticity coordinates has eliminated the radiance infor»
`,mation, and we are left with only two pieces of information
`The third dimension of color is obtained‘by a separate measure-
`ment of luminance. In a three-primary match to the reference
`white, the ratios of each of the three component luminances
`contributed by each primary to the total luminance are called
`the luminosity coefficients.
`
`C Color Wansfnmarions
`
`A colorimetrist’ designates color by a graphical representa»
`tion in a color space. This requires the data for a “Standard
`Observer“ that would be representative of the mixture data of
`color normals. As noted in Fig. 1. Guild [14] had obtained
`
`' luminance is a quantity measured in a photometer. it is the phony
`metric brightness of a uniform. sltiall field.
`
`PMC3683178
`
`PMC Exhibit 2039
`Apple v. PMC
`IPR2016-00755
`Page 2
`
`
`
`PMCAPL02442672
`
`The Commission international de L’Eclairage (CIE) in l93l
`defined the color-matching data of the Standard Observer to
`be the mean of the data of Wright and Guild, and the equal-
`energy white was adopted as the reference white.2 These spec-
`ifications enable us to define the color mixture data of the
`Standard Observer (see, for example (13, 17]). The data indi-
`cate how much of each primary is needed to match spectral
`stimuli of equal rradiance for the Standard Observer. The chro-
`maticity coordinates of the spectral colors for the Standard
`Observer based on the National Physical Laboratory primaries
`are shown in Fig.3. Corresponding tristimulus values for spec-
`tral stimuli ofequal radiance are shown in Fig. 4. Chromaticity
`coordinates express fractions of a whole mixture. On the other
`hand,
`tristimulus values express how much of a primary is
`needed in a match to a given spectral coldr.
`One form of graphical representation of this information on
`a chromaticity diagram using the r and g chromaticity co-
`ordinates is shown in Fig. 5. The spectral colors plot on' the
`elongated horseshoe shaped curve called the spectral locus.
`The straight line connecting the two extremes of the spectral
`locus is called the line of purples. Note that the spectral locus
`
`
`
`
`
`Fig. Z. Luminosity curve for the Standard’Observer.a At 555 rim, .(
`one W is equivalent to 680 lumens (from Wintringham [13]).
`
`IA
`
`
`
`COOHOIHAYESé
`
`
` CWATICIYY
`
`"7
`
`WAVELENGTH (/4)a 3
`
`
`The chromaticity coordinates versus wavelength of the spectral
`Fig. l.
`colors for-seven observers using Guild’s triehromatic calorimeter 5
`prim ‘es, and the National Physical Laboratory (NFL) reference
`white from Guild [14]).
`g.
`
`1
`‘
`data on 7 observers with a particular, set of spectral primaries
`and a particular reference white. Wright [15] obtained data
`for 10 observers using different spectral primaries and did not ‘
`normalize to the same white. In order to makeuse of the data
`of both Guild and Wright a transformation must be found be-
`tween the tristimulus values of a color for two arbitrary sets of '
`primaries. This has been solved by a number of workers for
`certain special cases. Wintringham [13] has treated the pro-
`blem in a very general form in which the two reference whites
`are not the same. The result can be expressed in terms of a
`3 X 3 matrix transformation.
`
`D. Chro malicin Diagrams
`
`“The (Hi [In] had already adopted a standard relative luminous
`effluirmzy function V,\ shown in Fig. 2 for photopia (“normal daylight
`vision“) (auditions.
`the function represents the results derived irom
`several different photometric methnd‘t ol' equating the brightness of
`spectral energy murr'm
`
`Mmfll
`
`700
`
`lino
`
`Fig. 3, Chromaticity coordinates of spectrum colors for the Standard
`Observer (from Wintringham [13] ).3 Primaries: 700.0 rim, 546.1nm
`and 435.8 nm (NPL primaries). Reference white: equal—energy
`white.
`.
`.
`
`extends outside the triangle formed by the three primaries
`which are, ofcourse, located Siam), (0,1) and (1,0).
