Graduate Students
Brian Leung
Ph.D. Candidate
Institute of Biomaterials and Biomedical Engineering
University of Toronto
Advisor: Dr. Tom Chau
E-mail: brpw[dot]leung[at]utoronto[dot]ca
Education & Training
Brian completed his Bachelors of Applied Science in Computer Engineering at the University of Ottawa in 2005 and his Bachelors of Science in Honours Mathematics, also at the University of Ottawa, in 2006. His experiences in industry include J2EE web application development with the Software Engineering Group at the National Research Council of Canada and ASIC verification with Tundra Semiconductor. In 2006, he took on a part-time research opportunity with the VIVA lab at the University of Ottawa, in the research area of Bayer demosaicking of digital colour images.
Research Title
Using Multiple Cameras to Realize Robust and Real-Time Facial Gesture Recognition for Children with Severe Spastic Quadriplegic Cerebral Palsy
Research Abstract
It has been recognized in literature on education that interactive learning, such as playing and using computers, is crucial to a child’s cognition and communication skills development. Children with severe cerebral palsy (CP) are at a disadvantage relative to typically developing children because they lack a dependable access pathway to use switch-operated devices for interactive learning. This has been identified as a primary risk factor for higher incidences of developmental and learning problems in children with severe CP.
Some children with spastic quadriplegic CP may be able to exploit facial gestures such as eye blinks and tongue protrusions to reliably operate a computer vision-based facial gesture user interface. Typically, such a system would capture facial gestures using the video input of a single camera placed at a distance from the user. However, single camera systems have limited utility for individuals with severe spastic quadriplegic CP because their extraneous head movements complicate the user task of facing the camera squarely and steadily for proper facial gesture recognition. This thesis will investigate the fusion of video inputs from multiple independent cameras to improve the reliability and detection accuracy of computer vision-based facial gesture user interfaces for these individuals. This treatment of multiple cameras is analogous to having several single camera systems monitoring the same user from multiple viewpoints.
Evaluation of the multiple-camera facial gesture user interface will involve three schoolchildren with severe spastic quadriplegic CP. Each child will be testing the facial gesture user interface once a week, for 20 minutes per session, over a period of four months. They will be using the interface to play a game and work on learning exercises on the computer.
The findings of this research may lead to the future development of a portable device that can be attached to another switch-operated device to readily implement a facial gesture user interface on the latter device. Such a portable device will open up access to a wide variety of switch-operated devices for individuals with severe spastic quadriplegic CP.
Figure 1: User and system setup.
Other Interest
Brian is an avid fan of video/computer games (from classic titles on the NES to the latest games on the Xbox360) and NHL hockey. He enjoys golfing and tennis in the summer and downhill skiing in the winter.
