@MASTERSTHESIS\{IMM2013-06575,
    author       = "S. Careelmont",
    title        = "Badminton shot classification in compressed video with baseline angled camera",
    year         = "2013",
    school       = "Technical University of Denmark, {DTU} Compute, {E-}mail: compute@compute.dtu.dk",
    address      = "Matematiktorvet, Building 303{-B,} {DK-}2800 Kgs. Lyngby, Denmark",
    type         = "",
    note         = "{DTU} Supervisors: Anders Lindbjerg Dahl, abda@dtu.dk, and Prof. Henrik Aan{\ae}s, {DTU} Compute",
    url          = "http://www.compute.dtu.dk/English.aspx",
    abstract     = "The goal of the thesis is to classify shots played during a recorded badminton match. The classification consists of a time chronological list of every shot’s start and end position. Such classification offers tactical information which can be used for coaching or display during broadcast matches. The common camera position in badminton match recordings is behind the baseline of the court. Compressed video files are considered to accommodate home recordings and saved broadcasts.
An algorithm is developed that outputs a classification given a certain input video. The algorithm extract moving objects in every frame and connects them through time. Next, a shuttle trajectory model is fitted to the extracted data. A classification is made based on the best fitting model appearance. The moving object extraction uses background subtraction and connected component labelling. Connecting the components through time is achieved by a custom matching step. The shuttle trajectory model uses a physics based shuttle trajectory description and a camera calibration to establish a model of the image plane behaviour of a shuttlecock.
The results show that on average 8 percent of the shots’ start and end position can be classified correctly. The overall performance is not satisfying but intermediate steps show promising results. Moreover, various bottlenecks are identified and an extensive list of suggestions for further research is given."
}