Method and apparatus for generating event registration entry. RU patent 2498404.

FIELD: information technology.

SUBSTANCE: method of generating an event registration entry for game events associated with a physical object present in a sporting event. The method includes a step of monitoring, within a sequence of video images, image characteristics corresponding to physical objects in a sporting event. Further, the method involved displaying on a display device a sequence of video images which includes the monitored image characteristics. A first displayed image characteristic is selected from the monitored image characteristics, which corresponds to one of the physical objects in order to denote that physical object as the selected physical object, wherein the first image characteristic is highlighted during selection. When selecting the first image characteristic, a given event list of possible game events is displayed next to the highlighted, selected first image characteristic and overlaid on the displayed sequence of video images.

EFFECT: faster automatic detection of a place where an event has occurred, which enables a user to quickly switch to the corresponding frames.

17 cl, 21 dwg

The technical field to which the invention relates

The present invention relates to a method and apparatus for generating a registration record of the event.

The level of technology

In the field of broadcasting of sports, for example, the report from the football matches, commentators usually visually identify the players on the captured during the match the video or in the course of the game, being in a certain point of observation at the stadium in a way that allows you to collect statistics about each player during the match. In addition, for programs covering the events, information about the position of each player on the playing field and its action on the playing field can be collected by the operator via video recordings made during the match, which uses the appropriate tools for viewing and editing.

However, viewing and editing of the recorded videos are time-consuming and expensive, and prone to errors operator. Although automated systems can be used to assist the operator in tracking each player using recognition technologies performed for recorded during the match footage, but for automated systems can be difficult to register an event, if the player played in this game events, how to kick the ball, blocking other player etc.

In addition, data with annotations relating to the status of the players during the game you can use to recreate the match, using a three-dimensional virtual modelling. However, in order to model looked realistic, data relating to the type-game event should be entered in a device that performs the simulation, so that the simulated players can meet gaming activities in accordance with the real events during the match.

In some systems, such as are described in the publication WO-A-02/071334, uses a lot of cameras to track the participants of the sports event such as a football match. These provisions are generated by the system, can have annotations for the relevant game, made with the help of an operator, using a separate list, which is presented separately from video games.

However, when talking about the many events may require considerable time for the operator to perform a large number of annotations for video, since he must choose a player from the list and visually confirm the video, what event should be associated with this player. In the present invention is an attempt to eliminate or reduce the problems described above.

The essence of the invention

In the first aspect of the proposed method of generating a registration record in-game events associated with the element present at a sports event, the way where:

track within the sequence of videos of the features of the image that correspond to the corresponding elements in the sports event;

choose from a tracked image feature first feature of the image, that corresponds to one of the elements, for the purpose of this element in the as the selected item;

choose game event from the events list, containing possible game events, to associate with the selected item; and

associate the selected game event with the selected item to generate a registration record the events for this item.

In accordingly, by tracking the peculiarities of the image, relevant, for example, players while playing football, and select the features of the image corresponding to the required player, the operator can select the appropriate game event (such as a kick or hit his head) from the list events to associate with this player. In addition, the feature of the image corresponding to the player, can be automatically selected, for example, using a workstation, manufacturing contents, and the corresponding event from the list of event associated with that player. Therefore, the cost and effort associated with markup footage to generate statistics about the match can be reduced. In addition, for example, resulting from the registration record of events can be used as input 3D modeling, containing the data tracking the players and the ball is associated with a real football match, thus improving the 3D modeling of real match; in this model, the player, thus, can be performed with the repetition of facial expressions and actions associated with the gaming event.

The method also includes: displays the sequence of video images of the first features of the image along with a listing of events; and

selection of game events from the events list in accordance with the choice of manual user-game event.

This helps the user in the choice of events and also speeds up processing of compiling the annotations.

The list of events can be displayed, essentially, next to the first characteristic of the image.

This allows the user to concentrate on the area of the screen displays the video image. It also speeds up the compilation of annotations.

The method can also contain a procedure, where: track by analyzing the sequence of video images provisions in each video of the game object associated with an event to generate traffic data, which relate to direction of motion of the object in the game in the video loop;

detect the occurrence of the event depending on the change of direction of a game object by parsing the data movement; and

mark the flag of one or more of the images, which were detected the emergence of in-game event to indicate that these video contain the emergence of in-game event.

Thus, it becomes possible to quickly automatically detect the place where the event occurred, which allows the user to quickly go to the appropriate personnel.

The game event can be , as happened, if the change in direction of the object in the game more than a specified threshold.

The method can contain a procedure, where: detected by the analysis of the sequence of videos of the provisions of the game object regarding the selected item; and

choose game event from the event list in accordance with regulations of the game the object relative to the selected item.

This is in particular useful as an aid in the selection of appropriate events. For example, in the following version of the incarnation, aimed at a football game, maybe the information about that event occurred when the game object (for example, the ball) was at its height approximately level with that of the head, means that the event was probably the event of hitting the ball head.

The method can contain a procedure, where: detect the relative distance between the object of the game and each of the elements to generate the data distances, which are related to the distance between the object of the game and each of the elements;

analyze the data distances to detect which of the elements was closest to the game object, when were detected the occurrence of the event; and

associate this game event with the element which were detected, as was the closest to the game object, when there was a game event. Can be used to help automate the registration of the event. Method may contain a procedure, where: a sequential sequence of video images;

establish a pause in a sequential display of a sequence of video images on the video, which was recognized as the flag, as containing the occurrence of the event; and

display the first property of the image along with a listing of events so that the user can choose the appropriate game event from the event list to associate with the selected item.

