The effect of the size of a stimulus on the perception of self-luminous colours

Transcription

1 The effect of the size of a stimulus on the perception of self-luminous colours Author: Jeroen Wattez Address: Schonenberg 16, 9550 Herzele Tel: 0495/ Promotors: Martijn Withouck, ir. Wim De Geest, Ing. Hugo Coolens, Ing. Kenneth Labiau Education: Katholieke Hogeschool Sint-Lieven; Light & Lighting Laboratory of KU Leuven Key words: Perception of colour, self-luminous colours, colour size effect. 1. Abstract Theoretically, when being exposed to stimuli of different sizes, the perception of colour can be partially predicted by employing eye-sensitivity curves. In this experiment, the difference between the perception of self-luminous, coloured stimuli with a FOV (Field Of View) of 2 and 10 will be examined. Based on an already existing experimental procedure, examinees give their idea on brightness, hue and the percentage of white for a sequence of stimuli with 2 and 10 FOV. A set-up is built and the psychophysical tests with the examinees take place. Afterwards, the difference in perception between 2 and 10 stimuli is analysed and compared with the theoretical expectations. It is found that there is no difference in percentage of white and hue between the two FOV's. For brightness on the other hand, some conclusions can be drawn. Stimuli with a FOV of 10 appear brighter than stimuli with a FOV of 2 unless the stimulus of 2 is more saturated. In that case, the stimulus of 2 looks brighter due to the Helmholtz-Kohlrausch effect. This effect seems to dominate the size-effect. 2. Introduction People s perception of colour depends on a variety of parameters such as the brightness of the background, the environment, the size of the object, etc... It is a problem people, such as interior designers, have to deal with on a daily basis. Dyes that are chosen in a shop based on a sample, do not appear the same when being painted on a wall. Additionally, the surrounding light will also play a role in the perception of the painted wall. Therefore, predicting the appearance of colours is necessary. The purpose of this experiment is to investigate the difference between the perception of self-luminous, coloured stimuli with a FOV of 2 and Theoretical Background As is well-known, the light-sensitive receptors in the retina play an important role in the understanding of how people perceive colours. These receptors, the cones and rods, are not proportionally distributed. Except for one spot, the rods dominate the cones in the whole retina. This special spot is very important in this investigation and is called the fovea. In the fovea, the cones are very concentrated and no rods are present. This spot lies right on the optical axis and is the place on the retina where images can be produced very accurately. The cones also are not distributed proportionally. In the centre of the retina (2 ) there are almost no β-cones. These non-proportional distributions are an important parameter in the difference between the colour perception of images with a FOV of 2 and 10. Stimuli with a FOV of 2 create an image on the retina that is completely focused in the fovea. Therefore, only cones will be stimulated. Stimuli with a FOV of 10 conversely, are also depicted on the retina outside the boundaries of the retina. Consequently, the ratio cones-rods and the types of cones that are stimulated vary, so a slight change in the eye-sensitivity curve is necessary. For this reason CIE provides two different standards of eye-sensitivity curves: CIE 1931 and CIE Both are illustrated in figure 1. It is clear that a significant difference occurs at short wavelengths, i.e. the blue spectrum. Previous studies investigating the colour size effect show that stimuli appear brighter and more colourful with an increase of size[2]-[9] but that there is no difference in the hues [2]-[5]. This article describes the observation of a number of psychophysical experiments to investigate a difference in colour appearance due to colour size effect. Therefore, examinees were exposed to a sequence of self-luminous,