`An important property of a chromaticity diagram concerns
`the calculation of the chromaticity of a mixture by a method
`analogous to a center of gravity system with the luminances of
`the components acting as weights. If two colors are additiver
`mixed,
`then the chromaticity of the mixture lies on the
`straight
`line between the two chromaticities of the compo-
`nents. For a three-color mixture the Chrornaticity of the result
`lies within the triangle formed by the three component chro-
`maticities. The extension of the spectral locus outside of the
`color triangleformed by the three primaries in Fig. 3 is a con»
`sequence of the necessity‘of adding one of the primaries to
`some of the spectral colors in order to carry out the match
`(equivalent to moving one of the terms from the right side of
`(I) to the left).
`‘
`(.
`Suppose we wish to calculate tlmchromaticity ofan object
`
`"The symbol “my“ on the axis is outdated and has been replaced‘by
`"om" in our tovt.
`
`PMC3683179
`
`PMC Exhibit 2039
`Apple v. PMC
`IPR2016-00755
`Page 3
`
`
`
`\r
`I
`'
`IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. COM-7:5, N0. 1 1, NOVEMBER 1977
`
`
`
`
`500
`
`540
`
`565
`Mm“)
`
`520
`
`560
`
`700
`
`740
`
`Fig. 4. Tristiinulus values?,§ andb of spectral stimuli of equal
`radiance for the Standard Observer (from Wintringham‘ii31).3
`r
`
`,_ waacmxS
`
`l
`
`..4_r,.,._
`
`
`
`
`
`,
`
`l RflleclM‘
`
`, RlMxEcihlné
`
`
`
`PMCAPL02442673
`
`over as large a range as possible. Calculation of the luminance
`of a color' would be made much easier if the luminosity co
`efficients of two of the primaries were equal to zero. The lumi-
`nance of a color would then be equal to the number of units >
`of the other primary used in the match.
`it was these considerations, among others, that led the CIE
`to propose a new set of primaries. The spectral locus is totally
`contained within the triangle formed by these new primaries
`(denoted by (X), (Y) and (Z) (as can beyseen in Fig. 5) imply-
`ing that all spectral colors can be matched with a positive
`quantity of each primary. The use of such “nonphysical” pri-
`maries should be no cause for concern: For méasurement pur.
`poses any real set Can be used and the results can be trans-
`formed by a 3 X 3 matrix to the nonphysical set."
`Luminosity information is obtained from the tristimulus va-
`lue ofgthe (Y) primary—the luminosity coefficients of the
`other two primaries are equal to zero. The resulting tristimulus
`values X, )7 and z— are shown in Fig. 7. Note” the all-B‘ssitive
`nature of the functions, and that)? is identical to the VA curve
`of Fig. 2._ The x}; chromaticity diagram for the 1931 CIE
`Standard ObServer is shown in Fig. 8. The equalenergy white
`(E) has the coordinates (1/3, 1/3) because it is the reference
`white for this system. All the color mixture properties that we
`have previously described for the we diagram are valid for this
`diagram, however, the equations for color mixture are espe-
`’
`cially simple for this case. If we are given the chromaticities
`’ x1. y1, x2; M and their luminances L1 and [,2 the chroma-
`a “City Of the mum” is Simply
`
`
`
`n(HIE;(Mxioo
`
`S
`
`u0
`
`3&8
`
`49
`no
`
`.
`
`540
`
`see
`Mmul
`
`527;
`
`see
`
`700
`
`no
`
`Products of the tristimulus curves'f‘ E‘and E with the reflect-
`. Pig‘s.
`ance of a colof sample, R()\),
`irradiated by Illuminant C, ECO)
`(from Wintringham {13] ).3
`
`Fig. 5. The rg chromaticity diagram for the Standard Observer (from
`Wintringham (13)). The wavelengths (in run) of the spectral colors
`appear on the horseshoe shaped locus, Point E represents equal-
`energy white, C represents illuminant C which is a standard bluish- -
`white source, P represents a specific color sample irradiated by
`Illuminant C and (X), (Y) and (Z) are the standard CIE nonphysical
`primaries discussed in Section ll. D.