This, in turn, can help to accelerate the process of registration of events.

Each item can be associated with a unique identity that uniquely identify the item.

Also provides a computer program containing a computer readable instructions that while they are being loaded into the computer configure this computer to complete method.

In another aspect of the proposed device to generate the registration record in-game events associated with element during a sporting event, a device containing:

- tracking tool designed to track the sequence of video images peculiarities of the image that correspond to the corresponding elements in the sports event;

a means of selecting the features of the image, intended for selection of monitored peculiarities of the image of the first features of the image, that corresponds to one of the elements to refer to this element as the selected item;

picker-game event, designed to select the event from the list events possible game events to associate with the selected item; and

a means of associating intended to associate the selected event from the selected element to generate the registration record of events for this item.

This device also contains the for display within a sequence of video images, the first features of the image along with a listing of events; and

where picker game event during operation selects game event from the event list in accordance with the manual selection a user-game event.

In another one additional aspect of the proposed device to generate the registration record in-game events associated with an item in a sporting event, a device that contains:

device tracking designed to track in the video loop features images that match the corresponding elements of the sports event;

selector image feature designed to select tracked peculiarities of the image, the first features of the image, that corresponds to one of the elements to mark the element as the selected item;

selector-game event, designed to select the event from the event list of possible events, to associate with the selected item; and

the device of Association, intended to associate the selected event from the selected element to generate the registration record of events for this element.

This device also contains a display device, intended for display, within a sequence of video images, the first features of images together with a list of events; and

where selector-game event during operation selects game event from the list events in accordance with the choice of manual user-game event.

The list of events can be displayed, essentially, next to the first characteristic of the image.

Device tracking during operation can monitor, through analysis, the sequence of video images, the situation in each video game object associated with an event to generate traffic data, which relate to the direction of motion of the object in the game within the sequence of video images; and

the device may additionally contain:

detector, intended for detecting the occurrence of the event, depending on changes in the direction of the game object, by means of analysis of traffic data; and

device flag set, designed to set a flag for one or more of the images, which is detected by the appearance of the game the event, which indicates that these video contain the emergence of in-game event.

Possible detection of occurrence of the event, if the change in direction of the object in the game more than a specified threshold.

Device tracking in detect analysis of a sequence of video images, the position of the object in the game regarding the selected item; and

selector-game event may, during choose the game event from the event list in accordance with position of the object relative to the game the selected item.

Device tracking during operation can detect the relative distance between the object of the game and each of the elements to generate the data distances, which are related to the distance between the object of the game and each of the elements; and analyze the data in distance to detect which of the elements was closest to the game object, when it was occurrence of the event; and

the device of Association at the time may associate the game event with an item that was detected as close just to the game object, when there was a game event.

The device may contain:

the display device is intended for displaying the sequence of video images;

in which the device is made with the possibility of:

install a pause in a sequential display sequence of video images on the video, which was marked with a flag, as containing the occurrence of the event; and

ensure that the device displaying the first features of the image along with a listing of events in such a way that the user can choose the corresponding game event from the event list to associate with the selected item.

Each item can be associated with a unique identifying information that uniquely identify this element.

Additional aspect of the proposed graphical user interface designed to produce the registration record of events for the game's events associated with elements of sports events, the event is consistently associated with the element interface, containing:

the features of the image, within one frame of a sequence of frames which correspond to the corresponding elements of the sports event;

selector features of the image, designed to select from the displayed peculiarities of the image of the first features of the image, that corresponds to one of the elements to refer to this element as the selected item;

selector-game event, designed to select the event from the events list, containing possible game events, to associate with the selected item;

the display device is intended for display, within a sequence of video images, the first features image along with a listing of events; and

where selector-game event during operation selects game event from the event list in accordance with the choice of the user manually game event.

Various additional aspects and properties of the present invention defined by the appended claims.

Brief description of drawings

1 shows a diagram of a system of tracking of the object used in a device and method of registration of events in accordance with the option embodiment of the present invention;

figure 2 shows a block diagram of the sequence of operations tracking method of the object used in device and method of registration of events in accordance with the variants of the embodiment of the present invention;

on fig.3 and 3C shows the schema for object tracking used in the device and method of registration of events in accordance with the option embodiment of the present invention;

on fig.4 presents illustration video, filmed on the stage, which is a football match with the tracked players;

on fig.4 presented linear image, which is equivalent to and technically identical fig.4;

on fig.5 presents an illustration of the video that was processed in the system object tracking for a model of the background image, by determining the average value, and the fig.5 shows a model of the background image, taking into account the dispersion;

on fig.5 presented linear image, which is equivalent to and technically identical fig.5, and fig.5D shown linear the image, which is equivalent to and technically identical fig.5;

on fig.6 presents an illustration of the video that was processed in the system of tracing of the object for presentation tracked provisions of players;

on fig.6 presented linear image, which is the equivalent of the technically identical to fig.6;

on fig.7 illustration presents two video images, which were taken by two different cameras, one on each side of the Playground, and an illustration of the virtual representation of a football match, in which the position of the players is monitored in relation to the time;

on fig.7 shown linear image, which is equivalent to and technically identical fig.7;

on Fig.8 presented the video of the football match, which tracked players tagged with labels;

figure 9 shows a three-dimensional representation of a virtual model of a football match, where you can to change the point of view of the match;

10 schematically shows the block diagram of the system to obtain a virtual model of a football match, in which the game presents synthesized elements are available for the device or through the Internet;

figure 11 shows the diagram of the image football ground together with the list of events, which enables you to associate the game event with the player in accordance with the option embodiment of the present invention;

on fig.12 shows a block diagram of the sequence of operations ways to generate the registration record of events in accordance with the variation of the incarnation the present invention;

on fig.13 shows a block diagram of the sequence of operation and method of detection of in-game event in accordance with the option embodiment of the present invention; and

on figure 14 shows a graph of the xy motion vectors used to detect a change of direction object under option embodiment of the present invention.

