Studies have previously been undertaken on the prediction of perceived spatial brightness aimed at achieving designs that take account of a feeling of brightness of space, and various estimation models and factors affecting the prediction of perceived spatial brightness are known. In this study the authors focused their attention primarily on factors other than average brightness to explain perceived spatial brightness in nonuniform luminance distribution with the aim of proposing an index which encompasses existing knowledge while being based on simple quantities.
The authors set 32 different lighting conditions in an office with north-facing windows and measured the luminance images for each condition. They then asked subjects to evaluate the perceived brightness from two seats, one with the windows in sight and with no windows in sight, thereby providing 64 different luminance distribution and estimation value samples. Ten contrast images of different spatial frequency were decomposed from each luminance image by wavelet transformation symlet6, then AD (ambient directivity, low spatial frequency) and CD (contrast detail, high spatial frequency) were defined on the basis of each variance of contrast image.
The authors used multiple regression analysis to obtain a multiple regression equation to explain spatial brightness using three variables: average luminance, or NB value; AD; and CD. AD lowers spatial brightness, while CD raises it. The estimated accuracy was greater than that achieved using a conventional equation.
After considering the above, the authors suggested NSB (Natural scale of Spatial Brightness), a spatial brightness prediction model based on luminance contrast.
