A MULTIMEDIA APPLICATION FOR TACTIC ANALYSIS OF BASKETBALL GAMES

A. TSIMPIRIS - Computer School, Serres, Greece, . TSAMOURTZIS - Demokritus University of Thrace, Department of Physical Education & Sport Science, Komotini, Greece,
N. SFINGOS, G. ZAGGELIDIS , S. ZAGGELIDIS - Aristotelian University of Thessaloniki, Department of Physical Education & Sport Science, Thessaloniki, Greece,

Key words: SportScout, video analysis, basketball, video information system, multimedia application

Introduction

Basketball is a fast flowing game of phases with a lot of statistical factors. It is essential for the coach to be familiar with the game and have the correct percentages. A coach that analyzes the game and is familiar with its percentages has the possibility of decreasing the probability of error for his team and of finding himself in an advantageous position against his opponents. Without concrete statistical data he is condemned to support himself with personal judgement and intuition. (Mikes, 1987).

The use of the PC constitutes an important aid in the direction of more valid and correct information, recording and analysis of games. PC programs are particularly widespread because of the important outcomes that result from their use, when given the help of a video the possibility of an analysis of any game and the output of conclusions so much on the attribution of the team and players as well as on the follow-up of any rival team.

The aim of this project was the development of a multimedia application that could analyze (tactic) all the Olympic sports and especially the sport of basketball via video analysis with the help of a PC.

Video information systems are informative systems that manage the entry signal of the video, process the video, filter the information, store, and index the video, in order to provide a complete collection of video data, for a large number of users in an easy and flexible way. Moreover, a video database (Row, Boreczky and Eads, 1995) is the core of a video information system, which manages the video, including input, update, and the deletion of information relative to the content of the video as well as the elaboration and execution of queries.

Such systems of information video have been designed for multiple usage aims (Chang, Lin and Lee 1995, Chang and Lee 1995, Chang and Lee 1996) as well as specialisations of such systems that are focused on the analysis of movement in sport. (Wu, Narasimhalu, Mehtre, Lam and Gao, 1995, Chang, Lee, 1997). We can say that the application that we have designed belongs to the category of Video information systems.

There are applications that report on tracking of multiple objects using colour histogram back projection and motion cues, and a video scene detection based on a colour histogram (Agbinya & Rees, 1999). One of the possibilities of our program SportScout is the use of different colours for the representation of information (ball, players, defence, attack etc.) on the court In mobile multimedia systems, the mobility of users, devices, and information produces new challenges for processing globally distributed information. Flach & Guenther (2000) focused their attention on new strategies for the adaptation and integration of database techniques regarding the management of multimedia data types at the particular conditions of mobile infrastructures.

Their major aim was to optimize the mobile data exchange by using content based information retrieval and reduction methods as well as media specific queries within the database system in order to reduce the amount of data to be transferred. The application that we have developed is a portable multimedia system that can be used by the users of a portable PC system independently or shared with the possibility of assembling the shared information of the portable systems in one central base in the end. The integrated management of standard structured data and time-dependent data is still too complex to be covered by a single system. So there will be distinct systems, such as multimedia DBMS and media servers for, e.g., video data, each doing only what it can do best. However, some functions desired by users and applications necessarily involve both systems. A prime candidate for this is the use of hypermedia link anchors in continuous media. Lindner, Berthold, Binkowski, Heuer & -Wegener (2000) had an idea behind this approach.

The idea was to enhance the client software to improve the interplay between structured data in the database (hypermedia structures and metadata) and the video stream. On initiating the play out of the video, the client receives a list of anchors from the MMDBS and organizes them in local structures. It also receives a handle for the video to be played out and uses it to contact the media server. Query and update processing will be distributed over the different kinds of servers.

In video database systems, one of the most important methods for discriminating the videos is by using the objects and the perception of spatial and temporal relations that exist between objects in the desired videos. Pissinou, Radev, Makki & Campbell (2001) developed a model that describes the spatial relationships among objects for each frame in a given video scene and the temporal relationships (for this frame) of the temporal intervals measuring the duration of these spatial relationships.

Their model representation also provides an effective and expressive way for the complete and precise specification of distances among objects in digital video. This model is a basis for the annotation of raw video. In the system SportScout that we have developed, we can represent respectively territorial information in proportion with the time of sequence video, for example the course of a ball or a player on the court.