`
`that is illuminated by a light of a specific spectral distribution.
`The spectral distribution of the reflected light may, of course,
`be th0ught of as being composed of an infinite series of spec-
`tral colors. To determine how much of each primary is needed
`in the mixture, 21 product is formed of each of the tristimulf'sx
`values with the spectral reflectance of the object as shown it ‘
`Fig. 6. The areas under each curve, as obtained by integration,
`are the desired tristimulus values R, G and B for the sample.
`x:
`Chromaticity coordinates can be calculated using (3) and‘ the
`point “P” is plotted in Fig. 5 ‘
`‘
`,
`‘ M
`»
`in 1931 such calculations were commonly/permit “ ton
`desk calculators, and the negative lobes of the furiCIions of
`Fig. 4 introduce negative product terms in which the negative
`sign is error prone with repetitive summing and differencing
`operations. it would be much better if there were no negative
`lobes, and it would be convenient if the quantities were mm
`
`x3 ~
`
`w vim Eel/:45?)
`(L 1/5’1) + (L 2/.1’2)
`
`J’s '
`
`1., +L2
`‘ (Li/yr) i «Lg/m
`
`and the luminance is
`
`PMC3683180
`
`PMC Exhibit 2039
`Apple v. PMC
`IPR2016-00755
`Page 4
`
`
`
`u
`
`l
`
`LXMB er
`2.0
`
`DIGITAL CODING OF COLOR YIDEO SIGNALS
`
` 'o
`
` A:
`
` TllaYlMULUSVALUES
`
`Fig. 7’. Tristimulus values 3:"; j! and I of spectral stimuli of equal
`radiance for the Standard Observer (from Wintringham [l3] ).3
`
`
`
`
`
`
`
`Fig. 8. 1931 ClE xy chromaticity diagram 00111an the spectrallocus,
`line of purples. and the chromaticity locations‘A,‘ B, C and E for
`CE standard Illuminants A, B. C and equalcnergy white (from
`Wyszecki and Stiles, Fig. 3.10 [17]).
`’
`’
`.
`
`L, =1;1 +L2'.‘
`
`(5)
`
`The 1931 Standard Observer was based on color matches
`using 3 2° fieldfln 1964, the one defined another Standard
`Observer, but based on a 10° field. The data base was obtained
`from Stiles and Burch [18] and Speranslcaya [19]. The
`location of certain wavelengths are somewhat shifted along the
`spectral locus in the 1964x10ym diagram as compared to the
`1931 xy diagram as can be seen by examining Fig. 2.17 of
`[20].
`”
`It is important to realize that the chromaticity diagrqu we
`have discussed pertain to color in the objective sense; They
`are based on color matches and not on color appearancc'. In—
`formation concerning the subjective color sensations of hue
`and saturation are not obtainable
`from a chromaticin
`diagram.
`-
`y
`
`1353
`
`E. Uniform Chromaticity Diagrams
`
`The ‘xy chromaticity diagram is still in wide use today, but
`it has a major shortcoming for some practical applications?
`Color samples which have chromaticities which are equally dis-
`tant from each 'other on'the diagram are not equally different
`in appearance. MacAdam has determinedthe loci of chromati-
`' cities that are equally noticeably different from each of 25 re-
`presentative colors for a constant level of luminarl'ce .[2l, 2?] .
`Such loci are ellipses and they‘are shown on the xy' diagram in
`Fig. 9. Notice how the differences vary over the diagram.
`There have been many attempts to‘make a diagram in
`which equal distances correspond to equal differences in per-
`ception under the restriction that it be'obtainedjby a'linear
`transformation of the xy diagram [l7] .4 However, the goal is
`impossible to attain strictly although improvements can be
`made. The 1960 ClE-UCS diagram [23] is onefsuch diagram,
`and it is shown in Fig. 10 with a plot of MacAdam’s ellipses.
