Date of Award:

5-2012

Document Type:

Dissertation

Degree Name:

Doctor of Philosophy (PhD)

Department:

Computer Science

Committee Chair(s)

Heng-Da Cheng

Committee

Heng-Da Cheng

Committee

Brett E. Shelton

Committee

Stephen J. Allan

Committee

Curtis Dyreson

Committee

YangQuan Chen

Abstract

Short track speed skating lends itself to intense competitions with a strong visual impact. Thus, the sport has become increasingly popular. In fact, in 1992, short track speed skating became an official Winter Olympic sport with four events, and four more events were added in 2002. Because of the sport’s growing popularity, there is a high demand from both coaches and TV broadcasters for a means of automatically gathering competition data such as trajectories, velocities, and 2D reconstruction animation. We call this vision-based sports video analysis.

In competitive short track speed skating, multiple skaters skate together on an ice track the size of a hockey rink. Different from traditional longer track speed skating, during short track competition, there are no visible lanes, and skaters usually cluster in groups, especially around curves. Due to this crowding, skaters’ velocity can change dramatically although still moving at a high rate of speed. Additionally, skaters on the same team wear identical uniforms, and the uniforms of different teams can be nearly identical. The clustering of the skaters while traveling at high speeds, and little visual difference between skaters can make sports video analysis difficult to achieve.

While there are many sports video analysis methods currently available, none of them lend themselves well to short track speed skating. This dissertation addresses this problem.

Checksum

71064f44afe483a2d0d3190f12c5b3c4

Download

Share

COinS

Copyright for this work is retained by the student. If you have any questions regarding the inclusion of this work in the Digital Commons, please email us at .