Detailed description of the invention

Disclosed method and device for generation of the registration record of events. In the following description provides many specific details to ensure a full understanding of options embodiment of the present invention. However, for a person skilled in the art must be obvious that these specific details need not be met in practice of this invention. On the contrary, specific details, well-known specialist in a given field of technology excluded to clearly illustrate the way to implement them.

1 shows a diagram of a system of tracking objects used in the device and method of registration of events, in accordance with the variants of the embodiment of the present invention. In the presented figure 1 is a variant of the incarnation of the tracked objects are football players (not shown) on the football pitch 30. Video (1920 by 1080 pixels) high definition (HD) site 30 removed by means of one or more high definition cameras. Although variations embodiment of the present invention can be used for tracking objects in the video, taken more than one camera, some examples show only one camera. It should be understood that the HD cameras are expensive, so use only one camera can reduce the amount of overhead required for the realization of the system, which uses this technology. However, the use of only one chamber provides only one two-dimensional view of the scene, on which the properties are located. As a result of tracking objects within the scene, presented the video, can be difficult because of the occurrence of obstacles in the event viewer, in which one object closes the other, etc. An example with single camera 20 is shown in figure 1, although, as shown by cameras 22.1, 22.2, if necessary, you can use two cameras, each of which is aimed at different halves of the football field.

In some versions of the incarnation secondary camera 22.3 can be used in combination with the camera 20 and/or cameras 22.1 and 22.2, to detect the position of the object, such as a soccer ball, relative to the football field 30. This will be discussed in more detail below.

Figure 1 camcorder 20 is in a fixed position at the football stadium, and is made to transfer the signals representing the video image captured by the camera 20 workstation 10 processing of content which carries out the processing of images and other operations for tracking the provisions of players on the playing court about the time. The data representing the position of the players regarding time, then register in such a way that can be generated metadata and statistics of the match, such as length of time that a particular player has spent in a specific part of the playing field, what is the distance of each player ran etc. Data representing the position of the players regarding time, generate the data path for each player, which relate to the way, which took place every player on the video. These paths are relatively three-dimensional model of a football field (the plane of the object) to present the information associated with the movement of the players regarding their position on the court, which is not well visible on the (two-dimensional) video. Such data generated paths can then be used to improve the sensations audience when the user footage of a football match is passed through the appropriate environment for viewers, or to help the coach during the trainings of the team. Object tracking, such as players, on the site of 30 will be discussed in more detail below.

In versions of the present invention workstation 10 processing contents uses the Cell processor jointly developed by Sony®, Toshiba® and IBM. The parallel structure of the Cell processor makes it particularly useful to perform processing tasks associated with the intensive treatment, such as image processing, image recognition, and object tracking. However, for a person skilled in the art would be clear that any relevant workstation and processing module can be used for the realization of options embodiment of the present invention.

In accordance with this technology the video that generate using the camcorder 20 high-definition, choreographed so as to remove the view of the entire site so that you can track the players on the court. The whole site removed from a static position of the camera 20, although, as noted above, more than one camera can be used to capture the image of the whole site. In one example, as noted above, you can use two cameras 22.1, 22.2, each of which focuses on a different half-court. In this example, the video generated by each camera can be combined together using workstation 10 processing content, as described in the application for patent №0624410.7 (published as GB 2444566)to form a video with ultra-high resolution. Thus, after the processing of the Association, the output channel of the cluster camera can be regarded as a single image with ultra-high resolution.

The layout of the super resolution has many advantages, including the ability to highlight certain features of the player, without having to use optical zoom and therefore no impact on the overall image of the stadium. In addition, the automated tracking easier, because the background image of the event is static and so can be tracked object with higher resolution screen.

Tracking object, used in the device registration of events and a way in accordance with the variants of the embodiment of the present invention will be described below, with reference to figure 2, 3 and 4.

Figure 2 shows a block diagram of the sequence of operations tracking method of the object. To track an object, create a model of the background image of those parts of the received video that detect remaining essentially static in the specified number of frames. At the first stage S30 video taken by camera 20, which is a football ground, a process to build a model of the background image. Model of the background image building to form a mask of the foreground image, which contributes to the identification and tracing of individual players. Model of the background image form on the stage S30, determining for each pixel in the average value of the pixels and pixel the variance of the values between subsequent frames, in order to construct a model of the background image. Thus, in subsequent frames in the case, when the average value of the pixels varies slightly, these pixels can be identified as the pixels of the background image, to identify the mask of the foreground image.

Such segmentation of the background image/foreground image is to process, which is known in the field of image processing, and this technology uses the algorithm described in the document authors Manzanera and Richefeu, called "A robust and Computationally Efficient Motion Detection Algorithm Based on Σ Nuclei Background Estimation", published in the proceedings of the ICVGIP, 2004. However, the actual technology should not be seen as limited to this known technology, and also known by other technologies to mask generation foreground of the image relative to the model of the background image for use in tracking.

