Ball Tracking System for Soccer Video by Switched Searching Method
Kazuki Yano, Tetuya Takiguchi, Yasuo Ariki · 2007
In this paper, we propose a ball tracking system for soccer videos by switching search methods between global search and local search. At first, the global search is employed to detect a ball on the first frame using a normalized cross correlation method. Then the search method is switched to the local search of the ball using a particle filter. While tracking, when the ball is lost, the search is switched to the global search method and the ball is tracked by the particle filter again. We carried out three ball tracking experiments; global search, local search and the proposed switching search method. As a result, the experiments showed the effectiveness of the proposed