The Surround Color and Color Matching Functions

Many research results indicated that the surround condition of image display could be a very important factor for image appearance. The most significant impact of the surround on image appearance is the change in perceived image contrast. The chromatic perception of the image will also depend on the color of surround. In this research, a psychophysical experiment was carried out to investigate surround color perception and its relationship with color matching functions and the impact of the macula. The first phase of the experiment was color matching across the center of the viewing field and the peripheral viewing field. The second phase of the experiment is verifying that the systematic color shift shown in the results of phase one was surround dependent but not device dependent. The result also showed that the 10° CMFs (color matching functions) are better than 2° CMFs to predict the tristimulus values of the display color when surround was considered. But neither 10° CMFs nor 2° CMFs could accurately predict surround color. A non-linear model was designed to predict the surround color perception. This model used a spectral filter, which was designed based on the density of Macula pigment. The results showed that this model could successfully predict surround color. The result of this research will help to better understand the impact of the surround on color and image appearance. Introduction When people attempt some color management tasks, the viewing conditions of the display will be a very critical variable; otherwise the color management might well be meaningless. Surround, as well as illuminant condition and some others, are very important components in the viewing condition. Based on the definition used in color appearance models [1] for the simple color patches, the surround could be defined as the viewing field, which is 10° outside of the central stimuli. But for imaging applications, the surround is dependent on the application setup. In some cases where images are displayed on the display panels, the surround could be thought as the area outside the display panels. In some other cases, where the images are displayed as hard copy, the surround could be light booth or the peripheral area in the viewing room. Previous research showed that the most significant impact of the surround on image appearance is changing the perceived image contrast. Based on the classical Bartleson’s result [2] or Hunt’s summary [3], the physical gamma ratio for average, dim and dark surround conditions would be 1:1.25:1.5 in order to perceive approximately same image contrast. Although some other research [4], [5] showed different gamma ratios for those three conditions, they all showed the same trends. Besides the perceived image contrast changing, the perceived chroma in image elements will also depend on the surround relative luminance. But this influence is a question that remains to be answered. [6] Another application, that is mixed chromatic adaptation, is also closely related to surround conditions. The mission of CIE TC8-04 is "To investigate the state of adaptation of the visual system when comparing soft-copy images on self-luminous displays and hard copy images viewed under various ambient lighting conditions." (http://www.color.org/tc8-04/) The surround condition can be treated as ambient light conditions if there are no other light sources in the viewing setup. Some researchers [7], [8], [9], investigated the color appearance influenced by ambient lighting conditions. Their results indicated that the ambient light caused subtle color shifting (10%~20%). These results were based on the achromatic color matching method and fixed state of chromatic adaptation. While, Katoh’s result [10] indicated more adaptation shift (40%) if the adaptation is not fixed. This research was directly motivated by the experiences from the previous surround research [5], in which the perceived image was measured under different surround conditions. In that experiment, the observers were forced to adapt to the surround condition every 30 seconds. During the adaptation time, the image display panel was colorimetrically set to the same color as the surround color, and the observers could freely look around the surround or display. But some observers complained that colors on the display panel and surround were not matched especially for those colors close to the neutral color. If the surround color and central display color were colorimetrically set to same color or same CIE 1931 tristimulus values, the observer should perceive the same color across central display and surround. In other words, the ambient light is metameric to the central display panel. We can easily draw the conclusion that it should be no different in fixed state of adaptation or unfixed state of adaptation. So why did those observers complained about mismatching across the central display panel color and surround color? There were two possible reasons to explain this shift. One possible reason is that the device models for LCD display and LED surround were not accurate. The other possible reason is that they were perceived as different colors by observer even if tristimulus values were the same. Let’s assume our device model is accurate enough, although we knew that it is impossible to make zero color error across these two different devices. When we go back to check the workflow, it was easily confirmed that the physical spectral distribution curves of the surround and central display panel are not the same, they are at best metameric matches. It was also found that the color matching functions were not the same in this scenario either, an issue of observer metamerism. The surround color was made up by diffused LED light, while the central display color came from an LCD display. They have significantly different spectral power distributions. They should be the same color when the tristimulus values were same to the degree the CIE color matching functions represent our observers. And we used the same color matching function applying on the two spectral curves to get the tristimulus values. The 1931 standard observer, also called the 2° color matching functions, was used in that research. But the viewing field is much larger than 2° viewing angle when we compare surround and central display panel. In this scenario, the color matching happened across the central retina and the peripheral retina. There was one big difference between those two parts of retina, which is the macular pigment. The macular pigment protects the fovea, located roughly in the center of the retina, temporal to the optic nerve. It is a small (the diameter is about 1.5 mm) and highly sensitive part of the retina responsible for detailed central vision. 1931 2° color matching function already included the impact of macula. But peripheral retinal vision does not have the impact of macula. This could be the potential reason for explaining the perceptual color shift across surround and central display panel. The hypothesis examined in this research was that the surround color was actually different from the central display even if the tristimulus values measured by colorimeter are same. If this is the case, different color matching functions will have different impact of measuring the surround color and central color. And a non-linear transformation of color matching functions using a filter based on the macula density could help to model the surround color perception. The experiment details below were designed to evaluate the hypothesis.