Color difference

Evolution of Color difference

Reflectance data
Tristimulous values (X,Y,Z) (2o & 10o) Improve uniformity of color and

introduction of CIE lab Color difference.

Difficulties in subjective methods

Color perception differ from person to

person. A dyer would go for wider tolerance whereas buyer might not. One specific person may approve a certain pair of shade one time and may reject the same pair in other time.

Parameters affecting subjective evaluation

Sample size and dimension.
Illumination level and its type. Distance between the samples under observation. Texture of material.

Color of the surrounding.

Observing condition. (Angle, distance and

duration) Observers variability (Age, attention, attitude, experience).

Backbone of color difference formula

When the color difference between a

standard and a sample is marginal, it may be either:

Just perceptible difference. Just noticeable difference. Just acceptable difference.

Backbone of color difference formula

Color difference formula are derived to

predict perceived color difference between pairs of color stimuli. In industry it is often require to specify the acceptability of color variations between a given standard and its production.

Backbone of color difference formula

When two batches are produced having equal

perceived color difference with standard one may consistently accepted while the other is consistently rejected. The reasons are:

Hue difference less accepted than chroma and lightness. Mismatch in the yellower side is less acceptable. Acceptance practice may differ from industry to industry, within an industry, also with economic condition.

Backbone of color difference formula

Backbone of all presently acceptable color

difference formula is CIE76. All the present formula has derived by dividing DL, DC, DH with a suitable weighing factor which should provide good result some experimental data. Many sets of experimental data has been used to develop a color difference formula. Industry are mostly interested in acceptability data in which sample is judged as commercially acceptable such as Munsell or MaAdam data set.

CIE76 formula
DE = (dL)2 + (da)2 + (db)2

= (dL)2 + (dC)2 + (dH)2 dL + denotes sample is lighter. dL - denotes sample is darker. da + denotes sample is redder or less green. da - denotes sample is less red or greener. db + denotes sample is yellower or less blue. db - denotes sample is less yellow or bluer.

JPC79 color difference formula

EJPC = [(L/Lt)2 + (C/Ct)2 + (H/Ht)2]1/2

Where, Lt = 0.08185L1/(1 + 0.1765L1)

Ct = 0.06381C1/(1 + 0.0131C1) + 0.638. Ht = TCt

T =1 when C1 < 0.638, otherwise T = 0.56 + |0.2Cos(1+168o)|, unless 160o<1<345o when T = 0.56 + |0.4Cos(1+35o)|

JPC79 color difference formula

Advantage

Substantial improvement over CIELab formula in terms of agreement with published data and industrial shade passing.

JPC79 color difference formula

Disadvantage

Poor agreement between formula and visual judgment for very dark colors having lightness value below 16. Anomalous E values for near-neutral samples. Anomalous hue angles and marked change in E values for samples with low tristimulous values. Varying tolerance in hue, chroma and lightness in acceptance data demanding various weighing factors.

CMC(l:c) color difference formula

Different lightness and chroma weights

are used for perceptibility and acceptability judgment.

For perceptibility l:c should be 1:1. For acceptability l:c should be 2:1.

CMC(l:c) color difference formula

E(CMC) = [(L/l.SL)2 + (C/c.SC)2 + (H/SH)2]1/2

Where SL = 0.0404975L1/(1+0.01765L1) unless L1<16 when SL =0.511 Sc = 0.0638C1/(1+0.0131C1) + 0.0638 SH = Sc(Tf+1-f) where f = [(C1)4/{(C1)4 + 1900)}]1/2

T values were kept same as JPC79 formula.

CMC(l:c) color difference formula

The formula was intended to be used only for small color difference and in this case there is no significant problem. This formula was extensively studied by the SDC color measurement committee using available experimental data over 3000 pairs of samples from 10 separate studies and was found to perform better than the CIELAB formula and slightly better than JPC79 formula. British Standard Institution has recommended CMC(l:c) formula for calculation of small color differences.

CMC(l:c) color difference formula

Disadvantage

Does not give satisfactory result in grey and blue region. Formula is too complicated. If standard is interchanged in a pair this would result a different E. If the two samples have a large color difference, the E values could be quite different.

BFD(l:c) color difference formula

Luo and Rigg observed that CMC

formula has some problem in saturated blue and other colors. So they did little modification on CMC formula.

CIE94 color difference formula

E(CIE94) = [(L/KL.SL)2 + (C/KC.SC)2 + (H/KH.SH)2]1/2

KL, KC and KH are parameteric factors that describe the action of change of external conditions of a visual task in relation to standard condition and SL, SC and SH are weighing factors.

Where SL = 1 Sc = 1 + 0.045CX SH = 1 + 0.015CX

CX = CS i,e CIELAB C of the standard sample a midst several

samples from different batches.

Under reference conditions KL= KC= KH=1 for perceptibility while

for acceptability, KL=2, KC= KH=1. Wherever, there is a deviation from the reference conditions, KL, KC and KH should be appropriately adjusted.

CIE94 color difference formula

Reference conditions: The specimens are homogeneous. The color difference between them is below 5.0(CIELAB). The sample is large enough to subtend an angle of 4o to assessor whose vision is normal i,e no color blindness. Samples are kept id direct edge contact. Samples are illuminated in at 1000 lux and viewed in object mode against a uniform grey background of L=50 under an illumination simulating D65.

CIE94 color difference formula

Advantage

Has recommended by CIE. Simplified. Has a problem in grey and blue region. Large discrepancies in predicting lightness difference.

Disadvantage

CIE2000 color difference formula

CIE demand that it is the best formula

among all the formula though it is the ideal one. It is obvious that CIE2000 is more complex than the other formulae. But still now this formula is not extensively used in textile field.

Evaluation of the color difference formula

Color discrimination set can be used for

deriving as well as evaluating performance of color difference formula. However a formula specially designed to fit a particular data set, may result a poor correlation between E and visual data. The reasons are :

Various set of color discrimination data such as Munsell, OSA, MacAdams, PGN etc. have markedly different characteristics and are based on quite different experimental conditions and method of assessments.

Cont.

Evaluation of the color difference formula

It is impossible to cover whole color space a

data set may covers specific color centers. A formula may fit one or more color centers satisfactorily, but fails for whole data set. Error in visual assessment such as fewer and inexperienced observers, improper experimental conditions and errors in instrumental measurements such as low precision instrument etc., mismatch of illuminating and viewing conditions in both visual and instrumental assessments may also result poor corelation.

Best formulae for Textile Processing

Though CIE 2000 is under the process of

evaluation. Among other formulae it has been observed that CMC(2:1) is the best for acceptable data set and being used most of the processing units. Though it is not the ideal one. Conceptually the development of color difference equations is quite advanced, however there is still no perfect agreement between observed and calculated color difference.

Future research in color difference field

To derive a color difference formula based upon

a uniform color space rather than the modification of CIELAB color space. To derive a formula based on color vision theory rather than the use of empirical approach. To derive a formula capable of taking into account different viewing parameters such as size of samples, size of color difference, separation background, luminance level etc.