The population equalization algorithm is proposed for color image quantization. It selects adequate colors fast, the number of which is the size of the video look-up table, so as to display an image with least degradation of image quality. After calculating 3-d histogram, partitions in the color cube are determined using the criterion that picture elements should be equally distributed among partitions. An enhancement of the error diffusion algorithm is also proposed to recover degraded image quality due to quantization. Combination of above algorithms makes it possible to reduce the number of colors down to 64 without perceptible degradation when a usual image is displayed on a high resolution CRT, whose pixel size is about 0.27mm by 0.27mm.
Subjective test results, which are concerned with the ease in TV watching when displayed in the small size and the sensitivity to noise disturbance, are presented. The optimum viewing distance is about 9H, where H is the picture height, and it is independent of the screen size. The optimum conditions of screen size and viewing distance are 13 inches and 1.2 m, respectively. The degree of disturbance due to random noise depends on the viewing angle, and the S/N ratio when the degree of disturbance becomes equal decreases about 6.6 dB/dec. as the viewing distance becomes far.
Apparent motion was investigated under light-adapted condition approximate to the standard viewing condition of television. Two 2°×2° square patterns were presented on monochromatic CRT display with a variety of spatial separation (SS) and stimulus onset asynchrony (SOA). Though the results of evaluation varied somewhat from subject to subject, good motion impressions were common to the subjects at a restricted range of SS (less than about 2°) and of SOA (less than about 100 msec). Exposure duration of the pattern had produced little effect on the evaluation. Explanation was given to these results on the basis of retinal receptive field model.
To determine if an apparent motion for chromatic stimulus occurs, we conducted subjective evaluation-tests using complementary color pair of equiluminous red-green/yellow-blue square patterns. Stimulus patterns were presented on the face of a color CRT over a wide range of the spatial separation (SS) and the stimulus onset asynchrony (SOA). At short-range of the SS (maximum about 15) and of the SOA (maximum about l00 msec), it was perdeived as a smooth movement both in red-green and in yellow-blue patterns.
Characteristics of vision, whose are spatio-temporal frequency response, spatio-temporal transient phenomena, phase correlation between two eyes to brightness channel and chromatiness channel of vision, and movement perception of chromaticness channel, were measured by psychophysical experiment. We found that experiment results can be explained well by parallel pathway of color vision. Especially, the phase of temporal dominant system in r-g channel has inphase to r-component of spatial dominant system and counterphase to g-component.