Skip to main navigation Skip to search Skip to main content

A novel ball tracking method using dynamic kalman filter with two-stage ball search

  • Xuefeng Bai
  • , Qiong Li*
  • , Tiejun Zhang
  • , Xianhua Song
  • , Xiamu Niu
  • *Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

Abstract

Tracking the ball in broadcast soccer video is an important yet extremely difficult problem. In this paper, we proposed a novel ball tracking method using dynamic kalman filter with an improved ball search strategy. In order to re-detect the ball when occlusion, a two-stage ball search strategy is used. In this two-stage ball search strategy, the ball is searched around several players. These players are determined by a distance threshold from the one that suspected to occluded the ball. This ball search result is used to update the parameters of dynamic kalman filter. Experimental results have verified the advantage of the proposed method.

Original languageEnglish
Pages (from-to)3373-3381
Number of pages9
JournalJournal of Computational Information Systems
Volume9
Issue number9
DOIs
StatePublished - 1 May 2013
Externally publishedYes

Keywords

  • Ball tracking
  • Dynamic kalman filter
  • Occlusion handling

Fingerprint

Dive into the research topics of 'A novel ball tracking method using dynamic kalman filter with two-stage ball search'. Together they form a unique fingerprint.

Cite this