Contrary to other approaches (Lee, Chiu and Yu (2002)) video database systems that aim to visualize, retrieve and manipulate video objects automatically, the application that we have developed records the information manually according to the philosophy of each user coach.

Software applications using video content in sports

There are software applications that help athletes and coaches to estimate distances, speed etc through motion analysis (Chang, Lee, 1997), or robotics simulation (Rosa, Vehia, Puyolb, 1997). The application that we have developed focuses only on the video analysis of tactic and statistics. One of the problems that appear in video analysis is the storage of different information that concerns the appropriate video clips as well as the difficulty in their indexing and grouping. Video analysis systems store these video clips in the hard disc of a PC with an attached comment. The result of this action leads is the complex grouping and search of required clips as well as the overloading of hard disks caused by multiple repeated video clips. Our system uses the entire video from one or more games and records the facts based on the time they commenced and terminated. We save in this way the creation of many unnecessary video clips in each search and the corresponding overloading of hard disks.

There are many other applications, which help athletes and coaches to process the videos of the games and to provide useful information. We point out the following applications Soccer performance, Sports Tec, Match Analyzer, LRS Sports, Pinnacle Systems Team Sports, SIMI 3D Motion of Analysis etc. which make corresponding analyses as SportScout.

SportScout's basic characteristics

The combination of video and statistics in a system is a flexible and effective way in order for coaches to be prepared and to study their competitor sufficiently. In this way, they will be ready to face him. Sporscout provides statistics, reports, printings as well as direct access to the desired videos.

The basic advantage of this software is the storage of the entire video in the hard disc and the naming of video sequences with the help of a user-friendly open form consisting of grouped information. Moreover, the installation and transport of the application in laptops and the management of videos in these personal computers turn Sportscout into a unique solution for individuals that do not need complicated video databases with servers and are not demanding as far as hardware is concerned. There is however, the possibility of the collection of shared information from other laptops to a central database. The open analysis form through which the naming of the video content is made, gives the ability to adapt the application to any sport. With a simple interface the user sets what he wishes to include in this form, once in the beginning.

General description of application's capabilities

SportScout has been designed to be interactive to users and data in an easy and fast way. This system initially functions as an intelligent scenes' recorder of one or more video clips which can and depict similar montage scenes in any sequence it is asked as well as statistic elements. The data is stored in a base of information and is connected to a mass of video files. 

The basic characteristics of this software generally are:

Extensive and completed digital video analysis of team games or individual sports, in very small time intervals, using the choices that are provided by the program (video, statistics, graphic representations), which can take a lot of parameters.

Creation of form of analysis specialised for each sport, based on its particular characteristics. Each form of analysis can have up to 10 main categories, 50 subclasses and in total 3.276.800 different criteria.

Classification and record keeping of phases that are fixed based on the time, and mechanism of search of concrete phases with criteria of search completely parametrized.

Automatic collating sorting and creation of video clips at the requirement of the user, with use of criteria of search and classification.

Creation of files with video with the possibility of management through a database and most rapid access in concrete scenes.

A' phase

Basic condition is that video exists in digital form (MPEG -, MPEG - II, AVI). Then it is loaded or is created from the beginning in the program, a file which will contain the key words that will be used at the characterization of the desirable video clip. There is also a file of picture (bmp, jpg) charged with the graphic depiction of each space of action, in the particular case of pictures from competing places. The content of video is characterized based on time. That is to say that the time moment that a sequence video begins is stored as well as the time moment that this sequence finishes.

After some certain initial regulations of the program which have been done (p.h speeds of slow/fast movement of the video etc.) are selected with the help of the mouse the desirable word keys that will characterize a sequence of video scenes and is stored each time a combination of key words, in a base of data.

B' phase

After the desirable key words are stored that will characterize the scenes from one or more video files, then the user can seek and project concrete grouped scenes that will fulfil the same criteria. The choice of these scenes is made with simple choice of key words from the cards in which they had initially been registered. The resulting grouped scenes can appear in various ways: a) with repeated video clips, v) with datasheet c) with graphic representations and d) with linear depiction in a selected background.

Figure 1 - Process of characterization of scenes and search of data

In Figure 1, you see the flow chart that describes the process that should be followed so that a characterization of scenes is made from various videos or a search of data from a base of information that is applied to a crowd of video files.

