Tennis video shot classification based on support vector machine

Video shot classification is one of the key technologies to achieve fast retrieval and browsing video. A novel approach of tennis shot classification based on SVM is proposed. After extracting sobel edge pixels ratios based on window as a classification feature, the optical flow measurements including foreground tracked points ratio (FPR) and mean length of motion vectors (MLV) are also calculated for classification. In the end, achieve the shot classification of tennis video by the way of support vector machine (SVM). Experiment shows that the method can better complete the shot classification of tennis video.