Font Size: a A A

Visual Soccer Analysis Based On Multi-view Associated Tracking

Posted on:2020-12-21Degree:MasterType:Thesis
Country:ChinaCandidate:X MengFull Text:PDF
GTID:2427330623963642Subject:Computer technology
Abstract/Summary:PDF Full Text Request
Football is a sport that has a long history and is widely loved.A successful football team is not just about a dozen people on the pitch,but also a complete training,analysis,coaching team behind it,and the basic youth training system.With the development of scientific concepts and advances in computer technology,more and more people are beginning to study the use of modern technology to replace some of the low-creative traditional human work,and use more convenient quantitative analysis,visualization and other technologies to assist football professionals.There are three key steps in this process: one is the collection of data.How to quickly and easily capture and convert the game video data into a series of statistics we need.The second is the further analysis and calculation of data,how to extract more valuable deep statistical data from the basic data,and quantify it.The third is how to visualize it in a clear and concise way,so that football professionals who are not good at computer technology and are not sensitive to numbers can easily understand and use such tools.The core of this paper is also to start from these three steps.First,this paper proposes a multi-view multi-target tracking structure to solve the basic tracking problem of this project.Because the stadium is wide and the player's goal is small,and the number of players is large and their appearances are similar,an algorithm suitable for tracking in large-scale scenes is designed.The information fusion of multiple perspectives with high confidence is obtained.Secondly,in order to solve the situation that tracking data needs more precise positioning,this paper improves the broad learning system and proposes a new broad learning method to assist the tracking module for further detection and positioning.At the same time,we use broad learning system to train a marking systems to assess the performance of players and teams,serving official competitions and youth training activities.Third,this paper designs a number of supporting visualization models for visualizing the football data information obtained.The information used includes traditional football statistics and some statistics presented in this study,some of which are used to assist the youth team and the other to quickly show the opponent's information and recent matches to the professional coach.Fourthly,this paper designs experiments for each proposed method,including the experiment of multi-angle tracking algorithm,the field tracking experiment of football daily training,the detection precision experiment of broad learning system,and the physical statistics and GPS comparative experiment of football game.At the same time,we have compiled the application of this research in the U20 stadium of the Tianjin National Games,and conducted a case study on the use of the coaches of the Shanghai team.The method proposed in this paper has achieved good results in multiple tasks such as multi-view tracking,physical statistics analysis and data visualization.Our research provides some new ideas for the topic of football combined with computer technology.Some of the innovative and challenging new issues we have put forward have laid the foundation for future research work.
Keywords/Search Tags:multi-view tracking, broad learning system, soccer tactics, visual analytics
PDF Full Text Request
Related items