| The concept of "high-tech winter Olympics,green Winter Olympics and People’s Winter Olympics" has been deeply rooted in people’s hearts,and scientific organizing has become the key word of this Winter Olympics.How to practice the core concept of science and technology winter Olympics,use science and technology to assist athletes training,help coaches optimize athletes’ movements,enhance athletes’ competitiveness,is a problem that needs to be considered deeply.This thesis focuses on the training of wheelchair curlers of the national team and studies the key techniques and systems of assisted curling training based on video analysis.The main research work of this thesis is as follows:First of all,in the existing curling broadcast video,in order to ensure the audience’s viewing experience,it is necessary to keep the current curling in the center of the camera.Therefore,the camera needs to continue to follow the curling movement,so that the video does not include the complete curling court from the overhead view,and thus the complete movement track of curling cannot be viewed.To solve the above problems,this thesis proposes a global motion based visual object alignment algorithm.Firstly,the target detection algorithm is used to obtain the curling position data,then the Angle matching between two frames is used to estimate the shot motion information between frames,and the data is used to realize the alignment of the curling court.Finally,the complete curling motion track is calculated by combining the curling position data,and the visualization results are output.The feasibility of this method is verified by test.Secondly,in view of the observed curling spinning condition in the process of training requirements,this thesis proposes a curling rotation detection algorithm based on template matching,on the basis of the traditional template matching method,according to the characters of curling surface number sticker place nearby pixels,calculate traversal pixel periphery and center of average,average of two area of the pixel to match respectively,The results were screened to quickly get the position of the number stickers on the surface of the curling stone.The number stickers on the symmetrical position on both sides of the handle were connected,and the vertical direction of the line was the rotation angle of the curling stone.The feasibility of this method is verified by test.Finally,the curling auxiliary training system is realized.Firstly,the system requirements are analyzed,and then the implementation framework of each functional module is designed.Finally,a complete curling panoramic analysis system is realized based on Java Script language,Python language and open source user framework React.The system includes the basic functions of video reading,the display of complete curling movement track,curling rotation information,curling aiming deviation prediction and other functions,and can transfer the track operation information to the virtual display system for display. |
