Abstract
At this stage, the technical and tactical information acquisition technology has become the key factor to improve the performance of athletes. In this paper, “manual + automation” information collection method was selected. At the same time, in order to improve the speed of information collection, a kind of intelligent prompt automatic completion algorithm was designed and proposed. In the algorithm, the optimization algorithm was improved in view of the poor convergence in the original algorithm, and then the intelligent completion algorithm with more restrictive was further proposed. By collecting the standard document of dynamic standard information and embedding the basic video file, the algorithm was beneficial to the algorithm and automatic completion and intelligent prompt visual features in the video retrieval and analysis. In addition, in terms of technology, the video semantic description was carried out based on the AVI format, and the video retrieval and video analysis based on sports tactical competition were realized, thus providing complete technical support for coaches and athletes in scientific competitions, and improving the level of athletes’ skills and tactics.
Introduction
In recent years, the level of sports competition has become an index to measure the level of development and progress of a country and a nation. In order to maximize the promotion of national competitive level, many countries have used advanced science and technology to help athletes and coaches analyze the game data in real time. Using information technology to analyze the technical index ability of athletes has become the trend of future development [1]. Moreover, many technical teams have made a comprehensive evaluation of the training level of athletes and coaches’ decision making through computer assisted technology, so as to improve the comprehensive competitive level of athletes. In the future, sports computing will be the focus of research in the field of sports development [2]. In view of sports competition and management, the new computer technology can be introduced according to the characteristics of the application object, and thus advanced computer technology can be used to solve various technical problems encountered in sports applications, so as to enhance the athletic ability of athletes [3]. In the field of sports video, the main direction of the research is the detection of scene class and event. Some researchers have used the neural network detection technology to collect information for the exciting scenes in sports videos [4]. According to the location and detection video technology, the research technology of information extraction and score recognition is established. On the basis of this research, knowledge detection of semantic events in sports videos is carried out. The repetitive study and comparative analysis have been carried out for tennis video. In terms of rule, the state of technical action is defined, and in terms of probability and statistics, a fusion analysis is carried out. Therefore, the semantic events of tennis have been completely detected. And according to the specific situation of football match, a separation method of football match is put forward, and according to the characteristics of the research object, the technical action of shooting or scoring is analyzed [5]. Although many research companies have developed sports research software, for commercial purposes, not enough manpower and material resources have been put into use in the sports competition. At present, Dartfish software can only be used for real-time cutting of video game, and should be decomposed in different video. Thus, in order to compare and analyze the application of human factors, video should be played at the same time for comparative analysis [6]. Such research methods can’t be synchronized with the requirements of the analysis of the video, and can’t carry out statistics for the true information of the game. In this paper, the athlete’s technical and tactical action information should be collected in real time, and the method of manual input and automatic information collection should be used to describe the minor movements of athletes [7]. The auto complete intelligent prompt algorithm can improve the speed of data information acquisition, restore the details of the tactical analysis of the game, and assist athletes and coaches to analyze tactics.
Key technology analysis
Collection method of tactical information in sports competition
The object of this paper is the live video, and the basic information is the basic tactical action of the athletes. Generally, in table tennis matches, the way players clap and play, as well as the reaction and effect after hitting the ball will be concerned. Athletes use the pre competition training design techniques, such as the “2+3” joint defense tactics in basketball matches, table tennis gun shooting and shooting tactics, etc. The collection of basic information is of positive significance to the analysis of the athletes’ movement and the training and guidance of the coaches [8]. The semantic information of sports games in this study is relatively complex, and therefore, the existing technology can’t extract information from sports information automatically. However, the accuracy of the description of semantic information will determine the feasibility of future data statistics. In order to accurately define and express the semantic information in the process of competition, in this paper, taking table tennis match as an example, the tactical meaning of table tennis matches is described, and the basic structure model of information collection is established. And through the description of the language script of the actual table tennis game, the semantic event information of sports events is collected from the perspective of the auxiliary language tool [9]. In the actual competition of table tennis, after classifying the technical movements, the tactical information is classified according to the classification method of basic line, and then the coding of tactical action and classification is put forward. In the information collection model, the tactics of table tennis are divided into batting style, completing basic action, shooting ball effect, and moving track of ball, etc. And the table tennis tactical information is divided into service, the ball attack, attack tactics stalemate class, sub class technology.