It should be understood that in the case when the field of view camera video covers some part of the audience, preferably, the audience must not be included in the model of the background image, because the people that make up the audience are likely to be in motion. This is undesirable, as it probably increases load when processing for Cell processor when running object tracking, and also because the majority of sports broadcasting stations hardly interested in tracking individuals among the spectators.

In the opened tracking technology object model of the background image is built in the beginning of the game, and it can be obtained even before the players go out on the field. Besides, the model of the background image can periodically be re-calculated during the game to take account of any changes in the lighting conditions, such as shadows, which may change during the game.

At the stage S40 model of the background image is subtracted from the input image coming from the camera, to identify areas of difference. Thus, the model of the background image is subtracted from the image, and the resulting image is used to generate a mask for each player. At the stage of S45 create a threshold regarding the values of the pixel in the version of the image, which was obtained after subtraction of the model of the background image. Model of the background image generate first determine the average value of the pixels on a sequence of video frames. On the average values of each pixel is possible to calculate the variance of each of the pixels on the nomination of the images. The variance of the pixels are then used to determine the threshold, which may vary for each pixel on all pixels of the video. For pixels that correspond to parts of the image with the high value of the variance, such as parts, including spectators, threshold may be set to a high value, while the part of the image that correspond to the Playground will have a low threshold, because color and content of the site remains essentially the same, except those present players. Therefore, a threshold allows to determine whether the elements in the foreground image, and so the mask foreground image can be properly identified. At the stage of S50 use probability to form on the basis of correlation with the average of model forms of a man to highlight the shape of the mask image in the foreground. In addition, particular colours are extracted from the image, to form a colored mask probability, to identify the player, such as color jerseys players. Thus, t-shirt colors each command you can use to differentiate players from each other. With this purpose workstation 10 processing content generates templates colors depending on the famous sets of colors teams from each of the football team. Thus, the required color of t-shirts each team, color t-shirts goalkeepers and color t-shirts judges. However, be aware that you can use other appropriate templates colors and/or processes and pattern matching.

Going back to figure 2, at the stage of S50 workstation 10 processing contents compares each of the pixels of each template colors with the pixels corresponding to the field of t-shirts in the image of the player. Workstation processing contents then generates a probability value, which indicates the degree of the similarity between pixels templates colors and selected pixels, for the formation probability colors based on the distance in the space of color hue-saturation-value (HSV, TDO) for this command and models of colours of the playing area. Besides, use probability forms for the localization of the players, which is based on correlation with the average of model forms of a man. In addition, the probability of movement is based on the distance from the position of the projected using a recursive estimation block, operating on the principle of least squares using the parameters of the original position, velocity, and acceleration.

Thus, stages S30, S40, S45 S50 and perform with regard to the processing of the camera image. After definition masks the foreground image, perform tracking players after the first perform sorting player-tracking by the degree of closeness to the camera at the stage of S55. Thus, players, which were identified as the closest to the camera, handle first, to resolve these players of the capture process. At the stage S60 provisions players update to ensure maximum probability of form, color and movement. At the stage of S70 build mask obstacles, which excludes already known area of the image that will be covered by other located closer players. This ensures that players with fully or partially closed by other players, can only be matched with visible areas of the image. Mask obstacles improves the reliability of the track, as it reduces the impact of the unification of the tracks (of which two tracks track the same player after the events of overlap). A special problem occurs when multiple objectives are similar, since they cannot be (difficult) color-coded. Mask obstacles allows you to assign pixels closer players and exclude them for down players, preventing the comparison of the two tracks with the same set of pixels and thus supporting their separate identity.

Then should the process of tracking each player by highlighting the features presented in the camera image, and display them on a three-dimensional model, as shown in fig.3 and 3B. Thus, for the corresponding provision in the two-dimensional image taken by the camera, appoint the three-dimensional position for the player, which provides the maximum probability of form, color and movement. In brief explanation, the selection and display of the player from a two-dimensional image on a three-dimensional model will be modified in case of detection of events overlap. For assistance when displayed from a two-dimensional image in a three-dimensional model on the stage, S65, tracked players initialize to the peaks of probability values of form and color displayed on the most appropriate choice of players. It should be emphasized that the initialization performed at the stage of S65, is performed only once, typically at the beginning of the capture process. For a good system initialization players should be well separated. After initialization any error in tracking players automatically corrected in accordance with the existing technology that does not require manual intervention.

To trace in a three-dimensional model based on the provisions of the two-dimensional image, perform conversion using the matrix R projection. Tracking requires that the provisions in the two-dimensional image can be mapped to the provisions of the three-dimensional model. Such a transformation is performed using the matrix (P) projection. Point in two-dimensional space corresponds to the line in three-dimensional space:

Point in two-dimensional space corresponds to the lines in three-dimensional space, since the third dimension, which is the distance from the camera, it is not known, and therefore should look, respectively, as a line through the three-dimensional model. The height of the object (players) can be used to determine the distance from the camera. A point in three-dimensional space are obtained by selecting a point along the line, which is located at a fixed height over a known level from the earth (the average height of people). Matrix R projection receive a priori, one time for each camera before the match, using the process of camera calibration, during which the physical characteristics of the Playground, such as angles 31A, 31, 31, 31D of the Playground 30 is used for determination of parameters of the camera that allows you to simplify the display of two-dimensional position identified players on the three-dimensional model. It performed with well-known technology that uses the established methods. That relates physical parameters of the matrix R projection includes the degree of optical zoom camera, the point of focus, the three-dimensional position, and three-dimensional rotation vector (where it is directed).

