discrimination to spectrally modulated lights Romain Bachy1,4, Jérôme Dias2,3, David Alleysson3, Valérie Bonnardel4
Université Pierre et Marie Curie, Paris, France Laboratoire de Caractérisation des Systèmes Photoniques, CEA-LETI Grenoble, France Laboratoire de Psychologie et Neurocognition, Université Pierre Mendes France Grenoble, France Department of Psychology, University of Winchester, United Kingdom
1. Summary
Using sinusoidally modulated spectral power distributions (SSPDs), we measured the minimum phase difference (Δφ) to discriminate two SSPDs of fixed frequency and amplitude. Hue sensitivity thresholds of two normal trichromat observers were measured under two adaptating backgrounds (low and high CCT) for 12 different phases varying from 0° to 360°. For a given phase, SSPD dominant wavelength is determined and discrimination curves expressed in function of wavelength (Δλ/λ) show similar profiles to wavelength discrimination curves obtained under different adapting lights [1]. Spectrally modulated lights allows us to generalise hue discrimination characterisation to non-spectral hues (i.e. hues located on the alychne).
adaptive procedure under 2 adaptative backgrounds. Their chromaticities located in the vicinity of the Planckian locus u’= 0.19 & v’=0.43 (High CCT, blue point) and u’=0.22 and v’=0.49 (Low CCT, red point) are within the stimuli elliptical contour (Fig 1, 4). Two of the authors VB & RB with normal colour vision served as observers.
3. Results
For each observer, discrimination curves are plotted on the same graph for the two backgrounds (high CCT, blue curve & low CCT, red curve) as Δλ in function of λd_c (Fig 2, right) and as Δφ in function of φ (Fig 3).
Δλ/λd_c Discrimination curves
Empirical discrimination curves (Δφ/φ) are explained by a 3-stages Müllerzone model with a Von Kries adaptation. A linear model (A + B sin(2φ+K), with K depending on adaptation) predicts post-receptoral mechanism parameters and adapting function.
methods
4 0
Relative Energy
700
0
0.2
0.4 u
0.6
0.8
4
500 600 (n m)
500
nm
600
700
Figure 1
Our light stimulator [2] consists in a Xenon light source (Fig 1, 1) illuminating a linear interference wedge. The transmitted light is sinusoidally modulated by an electronic mask (f=1.5c/300 nm; m=1) generated by a black and white LCD display (Fig 1, 2). The observer sees the resulting light (Fig 1. 3) at the aperture of an integrating sphere in Maxwellian view which produces an homogeneous 2° spot of 4cd/m2. When the phase varies from 0° to 360°, the chromaticities of the modulated light describe an elliptical contour in a chromaticity diagram. With 12 different reference phases (φ), discrimination was probed in various colours directions (Fig 1, 4).
Procedure
700
In a 3-interval forced-choice task, observers indicated the interval in which the test (φ+Δφ) was presented. Thresholds were determined by a staircase
180
4 330
400
270
90100
500 600 d (n m)
700
Δφ/φ Discrimination curves RB
VB 20
16
16
12
12
8
4
0
0
0
90
180 (°)
270
360
10 5
90
180 (°)
270
360
0
0
90
180 (°)
270
360
Figure 4
L(φ)= AL + BLsin(φ+KL) (same for M- and S-cones), with A, B and K determined by the experiments.
5. Conclusions
Δλ/λd_c Discrimination curves obtained with SPDs show similar profiles
to those determined with monochromatic light tests, with in particular the same background effect that improves hue discrimination in the vicinity of the adapting background colour;
A model accounting for adaptation and post-receptoral mechanisms provides an approximative fit to empirical curves, however, its inability to accurately predict inter-maxima distance and their shift with background adapation suggests the existence of non-linearities.
6. References
• [1] Jameson, D. And Hurvich, L. M., (1964). Theory of brightness and color contrast in human vision. Vision Research, 4, 135-154. • [2] Bonnardel, V. Bellemare, H. & Mollon, J. D. (1996). Measurements of human sensitivity to comb-filteres spectra. Vision Research, 17, 2713-2720.
8
4
10
15
with matrix M corresponding to Von Kries adaptation and opponent chromatic coding, theoretical thresholds are thus expressed by a sinusoidal function with A, B and K determined to best fit individual empirical curves (Fig 4).
210
Figure 2
20
15
Lo w C C T M o d e l Hig h C C T M o d e l Lo w C C T Hig h C C T
1/Δφ2 = X’(φ)TMTMX’(φ) = A+Bsin(2φ+K)
8
0
20
20
At a post-receptoral level accounting for Von Kries adaptation and opponent chromatic mechanisms, we can write :
For comparison purposes with canonical wavelenght discrimination curves, thresholds (Δφ) were converted into nanometers (Δλ, with 1° = 0.55nm), and plotted in function of the dominante or complementary wavelenght (λd_c). For the two observers, thresholds are lower in the orange-red part of the spectrum in the low CCT and lower in the green part of the spectrum in the high CCT condition. Dicrimination improvement in the vicinity of the background chromaticities is similar to the background effect reported in wavelenght discrimination by Jameson and Hurvich (1964) (Fig 2, left).
(°)
SPDφ(λ)
Relative Energy
0 400
Dependance of wavelenght discrimination on the background chromaticity with luminance level of test equal twice of that of the background. Upper: experimental measures Lower: theorical functions derived from opponent induction hypothesis. Adapted from Jameson & Hurvich (1964)
0.2 0
φ
φ
(n m)
600
0.4
0
Lo w C C T M o d e l Hig h C C T M o d e l Lo w C C T Hig h C C T
In our model cone responses to phase shift is expressed by:
120
90
(°)
v nm
100
12 270
RB
VB
0
RB 180
500
210
d
0.6
2
3
400
4. Model
5
330
0.8
Fφ(λ)
1
340
uv dia gra m
1
0 400
240
320
Experimental device 1
180
8
(n m)
2. Materials &
VB
12
low and 0-30° in the high CCT condition, and at 100° in the low and at 90-120° in the high CCT condition. From low to high CCT conditions, second maxima are shifted toward higher phases (from 180° to 210° for RB & to 210-240° for VB) and minima towards lower phases from 330° to 270° for RB and to 320° for VB.
(°)
1: 2: 3: 4:
(°)
Hue
0
90
180 (°)
270
360
Figure 3
Curve profiles show 2 mixima and 2 minima. The first maxima and minima are approximately at the same location for the two backgrounds: at 30° in the
7. Acknowlegdgements
RB thanks the Colour Group (Great Britain) for the attribution of the W. D. Wright Award to attend the 33rd ECVP. Author’s emails:
[email protected],
[email protected],
[email protected],
[email protected]