The tactical action of table tennis refers to the tactics to defeat the opponent. The basic difference between tactics and strategy: tactics is the analytical strategy to study the guiding significance obtained from past competitions, while strategy is the law that guides the overall situation of the game [10]. Generally, the tactical core of table tennis is to gain the overall advantage. However, table tennis technology is the basic method of game action, which belongs to technical problems. Only reasonable use of tactics can the technical characteristics of athletes be fully played out. The regional division of the table tennis table is shown in Fig. 1. Table tennis professional division method is: in the vertical direction of the net, the regional division is carried out, in which, the area inside the table is “4, 5, and 6 zones”, and the area outside the table is “1, 2, and 3 zones”. In this paper, the definition of the table outside the region includes the introduction of table tennis, “out of bounds” and “down net” are mistakes. “7, 8, 9 zones” are out of bounds area, and “0 zone” is for off net.
Table tennis table division area.
The tactical offensive routes of table tennis are classified, such as the “001” serial number route name is “1 districts ∼6 zone”, and “002” route name is “1 districts ∼7 zone”. After describing the technical movements of table tennis, it is necessary to encode the basic movements of the table tennis and the effects of the racket [11]. In the basic action alias, the corresponding action code is given. The Chinese phonetic alphabet is the encoded symbol, and the encoding rules are defined by Chinese Pinyin, but are not case sensitive. Process coding is a record of the action process of an athlete, which records and describes every completed action.
For example, when an athlete actually plays 10 rebounds, there are 10 coded words that should be recorded from the point of view of recording. This method can describe and learn intuitively, and at the same time, exposes the problems that each movement needs to pay attention to. Therefore, the test of the operator is relatively strong, and the pressure of field data acquisition is relatively large. Video recording after the game is convenient to collect data, which can pause when can’t keep up with the data collection [12]. For example, the script language left by a player who performs a technical event during a match is as follows: ZX16. FB66. FT62. ZH23. ZH33. ZH33. ZH31. ZH33. ZH31. In this way, scripting language can record the basic action of the game, and for the scripting language of the home game, “.” is added at the end of the script. The process code is used to collect the technical and tactical information of the game, and finally, a tournament semantic event field that covers the basic information of matches, balls and scripts is obtained. ZX15. ZC52. ZC25. BB51. TG11 The CS corresponds to the action of the home team and the visiting team as well as other tactical information, and the semantic information of table tennis depends on the external definition, which is determined by the professional knowledge of table tennis. In this way, the semantic information of table tennis matches can be described comprehensively and accurately [13].
For AVI video file calls, image quality and compression standards can be randomly selected, and it is also a widely formatted format for video files. And the general video capture device is also aimed at the AVI file format. In this paper, AVI video format and RIFF file format are converted, and the AVI file format is analyzed. Finally, detailed parsing is done at the file format byte level, so as to prepare for the final embedding of AVI document parsing in the video [14]. AVI video file format is a cross encoding RIFF file of audio and video data, and the more frequent ending suffix for the video format name is “avi” or “divx”. In the windows operating system, AVI is a relatively common format for video applications, and is also a relatively suitable file format for video storage in video capture devices. The standard AVI format file should contain the audio and video streams.
The commonly used AVI format file is divided into “AVI1.0” and “OpenDML” format, and “Open DML” file format usually includes two data streams: video stream and audio stream. However, only one video stream and audio stream are relatively legitimate at work. In the “AVI1.0” format file, there is a DV data stream, in which the file exists in a single stream, and the main advantage is that the storage space is relatively small [15]. The “AVI1.0” file supports the “DirectShow” filter. Although the “OpenDML” format is more commonly used, the “AVI1.0” file is easy to accept the latest technical support in the software market, and thus may become more popular in the future. But the two format files are based on the RIFF format file [16]. In the actual use of the process, there is a slight difference in the format of the data, which lies in the two modules of the document, namely “Strh” and “STRF” file type. The method to convert AVI format is shown in Fig. 2.
Standard AVI file structure diagram.
AVI RIFF file format, which specifies the combination of documents in a certain situation, is the similarity feature of various data type files on the arrangement. For the data in the file, there is no constraint on the encoding format, for example, “MPEG-1, Dvix” can completely become a basic document encoding a data format. And the requirements for data are: data can be without a certain amount of compression, and AVI data is considered as a storage container for data, and the knowledge of the format of the file itself adds data to easy patterns. The format of the file is: data is added to the easy mode, while for data, there are no basic constraints on coding and formatting [17]. In this study, a detailed analysis of data structure is carried out for a random AVI file, and then, the shunt and mixed flow algorithms can use the given data structure diagram. Figure 3 is a basic view of the 16 hexadecimal of an AVI file.
The basic view of the 16 Decimal System of AVI files.
The whole file is the later RIFF data block, with 4 character data block as a marked character segment. The actual size of the data block represents the real situation of the data, and the data block can also be a sub block. “0000–0003” byte number represents “RIFF block mark”, and the value is “52494646”, and the “0004–0007” byte number represents the entire file size of 43096 bytes, and the value is “58 A8 0000” [18]. And the byte number “0008-000A” stands for the “AVI” flag, and the value is “415649”. Then, the “6864726C” byte number represents the random “hdrl” block, and the information block is “LIST” block and will contain multiple “avih” and “Strl” modules in fact. The block in Hdrl module is avih block, the code is “61766968”, recording the whole information in avi video file, such as the number of streams in the information, the video image and the wide and high data. The basic data structure of “MainAVIHeader” is used to record all the information of avi files [19].
