Measuring colour appearance of red wines. - National Physical ...

NPL Optical Radiation Measurement Club
Optical Technologies and Measurement Network
Annual Meeting
Digital Measurement
of the appearance of red
wine – colour appearance
Wednesday 27th – Thursday 28th June 2007
Globe Room, Bushy House, National Physical Laboratory

Martin, M.L.G.; Ji, W.; Luo, M.R.; Hutchings, J.; Heredia, F.J.
(2007)
Measuring colour appearance of red wines.
Food Quality and Preference, 18, 862–871
Department of Colour and Polymer Chemistry,
University of Leeds, Leeds LS2 9JT, United Kingdom
Laboratory of Food Colour and Quality, Department of Nutrition and
Food Science,
University of Seville, 41012 Sevilla, Spain

Aim:
- To investigate the colour appearance of red wines in different
depths,
- To explore methods for quantifying the colour appearance of liquid
foods,
- To verify the Pridmore’s finding (2005) that perceived hue shifts
from red at shallow to green at deep for some red wines.
- To quantify the relationships between visual observations and
physical measurements of colour for a series of wines with reference
to the change of depth.

Four Wines: Table red, Oloroso,
Tawny port and Rosé
These four wines represent the colour range and
characteristics of colour change of wine products.

Four Wines:
These four wines were poured into a Petri dish at
different depths of 1.0, 2.5, 5.0, 7.5, 10.0 and 15.0 mm

Four Wines in :
I II
III
1200 mm
60°
Φ 82 mm
Φ7.5 mm
870 mm
1870 mm
45°
Φ7.5 mm
The three regions of wine colour in the cocktail glass
used for physical measurements and psychophysical
experiments

Physical measurement
Two instruments were used:
- a Minolta CS1000 tele-spectroradiometer (TSR)
and
- a calibrated single-lens reflex digital camera (Nikon D1X)
The measurement results were presented in CIECAM02
lightness (J), colourfulness (M) and hue composition (H)
(CIE, 2001).

• A VeriVide viewing cabinet with diffuse/0 geometry
• A diffused D65 simulator
• A GretagMacbeth ColourChecker DC chart was used
for camera calibration.
White backing
Black backing
Viewing windows
Petri dish
Camera

Visual assessment (magnitude estimation method)
in terms of lightness, colourfulness and hue.
•8 normal colour vision observers (Ishihara Test);
•between 30 and 45 min for viewing;
•3 females and 5 males;
•aged between 24 and 41
All observers had abundant experience in using magnitude estimation
methodology but for most of them it was the first time estimating colour
appearance of wine samples.
They were all research staff or research students either in the Department of
Colour Chemistry, University of Leeds (UK) or the Department of Food
Science and Nutrition of University of Seville (Spain).
The whole experiment was conducted at Leeds.

Wines
(4 ×
3 + 4 ×
6 ) × 3 ×
2 = 216
×
8 = 1728 observations

Lightness 100
Colourfulness 40
L* = 52
a* = 31
b* = 7
Munsell
four unique hues: red, yellow, green and blue
10RP5/6

Results and discussion

Coefficient of Variation
CV
=
Statistical Measures
100×
1
N
( Y − f × X )
2 sets of data: CV =0% perfect agreement.
Coefficient of determination
N
∑
i=
1
CV = 30% 30% disagreement
2
n xy x y
r
[ ( ) ] [ ( ) ]
2
2 2
2 ⎟ n x x n y y
⎟
⎛
∑
− ∑ ∑ ⎞
= ⎜
⎝ ∑ − ∑ ∑ − ∑ ⎠
i
Y
i
2
i
2

Mean (r2 / CV) Lightness Colourfulness Hue
Observer
accuracy
0.88 / 17 0.67 / 37 0.99 / 10
Observer
Repeatability
0.84 / 19 0.72 / 30 0.99 / 8
• colourfulness is the worst performance;
• changing in viewing and illumination for
liquid sample;
• repeatability > accuracy;
• mean results were used.

Observer data
b v
10.0Y
7.5Y
100
5.0Y
80
60
40
20
1
2.5Y
2
1
3
10.0YR
2
4
5
7.5YR
5.0YR
2.5YR
10.0R
5.0R
2.5R
10.0RP
7.5RP
0 1
6 2 3 5
5.0RP
5 4 4 6
-40 -20 0 20 40 60 80 100
-20
-40
3
2.5P
6
1
3
5.0P
a v
4
2
7.5P
5
6
2.5RP
10.0P
7.5R
Red
Tawny
Oloroso
Rosé
Lv
100
80
60
40
20
0
1
1
1
2
2
2
1
3
4 5
3 3
4
5 5 6
6
4
6
2
6
5 4
Constant hue loci for Munsell
Table red: depth ↑, lightness ↓, colourfulness ↑ then ↓, hue ≈
Others: depth ↑, lightness ↓, colourfulness ↑ ,
hue oloroso
≈ 10YR-5RP, Rose
3
0 20 40 60 80
Cv
Value5
2.5RP;
≈ 2.5RP-5RP;
tawny ≈ 7.5R-2.5R visually significant.
Red
Tawny
Oloroso
Rosé

What colour are they ???
I II
III

There is some visual evidence for a brownish green appearance of
the shallow oloroso (Pridmore et al., 2005).
Two out of eight observers reported a slight green appearance but
these were masked out when the panel averages were calculated.
The effect to the naive wine observer, however, can vary amongst
individuals, i.e. some viewers see the effect and some do not.
Members of the sensory panel used in the experiment are experts in
describing the colour appearance of opaque surface colours. They
are wine colour naive, presumably not expecting to see the counter
intuitive green tinge in red wine.
This sensory phenomenon requires attention in future work.

