Capture,Recognition And Analysis Of Badminton Players Swing

As people pay more attention to physical health and quality of life,more and more people are keen to exercise through various sports,such as badminton,which is suitable for all ages.With the development of the new era,artificial intelligence and robots have gradually entered people's lives.Among them,the badminton robot is an intelligent robot that can compete with people in real time.The robot can not only capture and track the high-speed movement of badminton,but also has a high-speed motion control system to accurately complete the swing action.In addition,if the module for capturing,recognizing and analyzing human movements is added to it,it can be upgraded to a badminton teaching robot,which solves the current situation of badminton scholars is much more than badminton teachers.Therefore,the main research content of this article is based on the badminton robot vision system to capture,recognize and analyze the badminton player's swing.The main works are as following:(1)A method to obtain the video segment of the athlete's swing motion based onthe flight direction and position of the badminton is proposed.In the vision system of the badminton robot,this method was used to collect and establish a dataset(BSMDataset)which containing 8 types of common badminton swings;(2)Improve the dense trajectory algorithm so that it can more effectively recognizethe badminton swing motion(in-class motion).Study the dense trajectory algorithm,propose a non-fixed-length dense trajectory feature,and use the target detection algorithm to detect the position of the athlete.By only extracting the feature points' trajectory in a small range that containing the athlete,which not only improve the robust of the algorithm,but also reduce the complexity of the algorithm.The effectiveness of non-fixed-length trajectories is verified through experiments,and the recognition rate of badminton swings is also improved;(3)Defining standard swing motions and analyze swing motions.From the BSMDataset data set,we select the standard motions of each type of swing motion,and characterize these motions with the non-fixed-length dense trajectory algorithm,and then use these standard motions as a benchmark to calculate the similarity between the motion to be analyzed and the standard motion Then use the scoring formula to evaluate and analyze the badminton swing;(4)Designing and implementing the software of badminton swinging action evaluation system.Using the badminton swing motion capture,recognition and analysis methods proposed in this paper,a badminton swing motion evaluation system based on a badminton robot was developed.The system can be used on the visual system of the badminton robot to capture,recognize and analyze the swing movements of the athletes.