Soccer Player Tracking Using Blob Assignation

이미지 블롭 할당을 이용한 축구 선수 추적

  • Park, Kyuhyoung (Dept. of Media Technology, Graduate School of Media Communications at Sogang University) ;
  • Changsoo Je (Dept. of Media Technology, Graduate School of Media Communications at Sogang University) ;
  • Yongdeuk Seo (Dept. of Media Technology, Graduate School of Media Communications at Sogang University)
  • Published : 2003.10.01

Abstract

In this paper particle filter is used as an underlying algorithm to track multiple objects, which are soccer players. Multi-object tracking becomes difficult when two or more players get close to and overlap each other because particles of the filters tend to move to a region of higher posterior probability. To resolve this problem, a blob assignation algorithm which identifies the separated image blobs after occlusion, based on the predicted states according to the dynamic model is suggested. This method performed well on the sequences under general camera work.

Keywords