Methodology

Technical characteristics

SportScout, is a 32- bit object - oriented application written in VB 6.0 running in Pentium PC's with operating systems Windows 95/98/ME / NT / 2000/XP.

This system needs video files in mpeg -1, mpeg-2, or avi format. Video files can be captured directly from a digital video camera or be from ADC digitalisation cards and be stored in the disk of the PC. With the current technology a one hour video is stored in space of an about 1 GB in quality mpeg -1. The large capacities of hard disks of PCs even of portable ones allow the storage of plenty of video hours in digital form. In figure 2 we present the typical equipment that is needed for the operation of the program SportScout.

The menu of the program is made so as to be able to be adapted into any language and to have publications for all the countries of the world.

The change of the language becomes easy, recording in a text file all the messages, the choices of the menu and submenu in the corresponding words of each language.

Video

Each video is presented (figure 3) in a handler picture box of VB which undertakes the role of a line of tools where the video control handlers are placed as for example the button play, stop etc.

A handler Active Movie with a name presents the video and with the handler MMControl the planning of guidance keys of the video is done.

The more basic element during the storage of an incident of the video is the opening moment of the incident and the expiration moment of the incident. In three handlers of textbox type are presented the time of beginning of registration, the time of termination as well as the running time of the video.

In this system and the time of reaction of the observer has been foreseen. The handler usually clicks the beginning and the precise running time of the video is stored. However, in case the operator has chosen that he wants some reaction time during the storage, then from the running time of the video this time of reaction will be removed so the time of beginning of the incident that will be stored will begin from the moment that it was selected minus the seconds that the user selected as the time of reaction. In this way the user saves time from the likely setbacks that he would make if this choice did not exist.

It has been observed that for a video analysis of a sport with two-hour duration in the conventional way, with use of two video appliances, about 2 days are needed. While with the use of this system the time has been decreased to the time of the duration of the game which is 2 hours. Of course this time depends also on the depth of analysis.

The more elements somebody wants to include in his analysis the more time of analysis will be requested. Naturally the opposite is also valid.

Figure 2 - Equipment that is required to work with SportScout application

Menu

Figure 3 - The main form contains: digital video, key words that have registered in the form of analysis and the graphic representation of the game

Form of Analysis

We call this form of analysis the space where the user can define the key words that will be used during the characterization of the video clip scenes.

How can the registration of choices of the form of analysis that is open include up to about 18.000 different criteria? This space is constituted of 10 handlers SSTab which we could make 99 using the maximum number that the VB allocates. In each SSTab there 5 ListBoxes were placed. Each ListBox can store up to theoretically indefinite number of data. In the particular case this number is limited to 216 = 65.536 different data. This restriction emanates from the variable maxlist that presents the value listcount of each lisbox that gives the capability of presenting the number of registrations of the list and is an integer type number and the VB gives to these numbers this maximum price (Global maxlist As Integer.)

Therefore we speak for the capability the user has to define 10 x 5 x 65. 536 = 3.276.800 different key words from which the 50 there could be used simultaneously during the characterization of a scene (that is say 10 SSTab cards myltiplied by 5 ListBoxes). We believe that this number even though it is arbitrary is big enough to describe a scene.

Storage

While the user begins the analysis he selects various combinations from key words. The combinations of these keys that are selected are stored in an Excel accountant sheet type of file with the help of the handler sheet. In this file as well as in the same registration for each combination of key words, the name of the video file is also stored usually of mpg type, the time moment that it begins and also the time moment that the characterized scene finishes. The maximum number of registrations that are allowed to be stored in this file by the function sheet of VB is 2¹⁴ = 16.384 scenes.

Search

The elements that have been stored in an Excel accountant sheet type of file with the help of the handler sheet are loaded in the handler table feedback so that the infiltration the regrouping and the treatment of elements that was asked can be done immediately. In the first 5 lines of the table that appears in this way the criteria are presented that were used as well as the stable elements of the project. The number of columns of this table depends on the number of lay ups and listboxes that were used during the previous recording of key words. The combination of key words that will be selected corresponds to a reasonable section of totals and a restriction of lines of the table. In this way the criterion of search is created with which time moments of the entire video will be located and shown. These time moments correspond to the lines of the table that fulfill the above conditions. As it can be understood the alternation in this way from a video in another one becomes very easy. When the values of the cells of the table that correspond to the column that includes the names of the video files are different from the video that corresponds to particular criterion is automatically loaded in the memory of PC. The result is that a lot of big in size video files are used, and the attribution of search becomes bigger depending on the number of video files that were characterized and used.

