Tracking of Soccer Players Using Optical Flow

This paper describes and gives an idea of moving objects’ detection and tracking which is useful for sports applications. In sports, the challenging task is to detect and track the motion of players in each video frame. This paper describes a hierarchical clustering algorithm to segment human region (player) based on color and depth followed by optical flow analysis using the Lucas–Kanade algorithm to detect and estimate motions of players. The optical flow used for analyzing large displacement and estimate motions in each frame accurately. For the past few years sports video analysis has received increasing attention, hence player tracking and recognizing trajectories are two major issues which are highly relating. In the proposed system, hierarchical clustering and optical flow together demonstrated on soccer dataset captured at Alfheim stadium to achieve96%of tracking accuracy. In Future, the given system is extended to find the trajectories of each player.