Methodology for improving visual content accessibility for low-vision users

Abstract

Nowadays, personalized multimedia services are drawing a significant amount of interest because of increasing usage of multimedia-enabled portable devices. As such, content adaptation according to the perception characteristics of users is important to enhance the accessibility of multimedia content, thus making possible personalized media consumption. This thesis is to propose a framework for image content adaptation for visually impaired people, using the MPEG-21 Digital Item Adaptation (DIA) standard. Among various perception characteristics of users, we focus on the spatial contrast vision characteristic of users, which is represented by a contrast sensitivity function (CSF). Four key contributions of the paper are discussed in more detail. First, we simulate the visual perception of users that are having different spatial contrast vision abilities. This is realized by incorporating the HVS model proposed by Pattanaik et al. Second, in order to measure spatial contrast vision, and thus realizing personalized content adaptation, depending on the severity of the visual ability of a user, we propose a method to measure the CSF for users on computer-based environment. Given a well-defined spatial contrast vision symptom and a measurement of its severity, we also show an example of how to represent this information in an interoperable way by using DIA description tools for low vision users. Third, we perform the actual content adaption, which is personalized in the sense that the adapted content is optimized to the given description of a particular symptom. Lastly, a new contrast metric called perceived contrast is proposed. This metric is used to calculate contrast value for any natural images as well as to determining the CSF. In order to see the effectiveness of the proposed adaptation methods, we performed a statistical experiment for users with the low vision symptom.