Statistically

The capability of presentation of searches based on the key words, in graphic representations (figure 4) is one of the strong points of this application. This is because of the convenience in receiving statistical and centralized results and direct presentation in a chart form. The realisation of the above objective was realised with the reception of information under the form of numbers. What happened was firstly the scanning of lines of the file that contains all the information that concerns the key words so that the number of registrations is measured.

Then with simple function sums, medium prices and percentages are easily calculated. The attribution of these statistical elements under that form of diagrams is done with the help of the functions of VB for various types of graphs. In the particular application we incorporated three different types of diagrams so that there is attribution of information in three different ways and the user can decide which is most suitable. The types concern charts of two dimensions (diagramm.chartType= VtChChartType2dBar), charts of three dimensions (diagramm.chartType = VtChChartType3dBar) and pies (diagramm.chartType = VtChChartType3dPie).

Figure 4 - The statistical part of SportScout offers basic statistics, which provide a fast picture round the elements of the game

based on the files of analysis

Linear depiction of places in the sportsground

The need of imprinting information on paper which depicts the sports ground is met in all the sports. It is observed that coaches, mark on papers or in big tables with pens the places of the players in the sportsground or they paint the course of the ball or the system that they are to use (Brown, 1983, Cousy and Power, 1970, Wooden, 1966). Taking into account the above we see that there are enough applications are developed in the PCs that can replace the paper or the table and the pens. These applications can depict the information statically in some backgrounds, or with animation movement. In this program we incorporated such a technique of static representation of information (figure 5).

Figure 5 -The graphic representation of the sportsground in order to fix the place of players or the course of the ball eg of the shots of the ball

in the sportsground

This is done as follows. Initially it opens with the beginning of the program and a picture of bmp or jpg format, which usually depicts the sportsground. On this background the user can place dots at any point of the picture, which are linked linearly. Each line has a different colour depending on which card or key words are selected. Each of these cards corresponds to a different activity. E.g. a card for defence, one for attack etc. Therefore from the colour of the line the user can also optically understand actions that took place. This information is stored with the time of beginning and end of the video clip that is presented and the key words that were selected. When in the end some certain key words are searched for, the static depictions with the lines, the dots and the numbers are also automatically represented. It can also be done with mass depiction of dots and lines on the picture that depicts the sportsground, so that statistical conclusions come out on the region where various events take place.

Practical Application in Basketball

SportScout can be used for various aims but particular emphasis was given to sports and particularly in basketball where there is great interest from coaches, journalists, channels, athletes, sports fans.

The most general objectives of this application were firstly the determination and classification of different elements and the growth of a method of classification according to concrete criteria of a coach. The improvement of training methods takes into consideration dialogic and dynamic exchange of information for the sake of the increase of training effectiveness, for the better and shorter evaluation of the positive and negative elements of teams and players separately, for the search and choice of talents during the game period that will staff the team, and finally for the improvement of espionage of the next opponent.

• Some of the possibilities of the program SportScout for the video analysis of games in Basketball are:
• · The programs of this type allow direct access in statistic information and possibility of creation of an espionage sheet.
• · It provides us with information from one or from a series of selected games.
• · The evaluation is multiple.
• · We can write comments on each possession.
• · As soon as the information is input in the computer, a database is created which we can recall at any moment for each likely combination.
• · The elements of search can be processed automatically and create video clips with each possible script that interests us.
• · It allows us direct access in statistical information and the possibility of creating espionage sheet.
• · We can see and print all diagrams of all the shots from each situation of the game that interests us. We can filter the diagrams with players, teams or with anything else.
• · It provides us with detailed reports of the tendency of the game for any situation we may want to analyze in full or half court e.g. to analyze how to break the press throughout the court when our possession has led to a successful shot. We can also print out these reports.
• · Observing in the video all the possessions that shape the reports of tendency of the game we can see the possibilities or weaknesses of our team as well as our opponents (e.g. recognize which defensive planning allows us to achieve more points).
• · We can create videocassettes with team, individual games in any sequence we want to watch them (Tsamourtzis and Siskos, 2001).
• · It combines video with roster sheets and statistics in a system.
• · The form of analysis is open and it can be altered and be supplemented by the coach according to his judgment and based on the needs of research. A typical form of analysis according to which the scenes of one or more games can be characterized and be sought is shown in figure 6.