The data is a LIST block that represents list describing information, and then explains the basic information of the file, and the file will be different data streams corresponding to different LIST block. There is a “strl” file list in the block store of “strh” and “strd”, which is used to save the configuration information in the decoder. And the name of the saved stream should be guaranteed to be in the optional state. Through the above data analysis process, a complete avi format file can be obtained. At the same time, the file has a certain combination of data, which needs to compose the basic data of audio and video data [20].
Design of sports video information collection system is to provide sufficient preparation for video semantic analysis in the process of sports competition. The data acquisition system of the system mainly completes the extraction of the scene data, establishes the basic relationship between the script data and the video production time, and establishes the embedded function of the script information. In order to realize the function of information collection and video annotation, the footstep data are analyzed preliminarily, and the editing function after video collection is added. Figure 4 is the port diagram of the data acquisition.
Port diagram of data acquisition.
According to the design requirements of the system, the system structure is divided into data acquisition and recovery module, game information data collection module, data script design and editing module, information management module. And the collection module of field data is analyzed emphatically. By using the basic functions of video capture, the specific view of the game is converted to the computer display, and by collecting data intelligently, the data can also be collected according to the video replay of the match. According to the characteristics of table tennis match, it is necessary to collect the language information of tactics. In script editing module, it is necessary to build an editing platform that embeds video basic data. The script time is adjusted to the basic function module, and the key frame is used as the benchmark. According to the Windows platform, the common AVI format information embedded module is established, and the format expansion function of the system interface is reserved. And Fig. 5 is the architecture of the system.
The architecture of the system designed in this paper.
In order to acquire the basic tactics data, according to the analysis of sports video semantic information system, as well as the technology research of each system acquisition module, it is hoped that the script data and video capture can be synchronized to describe the synchronous state analysis of language and video, so as to avoid the problem that the data of the script is incomplete when the game is too fast. And additional video capture method can also be carried out after the game. Figure 6 is the structural diagram of the video tactical data acquisition process.
Video tactical data acquisition process structure diagram.
Video input and output devices include cameras and camcorders. The filter formats the video files transmitted by the camera and displays them through the video card. Switching the basic video angle through the visual angle will greatly reduce the total fatigue feeling of the operator in the view collection. At the same time, the image collected on the scene is output into the storage device, and the useful athletes’ competition data in the process of competition is stored. Finally, the information is input into the database, and the coaches can analyze the tactics execution and application in the actual matches according to the competition situation. According to the application status of equipment management object and video information acquisition equipment object, the work time is divided into three parts: video playback, video preview, and video capture. After the initialization of the working mode, the audio equipment is packaged and processed. And the sequence diagram of the corresponding video capture and return visit is shown in Fig. 7.
Sequence diagram of video capture and return visit.
In the process of actual competition, auto complete intelligent prompt module should carry out analysis according to the skill database of athletes, and combine with the information provided by basic skills information base, and then reenter the script information into the database. After collecting the tactical information of different matches, the editing function of the skill database needs to extract and analyze the script data of each match, so as to carry out the automatically complete and intelligent service. In this study, the artificial video input and the auto complete intelligent prompt algorithm are used to collect data information, thus to ensure the integrity of athletes tactical information, as much as possible to improve the speed of information acquisition. And the intelligent completion prompting function of information collection depends entirely on the support of the basic information base and the skill pool of the team members. Figure 8 is the structure diagram of the data acquisition module.
Structure diagram of data acquisition module.
In the process of gathering and analyzing tactical information of sports events, it is necessary to analyze the differences of athletes’ skill performance. When inputting the cue, the information database of the athlete’s skill is searched. If the “skill base” does not exist, it does not achieve the corresponding query results, and then query athletes’ skills database will give specific tips. In order to collect the information of table tennis players, it is necessary to describe the script of table tennis single cricket. The athlete’s skill information is firstly extracted, and the basic collection source is the basic athlete information database. At the same time, the tactics combination and skill style are adopted to complete the incomplete functional information description in the information base. Figure 9 is a flow chart of the algorithm for automatic completion and prompt function. In order to improve the sports tactical information acquisition speed, an intelligent prompt automatic completion algorithm is designed, and the purpose of this study can be achieved by cluster analysis. Data is divided into categories of multiple objects, so that many objects of the same class have a high degree of similarity, while different classes have low similarity.
The flow chart of the algorithm for automatic completion and prompt function.