2 coloured stimuli with a FOV of 2 and a FOV of 10 and gave their opinion about the brightness, hue and the percentage of white. Experiments [10] have shown that naive observers found it difficult to evaluate colourfulness, that is why the amount of white is being asked to the observers. The experiments occur in a completely dark, black box. 4. Methodology Colour appearance for 30 different colours with a FOV of 2 and a FOV of 10 are assessed. These colours are selected from the CIEu' 10 v' 10 -chromaticity diagram in a way that they almost cover the whole colour gamut. These colours are projected in figure 2. Next, a set-up is built, confirming the FOV of 2 and 10. Finally, examinees are tested. In this experiment, ten persons aged between 20 to 30 are observed. They all passed the Ishihara 24 plate Test for Colour Blindness and the Farnsworth-Munsell 100 Hue test. The examinees are divided in two groups in order to avoid that some differences can be caused by the order of appearance. Figure 1: CIE1931 and CIE1964 eye-sensitivity curves [1] Figure 2: The 30 stimuli used, represented in the CIEu' 10 v' 10 -chromaticity diagram 5. Set-up In figure 3, the set-up of the experiments is visualized. The box is 74 cm wide, 54 cm high and 28.5 cm deep. The depth is selected in such a way that the opening with a FOV of 2 has a diameter of 1 cm and the opening with a FOV of 10 a diameter of 5 cm. It is important to know that the box is totally black inside, but it is more understandable when different colours are used to clarify its usage. The black parts in the image fancy movable objects. The large one is a slider that contain an opening with a FOV of 2. The 4 tiny blocks on the edge of the box, on the other hand, are used to hold this slider. This set-up can be used in three positions (figure 3). Led-modules are placed behind the two fixed openings. Figure 3: Set-up of the experiments

3 6. Psychophysical experiments Ten persons, 5 male and 5 female, with ages ranging between 20 and 26 are tested in this experiment. These observers are divided in two groups who will be asked the same questions and will be exposed to the same colours but in a different sequence. First, the persons endured a training, so they were familiar with the concept, the specific terms and the colours used. a. Group I In the first test, the examinees took place in front of the box and the slider was placed before the opening on top (as in figure 3b). Then, a fixed sequence of stimuli appeared at random, but specified, through the opening above or below, so just one led-module was always active. Each stimulus was available for 15 seconds and between two stimuli, no colour was displayed during three seconds. In total, 75 samples had to be evaluated (30 with a FOV of 2, 30 of 10 and 15 verification samples). In this test, the observers had to estimate the percentage of white they saw in the particular stimulus and also which hue (blue, yellow, red or green). If they saw two hues, they also had to verify the ratio of the two. In the following test, the observers had to match two stimuli in terms of brightness. In practice, one of the two ledmodules showed a reference-white, the second one another sequence of the 30 samples. First, the reference-white appeared above, the samples below (figure 4a) and the brightness of the reference had to be modified to the samples. This means that the person has to adjust the brightness of the reference-white by saying it has to be more or less intense. In a few steps, equality between the two stimuli is reached. This had to be done 30 times, for each stimulus. In the next series of 30 samples, each stimulus had to be justified to the reference white. Then, the positions of the samples and the reference-white changed (figure 4b) and the same had to be done. This switching is done to make the experiment place-independent. When all this is over, the slider was placed before the opening below (like figure 3c). Again, brightness had to be adjusted four times, in the same order as before (first like in figure 4c and then like in figure 4d). Finally, the first test had to be repeated, but also with the slider below. This means that the same fixed sequence was used and examinees had to evaluate the percentages of white and the hues. To have a certain reliability of the examinees, they also had to adjust brightness when the same stimuli were displayed above and below with the same FOV. Figure 4 Possibilities to match two stimuli in terms of brightness b. Group II The second group has done the same experiments, with the same colours and the same sequences, but the slider with the opening for a FOV of 2 was first placed below. This is also an addition to be sure that the experiment is place-independent. Another renewal is the adjustment of the brightness. As mentioned, the observers had to adapt the reference-white to the colours and vice versa. When in the first group the stimulus that had to be changed in intensity was initial more intense than the other stimulus, it was made less intense in the second group. When it was less intense, the second group saw initial an adjustable stimulus that was more intense. This change is made to restrain the phenomenon that people, averagely, stop too soon in changing the brightness. 7. Results a. Percentage of white Before comparing the results for stimuli with a FOV of 2 and 10, a comparison has to be made between the stimuli seen above and below (figures 5a and 5b). For both FOV's is checked whether there is a distinction between the two positions. Both graphs show that there is no difference so for this experiment, the results are placeindependent. Now, the percentage of white for stimuli with a FOV of 2 can be compared to those with a FOV of 10 (figure 6). Observers experienced difficulties estimating the percentage of white declaring the large error bars

