Assessing the application of an image color appearance model to basic self‐luminous scenes

Image color appearance models (Image CAMs) have been developed to predict the perception of complex scenes and are mainly used for image rendering and video reproduction applications. Among these Image CAMs, iCAM is an Image CAM that takes an image as the input and provides the perceptual attributes for each pixel. On the other hand, nonimaging CAMs are widely used and validated, but they always assume a simple test scene of a uniform flat stimulus, a quasi‐neutral background, and a surround. This study presents an evaluation of the performance of iCAM when applied to these simple self‐luminous scenes in predicting the influence of background luminance, background size, saturation, and stimulus size on stimulus brightness. The results show that iCAM is capable of predicting the effect of background luminance and some background size scenarios. However, for unrelated self‐luminous stimuli (dark background), the model predictions do not match the reference data. An evaluation of the effect of the filter kernel size and its relation to the physiological mechanism of image processing inside the visual system has been investigated. Furthermore, the impact of saturation and stimulus size on brightness seems to be underestimated by the model, because the Helmholtz‐Kohlrausch and stimulus size effects are not included. Hence, these findings call for an enhanced Image CAM.

Jaar van publicatie:2019

BOF-keylabel:ja

IOF-keylabel:ja

BOF-publication weight:0.5

CSS-citation score:1

Authors from:Higher Education

Toegankelijkheid:Closed