To reduce image processing effort, the user draws a simple shape over the video image, representing the confines within which the action unfolds: an arena. Additional shapes may identify zones of interest. Multiple arenas may be drawn if the video image looks down on multiple confined spaces. A 96-well plate (representing 96 arenas) is typically used when observing insect or fish larvae. The term arena setup is used for the collection of all user-defined shapes in a video image. Clearly, making an arena setup can involve a considerable amount of work.

It happens that a carefully drawn arena setup is misaligned with the video image during the experiment. This happens for instance, when an experiment is interrupted and resumed some days later, or when a camera accidentally gets knocked. The arena setup then has to be manually corrected for the new situation.

The object of this assignment is to automate the correction (and optionally creation) of an arena setup, by automatically (re)matching drawn arena’s to the underlying video image using principles of image recognition.

Your background
We are looking for students who are interested in software engineering in a scientific setting, in combination with computer vision. Some knowledge of Windows XP and C++ is required.