Keiji Uchikawa

Categorical color perception of Japanese observers: Comparison with that of Americans

Ten native Japanese observers named 424 colors of the OSA Uniform Color Scales set using monolexemic color terms of their choice. The results are compared with those from seven American subjects previously studied by Boynton and Olson. It is concluded, in full agreement with the original thesis of Berlin and Kay, that there are eleven basic color terms in each language, each of which describes a fundamental color sensation dependent upon an underlying physiology that does not differ between the two groups.

Influence of basic color categories on color memory discrimination.

Two color-memory experiments were performed to investigate whether observers tended to confuse colors with a smaller color difference in memory or colors in a same color-category region. We made color stimuli on a color CRT. Color difference was determined by a simultaneous color discrimination experiment. Color-category regions were obtained by a categorical color-naming experiment using the 11 basic color names: white, black, red, green, yellow, blue, brown, orange, purple, pink, and gray. The results show that two colors with a certain color difference can be confused more easily when they are in a same color category than in different color categories, and that colors identified with memory tend to distribute within their own color-category regions or their neighbor color-category regions, depending on their positions in a color space. These findings indicate that color memory is characterized by the color categories, suggesting a color-category mechanism in a higher level of color vision.

Saccadic suppression of achromatic and chromatic responses measured by increment-threshold spectral sensitivity

We measured spectral-sensitivity functions during saccadic eye movement by the increment-threshold method to test whether saccades selectively suppressed achromatic or chromatic responses. A circular monochromatic test stimulus of 12-deg diameter was presented for 10 ms on a 62 deg x 43 deg white background, and observations were made under three conditions: during fixation, during 6-deg saccades, and immediately after saccades. In two additional conditions the test stimulus was made to move during fixation and during 6-deg saccades at the same speed and in the same direction as the saccades. The during-fixation spectral-sensitivity function was found to resemble the relative luminous efficiency V (lambda) function in shape except for the case of short wavelengths, whereas the during-saccade spectral-sensitivity function showed lower sensitivity for all wavelengths and had three prominent peaks at approximately 440, 530, and 600 nm. These characteristics did not depend on whether the stimulus was stationary or moving. These results indicated that saccadic suppression was greater for achromatic than for chromatic response. A possible suppression mechanism was discussed involving the magno and parvo pathways.

Limitations of surface-color and apparent-color constancy

Color-constancy mechanisms have been studied and discussed in a number of investigations. However, there has been little attempt to reveal how color constancy deteriorates as the conditions for it become less than optimal. We carried out a series of asymmetric color-matching experiments, using two criteria: surface-color match and apparent-color match. With brief adaptation the degree of color constancy increased as chromatic cues were added in the surround. In the condition of black surround, the test stimuli appeared self-luminous, and chromaticities of the chosen matching stimuli were the same as the physical chromaticities of the test stimulus, indicating a total deficiency of color constancy. With 15 min of preadaptation to the illuminant, the surface-color matches showed almost perfect color constancy under illuminant change. In both adaptation conditions, the chromatic-shift of matches from what would be expected for perfect color constancy increased gradually between 1,700- and 30,000-K illuminant, as chromaticity of the illuminant departed from 6,500-K illuminant. Under 1,000-K illuminant the surface-color appearance became totally achromatic, and color constancy was completely lost. Our results show that, even with brief adaptation to the illuminant, the contribution of the surrounding stimulus is large enough to achieve a fair degree of color constancy, but complete adaptation to the illuminant helps to achieve almost perfect color constancy.

Example-Based Color Stylization of Images

We describe a new computational approach to stylize the colors of an image by using a reference image. During processing, we take the characteristics of human color perception into account to generate more appealing results. Our system starts by classifying each pixel value into one of the basic color categories, derived from our psychophysical experiments. The basic color categories are perceptual categories that are universal to everyone, regardless of nationality or cultural background. These categories are used to provide restrictions on color transformations to avoid generating unnatural results. Our system then renders a new image by transferring colors from a reference image to the input image, based on these categorizations. To avoid artifacts due to the explicit clustering, our system defines fuzzy categorization when pseudocontours appear in the resulting image. We present a variety of results and show that our method performs a large, yet natural, color transformation without any sense of incongruity and that the resulting images automatically capture the characteristics of the colors used in the reference image.

