Running state recognition in videos via frames' frequency and positions of two feet

In order to analyze and recognize the running state in athletic sports, this paper proposed a running state analysis method through image processing technique in videos. First, the runner's feet's positions in video sequences (i.e., images) were analyzed via image segmentation. Second, since the frames in every second is 24, then the times of the feet's positions' changes can be obtained, thus to determine the feet's frequencies, i.e., the runner's running state: fast running, slow running or walking, in which a state mode is built for general running state analysis. Experiments proved the method's efficiency.