Abstract
It is vitally important to estimate the visibility of all visual targets in real environments to assume and maintain visual safety. As luminance images in real lit environments are relatively easy to obtain, it is reasonable to attempt a method of estimating visibility of objects using these images.
The authors have proposed C-A graph that can estimate visibility from luminance images using a contrast profile method. The C-A graph presents C-value (luminance contrast) and A-value (adaptation luminance) as vertical and horizontal axes for varying object sizes respectively. This graph expresses the three factors necessary for visibility estimation. In the previous study, the visibility threshold C-value estimation model of circular objects, adoptable to wide range of object size, was established. Furthermore, it suggested that visibility evaluation of people with normal visual acuity can be estimated from the ratio of the visibility threshold C-value.
In this study, as preliminary steps of study for actual low vision people, visibility evaluation experiments were conducted for low visual acuity people using circular objects with uniform or non-uniform (checkered) background. Circular objects and backgrounds with varying background luminance, luminance contrast and object size, were displayed on 27 inch digital display screen. When background was non-uniform (checkered), luminance contrast and size of the checkers were varied, too. The subjects observed circular objects displayed on uniform or non-uniform background with both eyes from a distance of 2500 mm. They evaluated visibility of circular objects using a 5-step scale of 0-4.0=cannot see any object; 1=can see some kind of object, 2=can see a circular object but not clearly, 3=can mostly see a circular object and 4=can clearly see a circular object. Fourteen naked-eye subjects with low visual acuity participated in the experiments.
The experimental results were analyzed using a contrast profile method and were presented on the C-A graph. The experimental result of uniform background suggested that the estimation model of previous study estimates a little higher visibility threshold C-value than the result of low visual acuity people. When the background was checkered pattern, as the checker contrast and the checker size were decreased, the visibility evaluation is similar to the uniform background evaluation. As the checker contrast and the checker size were increased, it would become difficult to recognize the shapes of objects. Furthermore, the authors suggested that if the background has a luminance distribution, it would be better to calculate from a non-logarithmic (raw) luminance image, rather than a logarithmic luminance image.
In the next report, the authors will discuss experimental results for visually challenged people using the non-uniform (checkered) background with the C-A graph.