Temporal deterioration of wavelength discrimination with successive comparison method

Abstract The wavelength discrimination threshold (Δλ) was measured with a successive comparison method, in which two stimuli of different wavelengths were presented in the left and right half of 2.2° or 2.7° bipartite field, respectively, with a certain stimulus onset asynchrony (SOA). The stimulus duration was kept constant at 110 msec. The Δλ value for five different wavelengths, 430, 470, 520, 570 and 610 nm, was found to be constant up to SOA= 60 msec and increased gradually till SOA= 190 msec. The wavelength discrimination functions with SOA of 0 and 550 msec were also obtained for 430 through 650 nm in 10 nm steps. At all wavelengths the values with SOA of 550 msec were about twice as large as those with SOA of 0 msec. The present results indicate that the wavelength discrimination deterioration is complete within a relatively short period of time.

Influence of achromatic surrounds on categorical perception of surface colors

Color samples selected from the OSA Uniform Color Scales set were seen isolated in a dark field, illuminated by hidden projectors. These appeared as self-luminous aperture colors when thus isolated. We employed a categorical color-naming procedure to assess color appearance. Achromatic surrounds of 33 min width, if adjacent to samples subtending about 2.2 deg, were sufficient to render normal categorical surface-color perception. As the size of surrounds decreased, color naming shifted from that normally observed in the surface-color mode to that appropriate to the aperture-color mode. For isolated samples, brown was almost never seen, being most often replaced by orange; a white border less than one-sixtieth the width of the color samples was sufficient to restore its perception in an otherwise dark field. The reflectance of the surround and the gap between test and surround stimuli were also examined and found to be important factors in surface color perception, whereas the overall luminance level was not.

Apparent size of an object remains uncompressed during presaccadic compression of visual space

It is well known that compression of visual space occurs near the saccade goal when visual stimuli are briefly flashed at various locations on a visual reference just before a saccade. We investigated how presaccadic compression of visual space affected the apparent size of an object. In the first experiment, subjects were instructed to report the apparent number of multiple bars briefly presented around the time of saccade onset. The reported number of four bars began to decline at the 50 ms mark before a saccade and reached a minimum near the saccade onset. This confirms that the compression of visual space occurs just before saccades. In the second experiment, subjects judged the apparent width of a rectangle (a single element) or four bars (four elements) presented just before saccades. The apparent width of the four-bar stimulus was compressed just before saccades, but that of the rectangle stimulus was not compressed. Experiment 3 shows that the width compression of the four-bar stimulus is consistent with the width change predicted by compression of position. These findings indicate that the shape of a single object is not distorted at the saccade goal during presaccadic compression of visual space. In addition, experiment 4 indicates that the apparent width of a flashed stimulus just before saccades depends on the processing of global shape. This extends the definition of a visual object during presaccadic compression of visual space to not only a solid element but also a constellation of multiple elements. Furthermore, the results from these experiments suggest that presaccadic compression of visual space does not prevent object recognition underlying an attentional mechanism in generating saccadic eye movements.

Example-based color stylization based on categorical perception

We describe a new computational approach to stylize the colors of an image by using a reference image. During processing, we take characteristics of human color perception into account to generate more appealing results. Our system starts by classifying each pixel value into one of a set of the basic color categories, derived from our psycho-physiological experiments. The basic color categories are perceptual categories that are universal to everyone, regardless of nationality or cultural background. These categories provide restrictions on the color transformations to avoid generating unnatural results. Our system then renders a new image by transferring colors from a reference image to the input image, based on this categorizations. To avoid artifacts due to the explicit clustering, our system defines fuzzy categorization when pseudo-contours appear in the resulting image. We present a variety of results and show that our color transformation performs a large, yet natural color transformation without any sense of incongruity, and that the resulting images automatically capture the characteristics of the color use of the reference image.

Temporal integration of chromatic double pulses for detection of equal-luminance wavelength changes

Detection probabilities for wavelength changes were measured as functions of stimulus onset asynchrony (SOA) with the chromatic double-pulse method. Two test stimuli of a wavelength lambda t were successively exchanged with a reference stimulus of a wavelength lambda r in equal luminance for a short duration with a SOA. Durations were 5 and 50 msec, and the SOA varied between 5 or 50 and 2000 msec. Lambda rs were 571 and 518 nm, chosen from wavelengths near unique yellow and unique green so that wavelength difference delta lambda = lambda t - lambda r was perceived mainly as redness and greeness for lambda r = 571 nm and yellowness and blueness for lambda r = 518 nm. The results showed that temporal integration characteristics for these equal-luminance wavelength changes were quite consistent: complete integration with SOA up to 20 msec, partial integration with SOA between 20 and 200 msec, and probability integration with SOA greater than 200 msec. They did not show any inhibitory integration that was found for luminance changes.