Using the above analytic form we analyzed with SportScout 88 games of Basketball that were carried out by teams of men, in the various Championships that are organised by FIBA. A total of 15776 phases with 14093 possessions, 9558 shots, 4226 free shots, 3417 foul, 2235 turnovers, 5604 rebounds (3771 defensive and 1833 offensive) and 1096 steals. The total offences of teams begins after defensive rebounds at a percentage of 24%, after offensive rebounds at a percentage of 11%, after steals at a percentage of 7%, after baskets were at a the percentage of 36% and after out of bounds at a percentage of 22%. Fastbreaks were at a percentage of 14.3%, with secondary breaks at a percentage of 5.7% and with set plays (organised offences) at a percentage of 80%.

The total of offences of teams leads to shots at a percentage of 61%, to loss of the ball at a percentage of 14%, gained fouls at a percentage of 21.5% and to temporary out or renewals of time at a percentage of 3.5%.

Shots in the total of offences of teams are free throws at a percentage of 31%, two points shot at a percentage of 48% and three points shot in percentage 21%.

The types of shots in the total of offences of teams are jump shots at a percentage of 72.8% lay ups at a percentage of 15.6%, follow at a percentage of 2.2%, dunking at a percentage of 3.5% and hook shot at a percentage of 5.8%.

The total types of fouls which are taken again by the teams out of bounds at a percentage of 38.6%, changed to free throws at a percentage of 50.7%, are basket and foul at a percentage of 9.3% and only 1.4% concerns the remainder fouls.

The total of types of turnovers of the teams is made up of bad passes at a percentage of 33.8%, by steps at a percentage of 11.7%, by bad ball handling at a percentage of 33.8%, offensive fouls at a percentage of 16.6% and only the 4% concerns remainder errors (3, 5, 8, 24, double dribble). Particularly, we observe that time errors (3, 5, 8, 24) do not exceed the 3% in their total.

Finally, as for the tactic of defence, the types of defence on which these games were played, the man to man defence is used at a percentage of 72.4%, the zone defence at a percentage of 7.6%, the combination defence at a percentage of 0.1% and defensive transition at a percentage of 19.9%. The types of these defences are played in the 1/4 of the court at a percentage of 93.8%, in the 2/4 of the court at a percentage of 2.1%, in the ¾ of the court at a percentage of 1.3%, and in the 4/4 of the court at a percentage of 2.8%,

Conclusions

The applications of management of video are very useful in all the sectors. Specifically, SportScout is drawn so as to be adapted easily in all the sectors.

Flexible application with open form of analysis where the simultaneous characterization of scenes is enough to satisfy the most difficult requirements in a lot of activities.

Portable video base of database useful for users that move from place to place, coaches, journalists etc.

It is a multimedia application that combines video, sound, text, picture in a simple and fast way of interaction of user and choices. It incorporates all the advantages of a digital video which results in the direct search of phases and desirable video clips and their organisation in a compatible way for a simple pc user. Moreover, it provides the possibility of searching multiple phases from many Videos and even different stocking means.

The form of analysis that was created for basketball is very useful for further use of growth and exploitation from individuals of the athletic field. It can cover vast spectrum of requirements for complex criteria of searches. It is functional with regard to the rapid registration of characterizations of the sequences video the growth of an adaptable computational system that could be used in different sports and techniques of training. Its practical application in basketball attributed direct results for fast access to video sequences with complex criteria. The positive result that came from the application of this program in the sports and concretely in basketball was firstly the determination and classification of the different elements and the growth of classification method according to concrete criteria of the final user.

The optimisation of training methods of taking into consideration dialogic and dynamic exchange of information.

The increase of training effectiveness the best and shortest evaluation of the positive and negative elements of teams and players separately. The search and choice of talents that will staff the team as well as the improvement of espionage of the next opponent.

Abstract

The aim of this project was the development of a multimedia application that could analyze (tactic) all the Olympic sports and especially the sport of basketball via video analysis with the help of a PC. This system is able to provide: extensive and completed digital video analysis of team games or individual sports, development of a tactic analysis form specialized for every sport, classification and saving of video sequences that are fixed based on time in a data base, search mechanism of specific video sequences, automatic colleting, sorting and presentation of video clips. The results of the application lead to a more rapid presentation of data with video, statistical diagrams and graphic depictions of positions on the court. This data will help all coaches and more specifically basketball coaches to increase the effectiveness of training. The positive and negative factors of teams and players separately can be easily analyzed with this multimedia application by the coaches too.