In the study, the number of given objects is N, and the combination is D. The similarity distance is defined, in which the similarity is D × D → R, and the positive integer is K. The process of clustering analysis is as follows: the data set is divided into the unconnected part (C
li
, Cl2, …, C
lK
) of integer number, and the calculation expression is obtained:
In the formula, repi is the representative point of C li .
K - Means Clustering analysis algorithm is a commonly used clustering algorithm, and the number of set classification is C. For the sample set by cluster analysis, clustering center can perform clustering results expression. The objective function of clustering analysis is set, and the iterative updating algorithm is used to update the direction of the objective function. The result of cluster analysis makes the objective function obtain minimum value, and achieve the optimal cluster analysis result. Assuming that the sample set of clustering is X ={ x
i
|x
i
∈ R
p
, i = 1, 2, …, N }, a clustering center with a number of C is obtained: z1, z2, …, z
c
. If w
j
(j = 1, 2, …, c) represents clustering classes, then the calculated expressions are converted to:
The objective function is assumed to be:
In the formula, the number of samples obtained by induction can be used to measure and calculate the distance between sample data. According to the Euclidean distance (d
ij
) formula, the objective function defined in this paper is the square sum of the data points of the individual sample and the corresponding cluster center, and the calculated result is the minimum mean square error. The expression of Euclidean distance is as follows:
In this paper, the video file parsing and information embedding algorithm were tested. Aiming at the method of document information annotation in AVI video file, the parsing and mixed flow operation process of video capture file was adopted. In the research, the AVI video test file with different file memory size was selected, and more than 200 video files were selected. The comparative analysis results of AVI file parsing algorithm and mixed flow algorithm in terms of time consuming were obtained. Figure 10 is the performance result of mixed flow time algorithm. In Fig. 10, the abscissa is the size change of the AVI video file, and the ordinate is the change of the mixed flow time. And it can also be seen that with the increase of memory ratio of AVI video file, the mixed flow time gradually increases, and the size of video file has linear relationship with the mixed flow time of video file. And when the size of video files is 8000MB, the mixed flow time can reach 43000 ms.
Algorithm performance results of mixed flow time.
Figure 11 shows the relation graph between the analytical algorithm and the analysis time. As can be seen, the size of the video file has a linear relationship with the time of the parsing algorithm. However, when the video file is less than 400 MB, the parsing time firstly decreases and then increases, and the video file size is less than 400 MB, and the analysis time is not obvious. And when the video file is larger than 4010 MB, the analysis time is gradually increasing, and the parsing time of the video needs at least 710 ms.
Relation graph between analytic algorithm and analysis time.
When analyzing AVI format video files implanted directly into the system, after file fusion in the script information data of sports tactics, by using this research system, sixteen binary file analysis was realized, and the module can be viewed based on the structure of AVI file. From the result, the data position of embedded script can be obviously observed, thus to analyze the correlation between script data and file. Figure 12 is the AVI structure chart embedded with annotation information.
AVI structure chart embedded with annotation information.
As can be seen from Fig. 12, according to AVI documents, basic requirements for internal data distribution design of embedded service script contains data block strl′, which is set to txts′ in the standard field facctype. And the data segment ′movi′ of the script data written corresponds to the corresponding format file. In the test process, the tactical information described was about a table tennis game with memory of 420MB, and the format of the file was AVI. Among about 200 records, the embedded information video file increases by 2145kb. However, the increase of file ratio is proportional to the number of embedded scripts, but has no relationship with the memory ratio of the file itself. The player does not interfere with the description of the video file, nor cause damage to the video structure. The test system can read the basic information of video annotation within 3 seconds, and has been successfully applied to the later video retrieval. The video in the main interface is the result of the combination of script data and video source file in a table tennis match, the associated video is played in the playback software, and the script data embedded in the video can be displayed on the other side of the system.
In this paper, the construction of the information acquisition system of table tennis game tactics was taken as an example, and information collection and video annotation technology were discussed in detail. In terms of technical level, the method of auxiliary mark symbol was used to describe the script language, and the tactical characteristics and the coding characteristics of technical action were established. At the same time, the intelligent prompt automatic completion algorithm was established, so as to increase the speed of data input. Then the basic rule data acquisition method was improved reasonably, and the data acquisition dynamic process was combined with video. And in order to improve the performance of data processing, an ACARMI algorithm was designed to test and analyze the collected video information, and the effectiveness of video file acquisition and processing method based on semantic description was verified. In order to improve the sports tactical information acquisition speed, the intelligent prompt automatic completion algorithm was designed. According to the basic information and skill information of athletes, this algorithm can be used to calculate the frame structure of the athletes’ knowledge data source. And combined with data mining calculation method, it extracts new data that can be used for mining computation in stored script data.