Tracking of the motion path of a person from video streams for the overlapping case
Aye Pa Pa Mya, Myint Myint Sein · 2009
The tracking and recognition of human motion, action, and events using computer vision has recently gained widespread interest in both academic research and industrial, with much emphasis on real-time systems. This paper presents a fast and accurate method for tracking the motion path of a person from the video stream. The motion data of the people are acquired in possible cases such as occlusion or overlapping with each other. This system provides to extract the motion of each player individually from successive image frames. Especially, the trajectories for each player are obtained in the whole video sequence of sport game and in every instance of time for using the sport training system. Singular value decomposition of Gaussian Elimination method is applied to determine the matched person from previous image. The motion trajectory of a player is considered not only for normal case but also for overlapping case. Through the experimental results, the proposed system works adequately.