Tracking algorithm, which runs on the stage S60, is scalable and can work for one or more cameras, and only requires that all points on the floor were visible, at least for one camera (with sufficient resolution).

At the stage of S70, after the relative position of the players will be identified in the three-dimensional model, this position is then put back on the type of two-dimensional image football field, and the relative border project around a player identified on his position in three-dimensional model. Besides, at the stage S70, the relative border around a player then added to the mask overlay for this player.

On fig.3 shows a plan view of a virtual model 220 football ground. The technology shown in fig.3, players 230, 232 and 234 (left side of the pad) have been identified using the workstation 10 processing content, as players, who are wearing football shirts of different colors, different from the players 236, 238 and 240 (on the right side of the pad), which means that they belong to different commands. Discernment players, thus, allows more simple to detect each player after the events overlap because they can easily discern from each other by the color of their clothes.

Returning again to figure 2, on the stage, S60, the position of each player's track using known technologies such as Kalman filtering, although you should understand that you can use other technologies. Thus, tracking takes place, both in the field 210 of the camera view, and the virtual model 220. In the described technology forecasting of speed is carried out using a workstation 10 processing of contents, using the provisions of players in the virtual model 220 to facilitate the tracing of each player in the field 210 of the camera view.

Stages S60, S70 and repeat until all players on the field will not be processed, as shown diamond S75 decision. Thus, if not all players were processed, then processing proceeds to S60, whereas, if the processing was finished, she stops at the stage of the S80.

As shown in figure 2, exemplified by the method includes an additional stage S85, that can be useful if the image form more than one camera. The stages S30-S80 processing can be performed for the video images from each camera. Each of the players will be processed with the probability of detection for each camera. Therefore, in accordance with the stage of S85, the position of each of the players is assessed in accordance with the probability for each player for each camera, and evaluate the position of the player at the higher of the probabilities provided by each camera, so that the position with the highest probability for each player was identified as a provision for this player.

If you determine that an error occurred while tracking players on a football pitch, then track of each player can be re-initialized on stage S90. Detection of errors when tracking is performed when the probability of detecting a particular player is relatively low for a particular track and in line with this, the track re-initialize.

The result of the method presented in figure 2, is to generate data paths for each player, which represent the position of the player in each frame of the video that represents the path that player during the match. Thus, the data paths represent the position relative to time.

On fig.4, 5A, 5C and 6A approximate illustrations frames sample videos of a football match, in which the actual technology used to track players and formed a three-dimensional model of a football match as a virtual model. In order to decorate description and help in understanding the fig.4, 5B 5D and 6B are presented, respectively, linear image, the corresponding figures 4A, 5A, 5C and 6A. Linear image equivalent and technically identical personnel. They were included to assist in the understanding, if when you copy the application, the illustrations presented on fig.4, 5A, 5C and 6A, become obscured. On fig.4 presents sample illustration video football match, received one camera high definition. On fig.5 presents illustration video fig.4 on which the image was processed to obtain only the background image, using the mean values of each pixel, and the fig.5 presents illustration video fig.4 on which the image was processed to obtain background image only with the use of the variance of each pixel in the image. The corresponding linear drawings shown on fig.5D, you can see that the audience, form the greatest variance of the background image (shown by points on fig.5D). On fig.6 presents an illustration of the track, producing a limited rectangle around each player in accordance example shown in fig.3.

In the bottom half of fig.7 generated virtual model of a football match to represent the positions of players, numbered according to their position on the court, as they are viewed cameras on two-dimensional video in the top half of fig.7. So, the kind of three-dimensional model of a football match corresponds to illustrate the virtual model shown in fig.3. And again, in order to decorate the description and understanding of the linear image corresponding to the fig.7 is on fig.7. This is equivalent to linear image and technically identical illustrative personnel and was included to assist in the understanding, if when you copy the application illustration fig.7 becomes fuzzy.

In accordance with this technology, the tracking information generated in regard to the three-dimensional model of the two-dimensional image of a football match, as described above, may be added to the on-screen appearance taken with a video camera. Such an example is shown in Fig.8. As shown in fig.3, three-dimensional model of a football field is used to assist in tracking and detection of players on a football pitch. After the relative position of the players will on a three-dimensional model, mask for this player then projected onto a two-dimensional image and use to help in the detection and tracking of players on a two-dimensional image. However, after the position of the player will be identified with a relatively high probability, the position of the player on a two-dimensional video camera is known. In line with this, a graphic showing the identity of the players, the score of which was obtained by the algorithm tracking, may be imposed on real time video feed from the camera, use the workstation 10 processing of the content. Thus, as shown in Fig.8, for each of the players 300, 302, 304, 306 prescribes the label 308, 310, 312, 314, which is used to keep track of the player on the site, to track the identity of the player.

Also as an image, on Fig.8 shows two sets of images are selected 320, 322. Each of the parties on the football pitch presents one of the sets of selected images 320, 322. Each image represents an isolated part of the image provided from the camera 20, which is limited to the extent possible, isolate the player on the football ground. Thus, after identifying each of the players, the picture of the player from the video image can be identified and can be displayed with other players in each of the sets corresponding to each of the commands on the football ground. This representation of the selected images can provide automatic isolation of the type of a specific player, without the need for a separate camera to track a player during a football match. Thus, one camera can be used to capture an image of the entire football field, and each player can be monitored in the course of the match, as if a lot of cameras used to keep track of each player. As a result, can be achieved a significant reduction of the cost and complexity of the system.