4 (based on the standard deviations) represented in the figure. Nevertheless, it is clear that there is no real difference in perception between stimuli with a FOV of 10 and stimuli with a FOV of 2 for the amount of white in a sample. Figure 5 Comparing the percentage of white or the hue for the stimuli seen above or below b. Hue As in the previous section, stimuli above and below are compared in terms of the hue (figures 5c and 5d). Again, no hue-difference can be noticed between stimuli that are seen in both positions. The hue is transformed into an quantifiable number by an algorithm so it can be plotted. The results for stimuli with a FOV of 2 compared to stimuli with a FOV of 10 are plotted in figure 7. There is no notable hue-difference, just as in the previous section. In addition, observers had less difficulties defining the hues, declaring the small error bars. c. Brightness As mentioned, there are eight different situations for adjusting the brightness. This can be reduced to four conditions because the experiment is place-independent. This means that the situations in figures 4a and 4d are actually the same just as the situations in figures 4b and 4c. There are still four cases because once the referencewhite has to be adjusted and then the samples have to be modified.

5 Figure 6: Percentage of white for stimuli with a FOV of 10 and 2 Figure 7: Hue for stimuli with a FOV of 10 and 2 i. Case I In the first case, the set-up is like in figures 4a and 4d where the reference-white (2 ) has to be adjusted so it would appear equally bright as the samples (10 ). The abscissa in figure 8a is the luminance of the different samples. The ordinate shows the luminance of each reference-white that is adjusted to a sample. In the graph, each luminance of the reference-white obtains the colour equal to which sample the reference-white is compared to. This makes it easy to find out what samples differ the most in comparison with the expectations. In the graph, it is clear that the luminance of the reference-white is always higher than the luminance of the samples. This means that the reference-white needs more luminance to look as bright as the samples. For red, blue and green, the difference between the two luminances is the greatest. ii. Case II In the following case, the set-up is like in figures 4b and 4c where the samples (10 ) have to be modified so they would appear as bright as the reference-white (2 ). In figure 8b, the results are plotted. The stimuli lie in a vertical line because each sample is compared to the same reference-white (which has a constant luminance). The graph shows that the achromatic samples need more luminance to look equally bright as the reference-white. The chromatic colours on the other hand, need less luminance than the reference. The blue stimulus shows the largest difference. iii. Case III This case uses the same set-up as in the previous case, but now the reference-white (10 ) has to be adjusted so it would appear equally bright as the samples (2 ). A similar result is found (figure 8c). When the reference-white is compared to the achromatic samples, it needs less luminance to look equally bright. Being compared to chromatic stimuli, the reference-white needs more luminance. Again, the blue, red and green stimuli differ the most. iv. Case IV In the last case, the same set-up is used as in case I but now the samples (10 ) have to be modified to the reference-white (2 ) so they would appear equally bright. In figure 8d, the results are plotted. Except for one sample, each stimulus needs less luminance than the reference-white to look as bright. The blue stimulus has the greatest difference, just as in the previous sections.

6 Figure 8: Results for the brightness of the four situations 8. Discussion Based on the results, it can be stated that there is no difference in the perception of the amount of white and the hue for stimuli with a FOV of 2 and 10. This confirms the expectations and the results of previous experiments. For the brightness, some remarks are necessary. In the first case, the results confirm previous investigations, namely that the small size appears less bright than the large FOV. This is why the stimuli with a FOV of 2 need more luminance to appear equally bright as the stimuli with a FOV of 10. In the following case, the results are more complex. It emerges that the achromatic samples are still susceptible to the size-effect, which means that

