Development of a semi-automatic tracking system for soccer players by using particle filter

Introduction Soccer players are required to give high-level physical performances in a match. The total distance covered and the number of high-intensity running efforts are commonly used as a physical performance marker of soccer players in a match. Theseparameters canbeobjectivelymeasuredbyvision-basedtrackingmethods. however, thetrackingofsoccerplayersisoftendifficultduetoocclusionsituations in which a tracking target is shielded by other objects. The occlusion situation is the most important problem to solve in order to achieve improved tracking. Therefore, many image processing researchers have studied it as an attractive research topic and they have proposed various methods for the automatic tracking of players (D’orazio and leo, 2010). unfortunately, automatic tracking has not been perfected and the many proposed methods have sometimes tracked soccer players in only a part of a game.