As explained above, with reference to fig.3 and 3B, when processing tracking each of the players use three-dimensional model of the football venues in order to help identify the players and their location. After the collection of information related to the assessment of the positions of players, and to track this information between each of the video frames (data object path), it becomes possible to form a virtual representation of video images in real time through a synthesis of the image of each of the players and the presentation of these players on a three-dimensional model. In addition, the relative position in the form of model or synthetic camera position within the virtual space can be adjusted using established technologies for adaptation relative types of three-dimensional model of a football field. Thus, for each of the provisions of the players of the time determined by the type of image generated by the camera, perhaps re-create a virtual three-dimensional view of this is really captured the football match of the required position the camera.

As shown in the example shown in figure 9, a three-dimensional model was synthesized by applying data paths for each player in a three-dimensional model (as shown in fig.3 and 8), and the players were introduced to the model of each player in the position changes over time. In addition, because the appearance of three-dimensional model may change the relative position of the camera can be changed to provide a view of the match in a position where in reality Luggage is missing. Thus, as an example, if a penalty kick is awarded, as shown in figure 9, the relative position of the camera can be provided from a position behind the gate, to provide the kind of free-kick from the position where the actual camera is missing.

This is achieved as described above, using the matrix R projection and display the relative position of the three-dimensional position of the camera in the corners of the site. In addition, after evaluating the relative orientation of each player, as described above, then the relative orientation may also stipulate the data path of each player, and synthesized model of the player can be displayed with this orientation, which will change over time. In such a for orientation player, identified as part of the monitoring processing described above, will provide a data path to generate a three-dimensional model of a football match, and such orientation information is used to change the orientation of the player in the model, corresponding to the real image.

Additional technology is shown in figure 10. As noted above, it is possible to synthesize the three-dimensional virtual model of the real football matches, using the tracking information (data object path)generated using the algorithm of tracking and applying synthesized represent each of the players in combination with their relative orientation, as indicated by the lines orientation fig.3 to generate representation is actually played real football matches. In fact, this is a form of compression, because the video image in real time are in the form of synthetic form in three dimensions, by replacing monitored the positions of players on a football pitch synthesized models. The idea of a football match in accordance with the three-dimensional virtual model can be transmitted as a streaming via Internet in the client devices. This arrangement is shown in figure 10, where the workstation 10 processing contents accepts data from the storage 400 data, these data provide a pre-generated synthesized view of football players. View can also include information related to pose towards or behaviors of each of the players, in order to increase the virtual representation of a football match. Thus, the travel data obtained in the process of tracking, described above, may be supplemented and may include information relating to pose towards or manner of conduct of the player, and this information is used to select the appropriate synthesized the player model.

After taking the synthesized view from the vault 400 data workstation 10 processing content can generate a three-dimensional virtual representation of a football match and, if necessary, transfer the data, representing a three-dimensional virtual model of the device 410 subsequent processing. Processor 410 further processing can be used to package information associated with a three-dimensional virtual model of a football match, for its streaming server 420 distribution.

10 each of the many processors 422 client can request a particular form of three-dimensional virtual model of the football field by selecting options related to the preferred team or preferred mind a certain player, who then served in the server 420 distribution. Server 420 distribution can then be executed with the possibility of streaming data, representing a three-dimensional virtual model of a football match in accordance with the preferred options provided by terminals 422 clients. Terminals 422 - clients can then play your preferred form of three-dimensional virtual model, effectively giving the compressed version really captured video images recorded with the camera of 20. Thus, the data to reconstruct a three-dimensional model of a football match may be transmitted in the form of streaming in the client devices that provide narrowband version of real or near-real football match. Such data may simply include data object path, providing tracking of each player in respect of time, and may include the orientation of each player. Three-dimensional model of a football field can be generated software application in the terminals - customers who download synthesized player models and data are applied path to forming a three-dimensional model of a football match. Alternatively, a three-dimensional model can be used to create a computer game that reproduces or recreates the real match.

Options embodiment of the present invention, in which the registration record of events that contains the Association between the player and the event, such as kick the ball will be described below, with reference to 11-14. For clarity of presentation should be understood that the tracking objects may not necessarily represent the players in the game, but may also apply to judges, ball or other elements of the game or to the actual participants of the event, for example, equestrian sports, or in the course of racing. Therefore, the term «player» should be considered as a synonym of the term participant.

By the generation of the registration record of events, the recovery of a three-dimensional model of a football field, as described above, can be supplemented with data related to the actions performed by the players, thus improving the realism of the model. In addition, data stored in the registration record of events can be used to generate statistics match etc. for analysis after the match and to assist in training of players. As a rule, the registration record of events generate using the playback after the match video of the match, although in some versions of the incarnation, the record of events can be generated, essentially, in real time, as described below.

Figure 11 shows schematically the image of a football field, together with a list of events, which enables you to associate the game event with the player. In particular, on 11 are a lot of players (510, 510b, 510, 510d, 510e and 510f) on the football pitch 30, together with the soccer ball 520. Rectangle 530 selecting the player is shown around a player 510e. Rectangle 530 player's choice is associated with a list of events that presents a list of possible events that could be associated with the player. For example, a list of 540 events is possible events, such as "dribble", "header", "Blow left foot", "Strike the right leg", "Violation of the rules of the game" and "Footstep", although it should be understood that the event list can contain any relevant event to associate with the player.

In one variant of the incarnation list 540 events allows the operator to assign or to associate the event with the player for a certain frame or frames of an image. The method by which this is achieved will be described below, with reference to fig.12.

On fig.12 shows a block diagram of the sequence of operations ways to generate the registration record of events in accordance with the option embodiment of the present invention.

At the stage s100 workstation 10 processing monitors the content, within a sequence of video images, the features of the image corresponding to the players, as described above with reference to figure 1-8. The sequence of video images can consistently displayed and reproduced for the operator, using the corresponding display so that events can be associated with players when specific events occur during the game. Events can be associated with players either manually by the user that associates or assigns event for the player automatically using workstation 10, manufacturing, maintenance, which detects events and associate the event with a player or a combination of these two approaches.

At the stage of s105 choose feature of the image corresponding to the player (such as the player 510). In the variant of the incarnation, the operator can select the player, noting feature of the image corresponding to the player, which he wishes to choose (for example, a player 510), using the appropriate device user input, such as a mouse. However, it is understood that other relevant input devices can be used to provide opportunities for the operator to choose a feature of the image corresponding to the player.

Workstation 10 processing contents then assigns this player as the selected player and provides a mapping with the release of the selected player (for example, a player 510), enabling the box 530 selecting the players around the image of the selected player. The operator can thus visually confirm that the player he said, was appointed as the selected player. However, for a person skilled in the art, of course, it will be clear that you can use other appropriate methods for selecting the selected player.

In one variant of the incarnation, for the player to choose, the operator can set the video playback is paused when he notices that happened game event (such as a blow to the left or right foot), so that the player can be associated with this event. Alternatively, workstation 10 processing contents can detect the occurrence of in-game event and can set pause progressive displays the video image so that the user can assign or to associate the event with the player. The way in which detect the occurrence of the event, will be described in more detail below, with reference to fig.13.

In the other variant of the incarnation workstation 10 processing contents during selects the work, at the stage of s105, a feature of the image corresponding to the player in accordance with the detection of the occurrence of the event and the position of the ball 520 relatively players. This is described in more detail below, with reference to fig.13.

Once a player has been selected on the stage, s105, game event is selected from a list 540 events on the stage s110. In the variant of the incarnation, shown in figure 11, workstation 10 processing contents displays a list of 540 events, in Association with the rectangle 530 player's choice. The operator can then choose the event you want to associate with a player from the list 540 events. For example, a player 510 may have just hit the ball 520, and, thus, the operator can select "Kick left foot" from the list of 540 events, as shown by the gray highlight. Thus, in order to ensure that the operator the ability to quickly select the appropriate event of list 540 events, list of 540 events can be displayed, essentially, near the rectangle 530 choice of player, for example, in the video image or near feature of the image corresponding to this player. It preferably provides the opportunity to increase the rate of accommodation of the annotations in your footage, for example, a football match, using data of the events, because the operator list of events will be located close to the player 510. However, be aware that the list of 540 events can be displayed in any appropriate way.

In one variant of the incarnation workstation 10 processing contents at the time of work, chooses an event from the list 540 event, as will be described below, with reference to fig.13. In this case, the list 540 events do not want to display. Alternatively, a list of 540 events can be displayed to inform the operator about the fact that the event was , and what event was associated with any player. This gives the operator the possibility to change the selection if he considers that the workstation 10 processing contents incorrectly associate the event with a player or that improper event was associated with that player.

With this purpose, the operator can enter the appropriate input commands (for example, selecting pause in the graphical user interface) workstation with 10 processing content, for providing display a list of 540 events in such a way that the operator can select a player and a corresponding event, as described above.

As soon as the game event will be selected on the stage s110 from the list 540 events, working station, 10 processing contents associates selected game event with the selected player to generate the registration record of events. Registration of events contains the data pertaining to the time at which the event occurred, unique identifying information that uniquely identify the player, the Association between the player and the event and the event type. Example of record of the events is shown below.

ID player

Event type

The blow left foot

Hitting the right foot

Dribbling

In the example above, the time, record from the start of the match, although it should be understood that you can use any appropriate method of recording the time at which the event occurred. In addition, it should be understood that the event should not be limited to in-game events shown in the list 540 events and that you can use other events, such as "goal", "pass", "jerk" or events relevant to this sport, or an event that is associated list of events. In addition, if the event is not present in the list of 540 events, the operator can manually enter the event type, using the appropriate user interface, such as a keyboard. However, it may slow down the annotation footage of the match, and so the list 540 events usually contains game events that typically occur during a football match.

Variant of the incarnation, which can detect a game event using the workstation 10 processing of the content will be described below, with reference to fig.13.

On fig.13 shows a block diagram of the sequence of operation and method of detection of in-game event in accordance with the option embodiment of the present invention. Workstation processing contents detects the occurrence of the event, by tracking the movement of the ball 520 and detecting when there is a change in direction of the ball 520. For example, if a player kicks the ball 520, then the direction of the ball 520, for example, varies with the move left move right, when it is viewed from the point of view of the camera 20.

At the stage of s205 workstation 10 processing contents detects the changed direction of the ball 520, analyzing the motion of the ball. If a change in the direction of motion of the ball was not , the process returns to step s200, that tracks the position of a ball 520. However, if a workstation processing contents detects that the direction of movement of the ball 520 changed, then processing proceeds to s210.

