Validation of Image Filters for Studies of Visual Accessibility

摘要

An environment is visually accessible if a person can rely on vision to travel efficiently and safely through it, to perceive its spatial layout, and to update location and orientation within the environment. We are studying how architectural and interior design decisions and viewing conditions interact with vision deficits of people with low vision to determine the visibility of obstacles and other important features. Deficits in acuity and contrast reduce visibility of key features and increase confusability between features in a space, e.g., rendering a step invisible or confusable with a shadow boundary. Here, we describe an image-filtering method intended to reveal the featural information available in an arbitrary scene for an observer with a specified level of reduced acuity. Images were filtered using thresholded bandpass filtering techniques developed by Peli (1990), together with contrast sensitivity functions associated with different levels of reduced acuity and contrast sensitivity. We validated the filtering method by applying it to photographs of letter-acuity charts. We measured the performance of normally-sighted subjects using these filtered images to determine each filter's effective acuity. The resulting calibrated filters can be used to predict the visibility of features in architectural spaces.