Basketball coaches can also locate talents that will staff the team and finally can 'spy on' their competitors helped by the useful tools of this application.

Bibliography

1. AGBINYA, J.I.,&REES, D., MultiObject Tracking in Video. CSIRO Telecommunication and Industrial Physics, Image & Signal Processing Group, 1990
2. BROWN, D., The LSU Basketball Organizasion Handbook. New York, Leisure Press, 1983
3. CHANG, C.W. and LEE S.Y., Statistical and topological feature extraction and matching in video sequences. In: National Computer Symposium, Chung-Li, Taiwan, 1995, 693-700
4. CHANG, C.W.& LEE, S.Y., Indexing and approximate matching for content-based time-series data in video database. Processing of the lstlntema Ucnal Conferences Visual Informatics Systems, Melbourne, Australia Victoria Univ. of Technology, Melbourne City, 1996, 567-576
5. CHANG, CH. W., LEE SH.Y., A Video Information System for Sport Motion analysis.. Journal of Visual Languages and Computing, 8, 1997, 265-287
6. CHANG, C.W., LIN K.F. & LEE S.Y., The characteristics of digital video and considerations of designing video databases. In: Proceedings of the 4th International Conference on Informatics and Knowledge Management. Baltimore, MD. The Association for Computing Machinery, New York, 1995, 370-377
7. CHANG, C.W., LEE, S.Y., A Video Information System for Sport Motion analysis. Journal of Visual Languages and Computing, 8, 1997, 265-287
8. COUSY, B., POWER, F., Basketball Concepts and Techniques. Allyn and Bacon, N.J., 1970
9. FLACH, G., GUENTHER, N., Architecture for the interaction and access on multimedia database systems in the context of mobile environments. In: Proc Int Database Eng Appl Symp., 2000, 224-230
10. LEE, A.J.T., CHIU, H.P., YU, P., 3D C-string: A new spatio-temporal knowledge representation for video database systems. In: Pattern Recognition. 35 no. 11, 2002, 2521-2537
11. LINDNER, W., BERTHOLD, X., BINKOWSKI, F., HEUER, A., MEYER-WEGENER, MR., Enabling hypermedia videos in multimedia database systems coupled with realtime media servers. Proc Int Database Eng Appl Symp, 2000, 327-336
12. LRS Sports. Retrieved May 3, 2006 http://www.lrs.com/lrssports/default.htm
13. MIKES, I., Computer breakdown of percentage basketball. In: Scholastic Coach, 11, 1987, 52-55
14. Pinnacle Systems Team Sports.Retrieved May 3, 2006
15. http://www.pinnaclesys.com/PublicSite/uk/Home/
16. PISSINOU, N, RADEV, MAKKI, K, CAMPBELL, W.J, Spatio-temporal composition of video objects: Representation and querying in video database systems. IEEE Trans Knowl Data Eng. 13 no. 6, 2001, 1033-1040
17. ROSA LA, J.LI, OILER, A. VEHIA, PUYOLB, Soccer team based on agent-oriented programming, Robotics and Autonomous Systems, 21, 1997, 167-176
18. ROW, L.A., BORECZKY THE P., AND EADS, C.A., Indexes for user access to large video databases. In: IS&TISPIE Symposium being Electrical Inning Science and Technology, As jose, The International Society for Optical Engineering, The Society for Imaging Science and Technology, Springfield, Virginia, 1-10, 1995
19. SIMI 3D Motion of Analysis
20. Retrieved May 3, 2006 http://www.simi.com
21. Soccer performance organization.. Retrieved May 3, 2006 http://www.soccerperformance.org
22. Sports Tec International. Retrieved May 3, 2006 http://www.sportstecinternational.com/
23. TSAMOURTZIS, E., SISKOS, A., The computer in the service of basketball coach. In: Basketball Coach, 6, 2001, 26-29
24. WOODEN, I., Practical Modern Basketball. USA,The Ronald Press Company. 1966
25. WU, THE MR, NARASIMHALU, A.D., MEHTRE, V. M., LAM, C. P. and GAO Y., The CORE: a content-based retrieval engine for multimedia information systems. Multimedia Systems, 3, 1995, 25-41