TSR
b cc
10.0Y
7.5Y
1
100
5.0Y
80
60
40
20
2
2.5Y
10.0YR
3
1 2
1
2
7.5YR
5.0YR
2.5YR
10.0R
5.0R
2.5R
10.0RP
-40 -20
0
60
20 40 4
5
60
7.5RP
5.0RP
80 100
-20
-40
5 4
1
2.5P
4
5
6
3
3
3 6
5.0P
a cc
2
7.5P
5 4
6
2.5RP
10.0P
7.5R
Red
Tawny
Oloroso
Rosé
Jcc
100
80
60
40
20
0
1
1
1
2
6 5 4
1
2
2
3
3
6
3
3
6
5
2
4
4
4
5
0 20 40 60 80
Mcc
5
6
Red
Tawny
Oloroso
Rosé
similarity with observer results, except oloroso hue,
a hue change from the greenish yellow Munsell 7.5GY (approximately hcc
= 119, corresponding to Hcc = 132, i.e. 32% of yellow and 68% of green)
to 7.5R as depth is increased. (see oloroso Point 1)
This agrees with the results of the visual estimates of the smaller group of
panel members.
The instrument detected the green hue effect.

cc
TSR on cocktail glass at Regions I, II and III
10.0Y
100
7.5Y 5.0Y
80
60
40
20
2.5Y
10.0YR
7.5YR
5.0YR
2.5YR
10.0R
0
7.5RP
5.0RP
-40 -20 0 20 40 60 80 100
-20
-40
III
I
II
III
III
2.5P
II
I
I
II
5.0P
acc
I
II III
7.5P
7.5R
2.5RP
10.0P
5.0R
2.5R
10.0RP
Red
Tawny
Oloroso
Rosé
Jcc
100
80
60
40
20
0
III
II
I
III
III
I
I
II
II
0 20 40 60 80
Mcc
I
II
III
Red
Tawny
Oloroso
Rosé

Digital Camera
b cd
10.0Y
7.5Y
100
5.0Y
1
80
60
40
20
2.5Y
1
2
10.0YR
7.5YR
5.0YR
2.5YR
10.0R
-40 -20
0
0 20
3
40 60
7.5RP
5.0RP
80 100
-20
-40
2
6
2
54
5
4
3
2.5P
1
3
3
5.0P
a cd
6
4
4
7.5P
5
6
5
2
6
2.5RP
10.0P
7.5R
5.0R
2.5R
10.0RP
Red
Tawny
Oloroso
Rosé
Good agreement between TSR and Digital Camera
100
80
60
40
20
0
1
6
1
2
1
5
2
1
4
2
3
3
3
3
4
0 10 20 30 40 50 60 70 80
The digital imaging measurements also show the greenish-yellow
appearance (at 2.5GY in Munsell hue, and around 116 Hc value in
CIECAM02 hue scale) for the shallow oloroso.
Jcd
5
6
Mcd
4
2
5
4
6
5
6
Red
Tawny
Oloroso
Rosé

Compare
results
between:
Visual,
CS1000,
and
DigiEye
Lv
Lv
L_CS1000
100
80
60
40
20
100
80
60
40
20
100
0
0 20 40 60 80 100
0
0 20 40 60 80 100
80
60
40
20
J_CS1000
J_DigiEye
0
0 20 40 60 80 100
J_DigiEye
Cv
Cv
C_CS1000
80
60
40
20
0
0 20 40 60 80
80
60
40
20
0
0 20 40 60 80
80
60
40
20
C_CS1000
C_DigiEye
0
0 20 40 60 80
C_DigiEye
Hv
Hv
H_CS1000
360
300
240
180
120
60
0
360
300
240
180
120
60
0
360
300
240
180
120
60
0
0 60 120 180 240 300 360
H_CS1000
0 60 120 180 240 300 360
H_DigiEye
0 60 120 180 240 300 360
H_DigiEye

Compare results between: Visual, CS1000, and DigiEye
Observer vs. CS1000
Observer vs. DigiEye
CS1000 vs. DigiEye
r 2 0.83
CV 21
r 2 0.84
CV 20
r 2 0.99
Lightness
Colourfulness
0.87
21
0.89
21
0.90
CV 5 19 5
0.9 Hc
0.98
11
0.98
11
0.99

Conclusions:
1. A study has been carried out with measurements of
the colour appearance of different wine samples;
2. observer accuracy and repeatability were considered to
be acceptable;
3. The mean visual results from all observers were used in
the following comparisons with instrumental method;
4. Tele-spectroradiometer and digital camera results agreed
well;
5. Observer and physical measurements show good
agreements for lightness and colourfulness;

Conclusions (continuous):
6. For hue there was some discrepancy between observer
data and physical measurement: greenish-yellow hue was
reported;
7. Instruments detected this greenish-yellow hue;
8. All assessment and measurement methods successfully
depicted the changes of colour that occur with depth;
9. The use of colour appearance methodology represents a
significant step forward in the study of wines;
Finally, the digital non-contact method introduced here
has many advantages for quantifying colour appearance of
liquid food products;