BME/CENG/ELEC/SENG 499
Design Project

 

BME/CENG/ELEC/SENG 499 Projects

Supervisor: Alexandra Branzan Albu, aalbu@ece.uvic.ca (Nov. 2008)

AA1 Design of interactive image processing software for the K-12 curriculum

This project is part of a larger recruitment-oriented initiative aiming at rendering the field of computer vision accessible to high-school students. One of the most exciting features of computer vision is its direct applicability to a variety of domains: computer aided diagnosis in medical imaging, human motion analysis for video games, image enhancement for artistic photography etc. Therefore, we believe that computer vision is an inspiring example which shows that abstract mathematical concepts (for instance matrix manipulation and signal convolution) and algorithms have direct contributions towards solving real-world problems.

The project will consist in identifying the topic and, upon topic approval, designing one computer-based application that involves image processing. Regarding the choice of the topic, the challenge consists in finding the right balance between the fun element and the complexity of mathematical and algorithmic elements (which need to be understood at least at a basic level by highschoolers) involved in image processing. For instance, a face recognition algorithm using wavelets would not be an appropriate choice since it is almost impossible to explain wavelets using highschool math concepts.

An essential element for the success of this application is the design of the image database. This database should contain images with an appropriate level of difficulty (for instance, identification of human silhouettes in a cluttered outdoor environment could be too challenging for simple motion detection algorithms). Ideally, the user of the application should be able to enhance the database with his/her own images (taken with a cell phone for example).

The application needs to be designed in a programming environment that allows for easy, iterative prototyping. Matlab and Visual C++ are ideal choices, since both come with rich libraries of built-in image processing functions (Matlab Image Processing Toolbox and Open-CV library for C++). The design of a graphical user interface for direct image manipulation is also required.

AA2 Contact Supervisor

I am interested in supervising projects involving human motion analysis from video sequences. The database for such projects is already in place.