`These ellipses tend to be more circular than those shown in
`Fig. 9. The transformation from x, yfto u, v is‘as follows:
`
`it = 4x/(—2x + 123/ + 3)
`
`i u: 6y/(~2x + lZy + 3).
`
`'
`
`’(6)
`
`ill. THE COLOR TELEVISION SIGNAL
`
`' A. Relation to Calorimetry
`
`I.
`
`We can draw an analogy between a color television system
`and a colorirrieter. The three phosphors of the receiver corre-
`spond to the three primaries of the colorimete'rcand the cam-
`era taking filters correspond to the color mixture curves for
`them primaries. Each "color in the scene before the camera
`must be matched by suitably controlling the light qptput of
`the receiver phosphors. This goal will be achieved if the light
`contributions from the receiver pho’sphorslare adjusted tobc
`equal to the tristirnulus values appropriate to this sytem of pri-
`a.
`maries for each of the colors in the original scene. it we as'-‘_
`some for the moment that‘the television system is linear, then
`the three signals from the camera'should be proportional to
`theseiristimulus values. As can be surmised from Section ll,
`this is achieved by making the spectral Sensitivity of the color
`fitter in each of the three channels in the camera proportional
`to the corresponding color mixture curve of the Standard Ob-
`server for the receiver primary it controls. These relations were
`applied in the design of color television systems.
`
`B. Format of the-National Television System Committee
`{NTSC} Color Television Sigma!
`
`two basic re-
`The NTSC color television standards meet
`quirements: (l) compatibility with existing monochrome ref
`ceiVers and (2) bandwidth containment of the color signal
`within the existing bandwidth for monochrome television.
`\
`
`‘The latter condition is needed in order to retain the facility to
`carry out color mixture on the diagram by the method analogous to
`center of gravity systems
`i
`r‘
`
`PMC3683181
`
`
`
`
`Vii
`,
`51224-le
`sec
`‘0
`550
`620
`660
`Minn),
`-
`
`700
`
`7‘0
`
`7&6
`
`PMCAPL02442674
`
`PMC Exhibit 2039
`Apple v. PMC
`IPR2016-00755
`Page 5
`
`
`
`y
`IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. CUM-2 5, ND. ll, NOVEMBER 1977
`
`the three color signals must all be specified. Theprir
`izing
`maries were chosen such thattthere was ‘5
`large gamut of
`chromaticities contained within the triangle formed by the
`three receiver phosphors used as primaries. The three color
`signals, ER, EG and EB were adjusted to be equation. the ref-
`erence white (llluminant C). For the conditions specified by
`the NTSC, the luminance signal is
`
`BY = 0.305;2 + 0.59153 + 0.1153."
`
`(7)'
`
`The coefficients in this equation sum to unity‘so that when’
`illuminant C is reproduced
`..
`.
`,
`
`EY=ER=EG=EB.
`
`(8)
`
`1
`
`0.6
`
`Fig. 9.
`
`1931 C113 diagram showing MacAdam’sxellipses enlarged
`10 times (from Wyszecki and Stiles, Fig. 6.36 [17]).
`
`
`
`PMCAPL02442675
`
`Several advantageslaccrue if, instead'of transmitting thersig-
`‘nal by and we of the three color signals ER, EG any EB, the
`luminance si
`91 is transmitted along with two “color differ-
`.
`R
`‘uence" or “chrominance” si
`als such as E -
`EE.— 51,.
`The receiver can recover the E19, EG [and EB signals by a par-
`ticularly simple linear matrix operation. Variationsvin the two
`colorydifference signals do not affect
`the reproduced lumi-
`nance. The same is true if the other two signals were chosen
`from ER, EG andEB. This impewiousness of the luminance to
`perturbations in the“ other two signals is called the “constant
`luminance principle." A specific” advantage of using color
`difference signals and normalizing to the reference white is
`that for achromatic signals hoth color difference signals are '
`equal to zero. Variations in the relative strerflgths of the three
`signals do not affect the color‘balance of the reproduction of
`any level of grey. As will be pointed out in more detail in Sec-3
`tion V, the majority ofcolors in most scenes have chromatic-
`ities that are close to the reference white This reduces the
`amount of information that the color difference signals must
`carry.
