1、1 ScopeThis practice is intended to define the numericalprocedure for deriving reference video signals fortelevision camera evaluation. A standard reflectiontest pattern is assumed and the signal levels com-puted by this procedure are the signals to be expectedfrom an ideal camera as defined by the
2、televisionsystem color specifications.2 Normative referencesThe following standards contain provisions which,through reference in this text, constitute provisions ofthis practice. At the time of publication, the editionsindicated were valid. All standards are subject torevision, and parties to agree
3、ments based on thispractice are encouraged to investigate the possibilityof applying the most recent edition of the standardsindicated below.ASTM E308-96, Practice for Computing the Colorsof Objects by Using the CIE SystemCIE Publication 15.2 (1986), ColorimetrySMPTE RP 177-1993 (R1997), Derivation
4、of BasicTelevision Color Equations3 DefinitionsThe color reproduction objective of television systemsis to reproduce all colors as though they were illumi-nated by CIE illuminant D65. Thus, the ideal or refer-ence camera used in this practice always uses a D65illuminant. This practice utilizes CIE t
5、ristimulus values(XYZ) as given in CIE 15.2 to specify color samples,and the normalized primary matrix (NPM) as given inSMPTE RP 177 to derive the ideal camera linear videosignal levels (RGB).4 Input data4.1 The required data are first the CIE tristimu-lus values XYZ, computed using CIE illuminantD6
6、5, for each of the color patches on the testpattern to be used. One of the patches shouldbe a reference white that will be used for thewhite balance of the reference camera. If thetristimulus values for the test pattern are notavailable, they should be computed from thespectral reflectances of the p
7、atches using theweighting tables and procedures from CIE15.2 or ASTM E308. The reflectance datashould ideally be measured using 45/0measuring geometry as specified in CIE 15.2and be made against an absolute white reflec-tance reference.4.2 The second input data required is the nor-malized primary ma
8、trix, NPM, of the televisionstandard for which the camera is being evalu-ated. This NPM matrix should be derived fromthe system reference primary chromaticities us-ing the procedure given in SMPTE RP 177. TheNPM should be accurate to at least four signifi-cant digits and ideally all 10 digits carrie
9、d in itsderivation should be used. The reference sys-tem NPM uses a D65white reference by defini-tion.4.3 The final input data are the equations forthe opto-electronic transfer function (gammacorrection) for the television standard.5 General procedure5.1 The NPM defines the relationship betweenthe n
10、ormalized linear RGB video levels and CIEtristimulus values XYZ:Revision of RP 176-1993RP 176-1997 SMPTE RECOMMENDED PRACTICEApproved December 5, 1997Copyright 1997 by theSOCIETY OF MOTION PICTURE AND TELEVISION ENGINEERS595 W. Hartsdale Ave., White Plains, NY 10607(914) 761-1100Derivation of Refere
11、nceSignals for TelevisionCamera Color EvaluationPage 1 of 2 pagesREAFFIRMED 2003X Y Z = XRYRZRXGYGZGXBYBZB R G B = NPM R G B 5.2 The inverse of the NPM predicts linear RGBfrom XYZ. Since the XYZ values for the “N” testcolors are the starting data, the ideal RGBvalues can be computed by premultiplyin
12、g theseXYZ values by the inverse of the NPM:R G B N= NPM 1 X Y Z N5.3 The reference white on the pattern may notbe a perfect white; therefore, the reference cam-era must be white balanced on the white chip ofthe pattern. This is accomplished by normalizingthe R signals, the G signals, and the B sign
13、alsindividually to make the signal value 1.0000 forthe white chip.5.4 These normalized linear RGB values arethen transformed to the nonlinear RGB relativevideo levels using the opto-electronic transferfunction.Annex A (informative)Example calculationA.1 Given the set of CIE XYZ tristimulus values fo
14、r sixcolors and a white as listed below:and a normalized primary matrix:NPM = 0.4123907993 0.3575843394 0.1804807884 0.2126390059 0.7151686788 0.0721923154 0.0193308187 0.1191947798 0.9505321522 and an opto-electronic transfer function:V = 1.099 V0.45 0.099 for 1 V 0.018V = 4.500 Vfor 0.018 V 0A.2 T
15、he following set of ideal RGB signal values for thistest pattern to this set of reference primaries is com-puted. Two sets of RGB data are given: the firstthree columns (unprimed RGB) are the linear RGBsignals, and the last three columns (headed by RGB)are the nonlinear RGB video levels. In this exa
16、mple, thewhite patch is not a perfect white so the linear RGBvalues must be normalized to 1.0 (or white balanced)as indicated in 5.3:Color patch XYZ01020304050607WhiteBlueGreenRedYellowMagentaCyan0.946410.095900.166240.224850.624170.335160.167101.000000.070120.267310.132500.660750.221350.230081.0757
17、40.344800.110890.063400.104650.364930.45425RGBRGB010203040506071.000000.031290.073000.496630.961130.567660.03890 1.000000.052740.343790.033150.636780.105200.287571.000000.360630.072970.053260.010710.364430.448911.000000.132170.239450.703070.980570.752800.000001.000000.193400.580720.138270.798000.299940.528241.000000.595510.239390.194670.048200.598790.66743RP 176-1997Page 2 of 2 pages
