Analyzing Movements of Tennis Players by Dynamic Image Processing

Recently, we can easily obtain a fast computer and a digital video camera as their prices go down. In this paper, we describe a method and a system for analyzing performance of tennis players of a given singles tennis game. Our method consists of two phases. The first phase is an image processing phase, and the second phase is an evaluation phase using an evaluation function. First, we assume that a movie of the game has been shot by a digital video camera which is fixed at a place where the camera could film whole area of the tennis court and the two players. We transform the movie to a sequence of the constituent images on the computer. From the images, we detect the positions of the players for each time point by making differences of the images with a image of the court without the players, which we can synthesize from the sequence of the images. As the only moving objects in the images are the players and the players moves rapidly in the image, concentrating a specific pixel, the players images is not on the pixel for most time period. We, therefore, decide the background color for the pixel by computing the median of the sequence of the colors of the pixel of the time frames. Although we are able to detect player’s positions relatively easily because the players are large in the image and they do not move very fast, it is hard to detect the position of the ball in the image. In most cases, the image of the ball is blurred. So we detect the landing sites of the ball with human assistance. We, moreover, determine the position and the time of the hit with human assistance. We interpolate ball’s position in each frame from the landing site position and hit position. Using these data, we also obtain the speed of the ball. Second, we analyze the player’s action using the positions of the players and the ball detected by techniques we mentioned above. We analyzed the both players of the game by defining an evaluation function. The function is defined as a linear combination of six sub-evaluation functions. The function are depends on the positions of the both players, the landing site position and the velocity of the ball. We made a prototype of a tennis-player-evaluation system by implementing the present method (See Fig. 1 and Fig. 2). The system appropriately evaluated performances of players of some sample tennis games, although the evaluation function cannot analyze the player’s tactics enough. We have to work more to refine the evaluation function and develop a more reasonable evaluation function which is able to evaluate player’s tactics. That is for future work.