Beta 1


Title Tracking People in Sports using Video Analysis
Author Thirsgaard Rasmussen, Anders
Møller Rasmussen, Henrik
Supervisor Aanæs, Henrik (Image Analysis and Computer Graphics, Department of Informatics and Mathematical Modeling, Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark)
Institution Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark
Thesis level Bachelor thesis
Year 2008
Abstract This thesis describes the analysis and implementation of a system for tracking people using video data. People tracking has many applications in fields such as security, traffic, business, or sports, and the tracking system developed in this thesis is aimed specifically at movement analysis in badminton and tennis. Knowing your opponent's movement patterns and optimizing your own is useful for top athletes, and our system is made to assist in this analysis. The theory behind the system, however, is relevant for many other applications of people tracking. The tracking system in this thesis is created using blob based detection on a background subtracted image. Using 2D homography, the detected image positions are transformed into real world coordinates to generate a map based overview of the movement paths. Theory of the Kalman filter is presented and used in the implementation to model and estimate the movement states based on noisy measurements. Towards the end of the thesis, methods for improving the tracking system are presented, in particular tracking in stereo (using two cameras), and ideas for stabilization of shaking cameras and shadow removal.
Series IMM-B.Sc.-2008-11
Fulltext
Original PDF bac08_11_net.pdf (4.38 MB)
Admin Creation date: 2008-06-25    Update date: 2008-06-25    Source: dtu    ID: 220921    Original MXD