Cardiac interbeat interval sequences have been measured during VDT works under the three ambient illuminance levels of 180, 320 and 560 Lx.CFF (Critical Flicker Fusion Frequency) were measured before and after VDT works.The light sources were fluorescent lamps with the color temperature of 4200K.Three luminance contrasts of 3: 1, 5.5: 1 and 10: 1 were adopted on a VDT surface at each of the above-mentioned illuminance levels.
Spectral analysis of the cardiac interval signals showed the presence of the vasom otor blood pressure components centered at about 0.1Hz.The mean heart rate, the power of the vasomotor components of the cardiac signal, and CFF were found to have significant relations with the ambient illuminance and the luminance contrast on a VDT surface.
The results obtained suggest that the visual condition of the ambient illuminance level of 320-560Lx and the luminance contrast of 5.5: 1 causes the minimum physiological load.

View full abstract

The study of pattern recognition are performed by various ways in many labolatories. But no one establishes the definitive theory of pattern recognition. So we tried to perform three ways of character recognition, considering the function of human visual pattern recognition. Especially, we examined the abilities about neural networks that are spotlighted, recently.
The method of temmplate matching that we operated is effective in the printed characters.But this method is weak in the noisy visual patterns and characters getting out of shape. The method of structural analysis we operated is effective in the noisy patterns, but it was impossible to recognize the characters in case of lacking of the feature in part. And it needs a complicated programming in the large system.
The structure of three-layered neural networks is simple, and it copes with input patterns by learning, flexibly. But the learning of the system takes much time. So we designed the improved learning method, “improved backpropagation”. And we successed to decreases approximately 30% of the learning times, using the “improved backpropagation”.

View full abstract

The visual experiment in which observers subjectively estimated chromatic differences for colour chips with lower saturation than Munsell Chroma 5 by means of ratio estimation under the daylight D₆₅ and the illuminant A was carried out, and the estimated vlues were compared with colorimetric colour-differences in the CIE 1976 L*u*v* and the L*a*b* uniform colour spaces.
Three groups of colour chips, namely, 10BG-5R, 6.5PB-5Y and 7.5GY-5P with constant Value 6 were used, which had continuity for transition of saturation extending over opposite hue beyond the neutral point.
Results obtained are summarized as follows:
(1) The correlations between relative values of perceived colour-differences obtained under the daylight D₆₅ and those under the illuminant A are high for all colour groups, and those values change little even if the illuminant changes, irrespective of the degree of saturation.
(2) The correlations between relative values of colorimetric colour-differences obtained under the daylight D₆₅ and those under the illuminant A also are high; the correlation coefficients are nearly 1 for all the colour groups in both the colour spaces.
(3) The L*a*b* colour space is superior to the L*u*v* colour space in terms of compensation for chromatic adaptation. And, as for the colour chips of 7.5GY-5P, the compensation for chromatic adaptation is incomplete for the colours of low saturation in both the colour spaces.
(4) The ratios of perceived chromatic differences to colorimetric ones become relatively large with decrease of saturation, irrespective of the kind of illuminants and colour spaces.

View full abstract

In this paper, the spacings of hue on the Munsell renotation system are examined by use of data of perceived colour-differences: which was obtained in visual experiments by means of the ratio method under an artificial daylight D₆₅, for many pairs of colour chips with constant value and chroma 6/8 and hues as a colour circle divided into 160 nearly equal parts.
The following is cleared as a result of it. It is generally said that the Munsell renotation system will represent perceptual uniformity in each attribute, but the saying is not true as far as at least the attribute of hue respects. The hue-differences in the Munsell renotation system are relatively larger than the perceived ones at about the hue ranges of R to YR, Y to GY, and G to B, and smaller at about the hue range of PB to RP. And the hue of yellow, green, blue and purple in a perceptual uniform colour space would nearly correspond to the hue of 7Y, 10G, 2.5PB, and 9P in the system respectively, assuming that the hue of 5R corresponds to the perceptual red.

View full abstract

The human face has been taken up as one of the main visual targets and the provision of its brightness has been a theme. There are some studies with regard to the necessary luminance of the face. However, with lighting design there are many cases where it is difficult to deal with the luminance, therefore it is desirable to convert this to illuminance, which is easier to work with.
Before now, the face was assumed to be a perfect reflacting diffuser and a method whereby this was converted to vertical illuminance was employed.
From the results of experiments regarding human identification in residential areas, J.F. Caminada et al have concluded that the semi-cylindrical illuminance is more pertinent than vertical illuminance and hemispherical illuminance. From the results of experiments in a dark room, the author has concluded that we can provide for brightness of the face by semi-cylindrical illuminance under the condition where an excess of shadow does not occur.
In this paper, the author investigates the influence of shadow when converting luminance to illuminance in real environments (general indoors, underground passages, parks and residential areas at night), and found the following:
(1) Within these kinds of environments and without regard for the influence of shadows, the face can be assumed to be a perfect reflecting diffuser and converted by means of the formula.
Esc=π·Lf/ρ
where Esc: semi-cylindrical illuminance.
Lf: luminance of the human face
ρ: reflectance (ρ=0.3)
(2) Semi-cylindrical illuminance is more pertinent than vertical illuminance in real environments.

View full abstract