`.
`'
`Of special interest in terms of color television transmission
`is theaability of the visual system to detect spatial detail in the ,
`color difference signals. Pearsonl[24] gives a short acconnt of
`most of the applicable early work. Briefly stated, the NTSC
`color difference signals denoted as I and Q were chosen to be
`
`
`
`
`
`Fig. 10.
`
`1960 ClE-UCS diagram showing MacAdam’s ellipses (from
`Wyszecki and Stiles, Fig, 6.48 [17]).n
`
`This is achieved by transmitting a luminance signal represent-
`ing the monochrome information and another signal compris-
`ing two “chrominance” components which supply the geldi-
`tionalainformation needed to represent the chromaticities of
`the original scene. Frequency multiplex techniques are used to
`simultaneously transmit
`these signals in the bandwidth previ-
`ously allocated for monochrome television. In order to make
`the most efficient use of the bandwidth oi; the channel the ’
`properties of the human visual system are utilized. The two
`chrominance components were chosen such that they could
`be accorded much less bandwidth than the luminance signal.
`The exact form of this signal will he described shortly.
`in order to form the luminance signal.
`the chromaticities
`of the three receiver primaries and the condition for normal-
`
`£1 = ((53? Ey) cos 33°,I1 .14)
`
`’ — ((EB —. Ey) sin 33°/2.03)
`
`50 = «ER ~ Ey) sin 33°/1_14)
`
`+ ((5H .7 By) 905 33°/2.03).
`
`(9)
`
`This choice is near optimum for the constraints involved. The
`E, signal carries orange-red to cyan information and the 50
`signal carries green to magenta information. The chrmninance
`bandwidths assigned prior to formiri‘i; the composite signal are
`as specified in Table l. The luminance bandwidth is a nominal
`42 MHz.
`v4
`,
`The Llifirity in the relative apportionment of the band»
`width for the three transmitted Signals results in a
`“mixed
`highs" system. For large area color the clu‘oruaticily is reprw
`
`PMC3683182
`
`PMC Exhibit 2039
`Apple v. PMC
`IPR2016-00755
`Page 6
`
`
`
`1355
`
`in use in France, Iran, the Middle East and Eastern Europe.
`(See Carnt and Townsend [26] for a’ description of these sys~
`tems.)
`“
`In PAL and SECAM the chiominance components are des-
`ignated as
`,
`‘
`
`4U: (37 20/203
`
`V= {R L Y)/i.ld
`
`9
`
`(12)
`
`where the factors 2.03 and l.‘l4“are the same as those appear—
`ing in the NTSC signal, but the cosine and sine terms are ah;
`Sent.
`‘
`>
`5
`
`in the lngSC system,. deflation by mo‘ie than 600 kHz‘
`above the subcarrier frequency exceeds the, nominal band—
`width (4.2 MHz). Crosstalk between the“! and Q channels is
`avoided by restricting the Q bandwidth to 600 kHz which
`allows enough room for a double sideband (therefore no cross-
`talk). and confining sideband distortion to the 1 signal which
`begins to roll off around 1.3 MHZ.
`‘
`
`“is
`
`‘5‘»...
`
`l
`
`LIME at at: DIGITAL CODING OF COLOR VIlDEO SIGNALS
`
`(
`‘
`TABLEI
`BANDWIDTH OF THE CHROMINANCE SIGNALS
`WWW—“é
`mac g-Channei
`. "
`.
`
`lloo kliz less than 2 dB Clown
`500 kHz ress than 6 are down
`600 kHz at:
`least 6 dB down
`
`W
`
`1.3 MHZ less than 3 dB down
`
`3.6 m2 at,
`
`least 2018’ down
`
`% PAL gs-[stems a; c, a and :2:
`‘
`and V- Channel
`‘at 1.3 MHz less than 3 dB dmjn
`at “from at least 20 (35 down
`‘i?
