Low vision affects many people, both young and old. Low vision conditions can range from near- and far-sightedness to conditions such as blind spots and tunnel vision. With the growing popularity of mobile devices such as smartphones, there is large opportunity for use of these multipurpose devices to provide low vision assistance. Furthermore, Google's Android operating system provides a robust environment for applications in various fields, including low vision assistance. The objective of this thesis research is to develop a system for low vision assistance that displays important information at the preferred location of the user's visual field. To that end, a first release of a prototype blind spot/tunnel vision assistance system was created and demonstrated on an Android smartphone. Various algorithms for face detection and face tracking were implemented on the Android platform and their performance was assessed with regards to metrics such as throughput and battery usage. Specifically, Viola-Jones, Support Vector Machines, and a color-based method from Pai et al were used for face detection. Template matching, CAMShift, and Lucas-Kanade methods were used for face tracking. It was found that face detection and tracking could be successfully executed within acceptable bounds of time and battery usage, and in some cases performed faster than it would take a comparable cloud-based system for offloading algorithm usage to complete execution.
Library of Congress Subject Headings
Assistive computer technology--Design; Low vision--Patients--Services for; Blind, Apparatus for the--Design and construction; Mobile computing; Computers and people with visual disabilities
Department, Program, or Center
Computer Engineering (KGCOE)
Stump, Mark, "Low vision assistance with mobile devices" (2011). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus