Nathan Kendrick (MSc completed 2005)

I am currently involved in gait analysis work. My thesis involves filming people running on a treadmill. I used an equal number of trained track and field runners and untrained people from the general public. Here is a frame of one runner:

If you notice, this runner is wearing a number of reflective balls that can be picked up by 8 motion capture cameras (two of which are seen in the above photo). Thus, in addition to conventional video, motion capture data is also stored. This data can be used to make computer models of each runner. The following picture is a model of a different runner than the one above:

My research on this data involves two parts. First, I can show videos of these computer models to 'experts' in the field of running. These include national and international track and field coaches and athletic therapists. These experts tell me what they notice about each runner, what they feel is good or bad about their form, what could be improved, and who they feel is trained or not.

The second part of my research on this data is programming the computer. I can use a variety of techniques on the conventional video to find the runners, and analyze their forms. Below are ten binary images, with the white parts being what the computer thinks is the runner.

Many other things can be done, such as making a silhouette of the area a runner carves out when running:

Use techniques such as video phase locked loops, eigenvalue decomposition, as well as data such as stride length and body mass, the computer can try to emulate the experts. It can make decisions on a runner's form, and ultimately tries to make the same decisions (or even better decisions) than the experts can make.