`.
`
`duced correctly. As the area becomes smaller and smaller, first
`the. E9 signal and then the 1:} signal is attenuated, and finally,
`only the high frequencies of the luminance signal—remain.
`In order to fit the chrominance signals into the luminance
`bandwidth'they are amplitude mohulated onto two color sub-
`carriers of the same frequency (3.58 MHz) but shifted in phase
`by 90°. The exact frequenCy for the ‘color subcarrier is chosen
`to be an odd multiple of half the horizontal line frequency so
`that the clumps of energy in the subcarrier fall between: the
`' clumps in the luminance signal spectrum, and so are of mini-
`mum visibility ivhen viewed "on a monochrome receiver. The
`form of the composite solo} signal is:
`
`E=Ey' + [561' sin (wt + 33°) + E,’ cos (cut and]:
`
`(10)
`’
`The. two chrominancebcomporxents’in (10) can beithought'of
`as modulating the’amplitud‘e and phase of a single subcarrier.
`The primes on the symbols refer to the gamma corrected volt-
`ages. Up to this point we have been assuming a linear system
`which is not
`the case in practifl. For the recommended
`gamma correction of2.2,E§:G_B = ERtG‘Bl/‘z-z and
`
`EY‘ =0.3OER’+0.59EG'+0.HE,,'.
`
`(11)
`
`The chrominance components can be plotted on the 1960
`CIE UQ‘I diagram in terms of the normalized amplitude and
`subcarrier phase as shown in Fig. 11 (Townsend [25]). These
`loci for a gamma of 2.2 are not circles andflstraight lines as
`would be ideal.
`'
`
`C. PAL and SEC/1M
`
`The Phase Alternation Line (PAL) color television system is
`in use in most of Western Europe, and the SEC/KM“ system is
`
`It refers to the use of a
`’The name SEFAM in not an ucmnym.
`millenflsl chmminnncc signal and a merrer device (sequentian al
`mémoire)
`r
`
`PMCAPL02442676
`
`An alternative technique of avoiding U/ V crosstalk.‘i‘s
`adopted in the PAL system whereby the phase of thé Vsignal
`is changed by l80° between successive lines in the same field.
`‘Complementary’ crosstalk errors therefore occhr on adjacent
`lines and can be reduced significantly by averaging in a delay ,i
`line decoder.‘ Similar bandwidth may therefore be allowed for
`the U and V signals (Table I) which are transmitfcd, as in
`NTSC, in phase quadrature Lsuhoarrier is located at 4.43 MHzl.’ '
`However, since a subcarricr which is an odd multiple of half
`the “line frequency (NTSC) would now cause the line alternat.
`ing V component to crosstalk into vertically correlated lumi-
`. n'ance energy,- the PAL subcarrier is chosen to bear a quarter
`line offset. Alternation of the V component now results in
`spectral separation of the U and V signals between line har-
`‘ monies, as in Fig. 1221 as compared with the NTSC spectrum
`shown in Fig. 12b. An additional subcarrier offset of 25 H2
`relative to a quarter-line harmonic gives appreciable improve-
`ment
`in compatibility (with monochrome reception) and
`crosscolorg, but requires eight fields to complete the spatial
`subcarricr cycle.
`In addition to the advantage mentioned, in
`respect
`to sidebanding.
`the alternation of the Vcomponent
`gives a degree of protection against phase errors (erg, differen-
`tial phase distortion) and‘multipath reception. These errors
`change the saturation but not the hue As a results of averag— '
`lug out phase errors, hue and tint controls are not required
`in the PAL receiver.
`'
`
`Immunity to gain and phase distortion of the chrominancc
`is provided in the SECAM system where frequency modulation
`of two subcarriers is used to transmit the U and V signals indi~
`vidually on alternate lines,
`thus precluding‘the possibility of
`U/V crosstalk Unfortunately the system is still susceptible to
`different