Detection And Tracking Based On Number Identification Player

Sports are very popular in modern world. The software of game video analysis is favored by many of the sponsors and the fans. So the research of game video analysis technology is of great significance. Based on the insufficient of current game video analysis technology, is proposed a method in this pape based on recognize of player number to track a certain player. On the basis of the method with player number of the location and identification, we achieved for a particular player tracking, and its trajectory is described. The main content of this paper as follows:Firstly, player number localization. In order to improve the applicability of this method, image segmentation should be done for the frames of the game video first. Then we establish a sample template set combine shape and character feature compare with the candidate area in binary image. We successfully position the number on the player movement region and lay a foundation for the next step of number identification.Secondly, identify player’s team. For the player in different team of has the same number, we must identify player’s team so as to make sure the results of player number recognition are unique. In this paper we extract the image of player region based on the position of player number. Then we use the similarity measure to estimate the player’s team.Thirdly, player number recognition. We preprocess the image of player number normalization in order to improve the accuracy of identification. Because of all kinds of action is possible to make in the game, so the players number may appear all sorts of different degrees of distortion. Combination this particularity of the number, this paper puts forward the template matching recognition algorithm based on affine transformation to recognize the player number. Through experiment, the method of this article proposed has good adaptability to the special environment of the court. The average of different number identification accuracy reached more than80%.Finally, the certain player tracking and trajectory description. We applied the method of player number identification in the tracking of a particular player. In this paper, we adopt the algorithm of the SSD tracking based on regional information to track a certain player. Finally, we correspond to the tracked result and the course of the template image. Thus the complete trajectory of the player formed.The automatic technology of analysis in sports video has been continuously improve, we will strive to this certain player tracking method based on the player number identification into the actual broadcast of the game. So as to make the existing sports video analysis system more perfect.