7 they look less bright in a smaller FOV. But, chromatic samples seem to look brighter than the reference-white, even though they have a smaller size. This is a consequence of the Helmholtz-Kohlrausch effect, which states that a sample looks brighter than another one with the same luminance level if it is more saturated. So if the chromatic samples (2 ) are compared to the reference-white (10 ), they still look brighter because they are more saturated. This experiment shows that the Helmholtz-Kohlrausch effect is stronger than the size-effect. This is confirmed by case III, which has the same results. In the last case, the size-effect and the Helmholtz-Kohlrausch effect add up. Samples (10 ) are compared to a reference-white (2 ). The chromatic samples look already brighter because they are more saturated, and this is reinforced by the fact that they have a greater size. 9. Conclusion It can be concluded that stimuli with a FOV of 2 and 10 appear to be the same in terms of amount of white and hue. If there is no colour-effect, the stimuli with a FOV of 10, look brighter than those with a FOV of 2. Nevertheless, if one stimulus is more saturated than the other, this one will appear brighter. Thus the Helmholtz- Kohlrausch effect dominates the size-effect. In further experiments, it can be investigated how saturated a stimulus with a FOV of 2 has to be to appear as bright as a reference-white with a FOV of 10. Also, the same experiment can be done in other circumstances (luminous background,...). References [1] Udal, A. (2012) LUMEN (Photometry and Colorimetry) [Online]. Retrieved May 18, 2015 from [2] Hsieh T.-J. and Chen I.-P., Colour Appearance Shifts in Two Different-Sized Viewing Conditions. Color Res. Appl., 35: doi: /col.20580; 2010 [3] Fu, C., Li, C., Cui, G., Luo, M. R., Hunt, R. W. G. and Pointer, M. R., An investigation of colour appearance for unrelated colours under photopic and mesopic vision. Color Res. Appl., 37: doi: /col.20691; [4] Xiao, K., Luo, M. R., Li, C., Cui, G. and Park, D., Investigation of colour size effect for colour appearance assessment. Color Res. Appl., 36: doi: /col.20610; 2011 [5] Xiao, K., Luo, M. R., Li, C. and Hong, G., Colour appearance of room colours. Color Res. Appl., 35: doi: /col.20575; 2010 [6] Xiao, K., Luo, M. R. and Li, C., Colour size effect modelling. Color Res. Appl., 37: doi: /col.20650; 2012 [7] Kutas, G. and Bodrogi, P., Color appearance of a large homogenous visual field. Color Res. Appl., 33: doi: /col.20367; 2008 [8] Gombos, K., Schanda, J. Interrelationship between size and brightness dimensions of appearance. In: CIE Expert Symposium on Visual Appearance, Paris, France; 2006 [9] CIE TC1-68, Effect of stimulus size on colour appearance; Vienna, Austria: Commission Internationale de l'eclairage; 2011 [10] Withouck, M., W.R. Ryckaert, K. Smet, and P. Hanselaer. 2012b. Colour Appearance Modelling for selfluminous colours. Paper read at Predicting Perceptions: Proceedings of the 3rd International Conference on Appearance, April 17-19, 2012, at Edinburgh, UK.

8 Curriculum Vitae Personalia Name: Jeroen Wattez Address: Schonenberg 16, 9550 Herzele Date of birth: 2 februari 1990, Zottegem Nationality: Belgian Cell phone: 0495/ jeroen_wattez@hotmail.com Driving license: B LinkedIn: be.linkedin.com/in/jeroenwattez Education Master of Science in Photonics Engineering Thesis: Shape & size effects on the luminescence of single microparticles University of Ghent Master of Science in Industrial Engineering: Elektronics-ICT, main subject Electronics Engineering (Magna cum laude) Thesis: The effect of the size of a stimulus on the perception of self-luminous colours Katholieke Hogeschool Sint-Lieven Bachelor of Science in Industrial Engineering: Elektronics-ICT, main subject Electronics Engineering Katholieke Hogeschool Sint-Lieven ASO Sciences-Mathemetics (8 hours) Onze-Lieve-Vrouwcollege Zottegem, campus Deinsbeke Summer Job Thuiszorgwinkel CM Zwijnaarde Driver/Storehouse Language skills Speaking Reading Writing Dutch ****** ****** ****** English **** ***** ***** French *** **** **** Skills - MS Office - Java - LabVIEW - MATLAB Hobbies - Football (18 years in competition) - Cycling - Swimming