Publication detail

Automatic Determination of the Position of a Tennis Player Using a Pair of Cameras

KŘÍŽ, P. ŘÍHA, K. DUTTA, M.

Original Title

Automatic Determination of the Position of a Tennis Player Using a Pair of Cameras

English Title

Automatic Determination of the Position of a Tennis Player Using a Pair of Cameras

Type

conference paper

Language

en

Original Abstract

This paper describes an image processing method for an automatic determination of the position of a tennis player on a playground for further analysis of his motion. A testing code was written in C++ with use of OpenCV libraries. The processing chain consists of several image processing techniques such as a camera calibration, background subtraction or finding the largest contours. The main objective has been achieved by finding and using a 2D homography (perspective projection matrix for points which lie on the pedestal plane) between a reference image of the first view and sequential captured images of the second view and vice versa. There were also used standardized dimensions of the tennis court for a determination of appropriate world coordinates system. Finally, there was used a linear triangulation technique for calculating 3D coordinates of corresponding points and output results discussed in context of real measurements and compared with detection using a pair of cameras. The advantages of this described method are especially variability and simple installation of a pair of sport cameras and real time data processing. The method was chosen and tested in the context of a future work. A camera calibration using external special patterns is not required.

English abstract

This paper describes an image processing method for an automatic determination of the position of a tennis player on a playground for further analysis of his motion. A testing code was written in C++ with use of OpenCV libraries. The processing chain consists of several image processing techniques such as a camera calibration, background subtraction or finding the largest contours. The main objective has been achieved by finding and using a 2D homography (perspective projection matrix for points which lie on the pedestal plane) between a reference image of the first view and sequential captured images of the second view and vice versa. There were also used standardized dimensions of the tennis court for a determination of appropriate world coordinates system. Finally, there was used a linear triangulation technique for calculating 3D coordinates of corresponding points and output results discussed in context of real measurements and compared with detection using a pair of cameras. The advantages of this described method are especially variability and simple installation of a pair of sport cameras and real time data processing. The method was chosen and tested in the context of a future work. A camera calibration using external special patterns is not required.

Keywords

background subtraction; camera calibration; corner detection; contour; homography; triangulation

Released

05.07.2017

ISBN

978-1-5090-3981-4

Book

Proceedings of the 40th International Conference on Telecommunications and Signal Processing (TSP 2017)

Pages from

763

Pages to

768

Pages count

6

BibTex


@inproceedings{BUT138285,
  author="Petr {Kříž} and Kamil {Říha} and Malay Kishore {Dutta}",
  title="Automatic Determination of the Position of a Tennis Player Using a Pair of Cameras",
  annote="This paper describes an image processing method for an automatic determination of the position of a tennis player on a playground for further analysis of his motion. A testing code was written in C++ with use of OpenCV libraries. The processing chain consists of several image processing techniques such as a
camera calibration, background subtraction or finding the largest contours. The main objective has been achieved by finding and using a 2D homography (perspective projection matrix for points which lie on the pedestal plane) between a reference image of the first view and sequential captured images of the second view and vice versa. There were also used standardized dimensions of the tennis court for a determination of appropriate world coordinates system. Finally, there was used a linear triangulation technique for calculating 3D coordinates of corresponding points and output results discussed in context of real measurements
and compared with detection using a pair of cameras. The advantages of this described method are especially variability and simple installation of a pair of sport cameras and real time data processing. The method was chosen and tested in the context of a future work. A camera calibration using external special patterns is not required.",
  booktitle="Proceedings of the 40th International Conference on Telecommunications and Signal Processing (TSP 2017)",
  chapter="138285",
  howpublished="electronic, physical medium",
  year="2017",
  month="july",
  pages="763--768",
  type="conference paper"
}