Abstract
The detection of ramps and steps is important for the safe mobility of people with low vision. We used ramps and steps as stimuli to examine the interacting effects of lighting, object geometry, contrast, viewing distance and spatial resolution. Gray wooden staging was used to construct a sidewalk with a transition to one of five targets: a step up or down, a ramp up or down, or a flat continuation. 48 normally sighted subjects viewed the sidewalk monocularly through blur goggles which reduced acuity to low-vision levels. In each trial, they indicated which of the five targets was present. Here, we report on a probabilistic cue-based model to explain data in the resulting target/response confusion matrices. A set of cues for distinguishing among the five targets included contrast at the transition from sidewalk to target, discontinuities in the edge contours of the sidewalk, and variations in the height in the picture plane of the targets. We formulated the problem of recognition in two parts: the independent probabilities for detecting the cues, and the optimal use of the detected cues in making a recognition decision. To estimate the cue probabilities, we derived and solved equations relating the cue probabilities to the conditional probabilities in the cells of the confusion matrices. We found that the high probability for detecting the contrast cue explained superior visibility of step up over step down. Cues determined by discontinuities in the edge contours of the sidewalk were vulnerable to changes in viewing conditions. Cues associated with the height in the picture plane of the targets were more robust across viewing conditions. We conclude that a probabilistic cue-base model can be used to understand the effects of environmental variables on the visibility of ramps and steps.