Research And Realization Of Tennis Trajectory Measurement System

The tennis trajectory measurement system can overcome the shortcomings of human beings relying on the naked eye to observe.It assists the referee to make more accurate judgments,effectively curbs the referee’s motivation to perform black-box operations,and ensures the fairness of sports competitions.However,the current tennis trajectory measurement system is complicated and difficult and the high deployment cost limits its application.This paper studies and designs a set of tennis trajectory measurement system with high reliability,high accuracy and relatively low cost.The specific work content is as follows:(1)A high-precision extraction algorithm for tennis contour based on three-frame difference,local dynamic threshold and improved YOLOv4(You Only Look Once)model is realized.The tennis contour is roughly extracted in real time by the three-frame difference method.After the initial trajectory of the tennis is detected,the large tennis contour is detected twice through the local dynamic threshold to obtain a more accurate tennis contour.Finally it uses the improved YOLOv4 model to judge the contour of the tennis ball,which reduces the possibility of incorrectly identifying the tennis ball.(2)A calculation method based on the characteristic points of the calibration board and the fixed points of the venue,and the conversion matrix of the venue coordinate system based on RANSAC is realized,which improves the situation that the calibration board is used alone to calibrate the venue.Because the calibration point of the calibration board is not on the ground plane of the tennis court,there is a problem of large errors in the tennis coordinate measurement.At the same time,a calibration board area detection algorithm is implemented to improve the reliability of the traditional calibration board area detection algorithm.(3)A tennis dynamic model and a tennis trajectory recognition algorithm are established.The algorithm evaluates the trajectory candidate points through the four aspects of the tennis trajectory point’s motion direction,motion speed,motion acceleration,and coplanarity,and adopts the concept of full connection.Trajectory matching and reconstruction realizes accurate and real-time identification of tennis trajectory.It also implements the UKF-based tennis landing point prediction algorithm,which can accurately predict the landing point coordinates that meet the accuracy requirements.(4)The design and realization of tennis trajectory measurement system are finished,including tennis contour extraction,coordinate calculation and conversion,tennis trajectory recognition and landing prediction algorithms,and the three dimensional reconstruction of tennis trajectory modules.