As described above, processing tracking generates motion vectors for the ball 520 that map to the difference in the position of the ball 520 between successive frames of the image. Therefore, in the form of embodiment of the present invention, a workstation 10 processing contents during work detects the change of direction of the ball 520, displaying a graph of a direction of movement of the ball for each plane coordinates (for example: in the form of dependence of the horizontal component of the vertical component y (xy graph); horizontal x component from the component z depth (graph xz); and vertical component y component z depth (schedule K)). However be aware that other graphics you can use and that other coordinate systems, such as cylindrical polar coordinates, and spherical polar coordinates, respectively, can be used to represent the motion vectors.

For clarity, it should be understood that the terms "schedule" and "display on the graph" shall not be considered as meaning that the graphs you want to display. In the present description these terms refer to the generation of data values so that additional data manipulation can be as marked on the charts points, as described below.

Approximate schedule xy four vectors v 1 , v 2 , v 3 and v 4 motion presented on figure 14. Workstation 10 processing contents during work associates generated motion vectors with each other to generate the data path. In the example shown in figure 14, the data paths contain the vectors v 1 , v 2 , v 3 and v 4 movement. Workstation 10 processing contents at the time of operation can calculate the angle between each vector v 1 , v 2 , v 3 and v 4 movement and the line passing through the motion vector and parallel to the axis of X, to generate angles θ 1 theta 2 theta 3 and space 4 . Angles θ 1 theta 2 theta 3 and space 4 , thus, represent the absolute angles relative to the fixed the reference direction.

Workstation 10 processing contents then compares the angle relative to the X axis for the motion vector angle associated with the previous motion vector. Change of a direction of movement of the object corresponds to the difference of the angle between a single motion vector and the following vector of movement. In other words, the direction of movement of the ball similarly corner associated with the vector of movement, as shown in figure 14.

Those frames, change of a direction of movement of the ball 520 probably contain gaming event. So the video image that are detected as being associated with a change of direction of the ball 520, detect and mark flag, using workstation 10 processing of the content on the stage s210 how contains events.

In the example shown in figure 14, θ 1 is equal to the angle θ 2 . Therefore not detected object movement direction changes between v 1 and v 2 and, consequently, personnel associated with those motion vectors, are unlikely to contain the event. However, in the example shown in figure 14 theta 2 does not coincide with θ 3 , and can therefore change in direction between v-2 and v 3 , which will workstation 10, and, therefore, personnel associated with those vectors v 2 and v, the movement is likely to contain the event.

In some versions of the incarnation of each frame, which can detect a change in direction of the ball 520, marks with a flag, as containing events. However, if a change of direction is relatively small (for example, between v-2 and v 3 ), it is possible that the ball 520 just deviated from the player or deviated during the flight or was twisted. Therefore, in some versions of the incarnation, workstation 10 processing contents only then detects an event, if the change of direction of the ball 520 (that is, the angle between the vector of movement and the X-axis) is greater than a specified threshold. If necessary, the event detect, if there is a change of the sign of absolute angle at which directed the vector of movement, relative to the reference direction.

This is shown in figure 14 using the vectors v 3 and v 4 movement. In this example, the negative angle θ 4 more than the positive angle θ 3 (where a positive value is accepted as a deviation counterclockwise from the X axis and a negative value is accepted as a deviation clockwise from the X-axis), and angles have different signs. Therefore video assosiated with vectors v 3 and v 4 motion will be detected workstation 10, as containing events. Workstation 10 then flags the flags of these video as containing events.

Processing schedule and detection angles, described above, is carried out for each graph (xy, xz and yz). Workstation detects the event as described above, whenever there is a change of direction of the ball's movement 520, as indicated on any one of the three graphs. As an alternative, they say that the event occurred, only if there is a correlation between the graphs, for example the change of direction is indicated, as in figure xy, on the xz plane.

It should be understood that any appropriate number and types of graphs can be used for detection of the change of direction of the ball 520. In addition, it should be understood that angle, calculated for each vector of movement, does not necessarily represent the absolute value, but can be a relative angle relative to the previous motion vectors, as shown angles α 1 , a 2 and a 3 on figure 14. Therefore, in this example, a change of direction corresponds to the relative angle between two successive motion vectors, not equal to 180 degrees.

Let's come back to fig.13, in the variant of the incarnation, processing device content, also during operation, automatically associates detection game event with the player. This technology relates to the fact that a player who kicks or hits head the ball (that is, involved in the event), probably, will represent the closest player to the ball and, therefore, this player must be associated with this event.

To achieve this, as soon as the flags on the video, detected as containing events, working station, 10 processing contents at the time of work, and detecting, at the stage of s215, the relative distance between the ball 520 and each of the players, to generate the data of distance, containing the values of the distance for each players who refer to the distance between the ball 520 and the relevant players. As a rule, this is done only for those video images or frames that were flagged as containing events, to reduce the resources used for processing. However, the distance between the players and the ball 520 can also be calculated for each video, to generate the registration record of the distance between the players and the ball 520 for each frame or image.

Then, at the stage s220, working station, 10 analyzes the data from the distance, to detect out the players who was closest to the ball 520, when it was game event at the stages of s205 and s210. As a rule, the data analyzed for detection of the smallest distance, and workstation chooses this value, as appropriate nearest to the player, although it should be understood that you can use other technologies.

At the stage of s225, working station, 10 processing contents associates game event for this frame with the player associated with the smallest distance value. This Association can then be saved in the registration record events as described above.

In some versions of the incarnation workstation 10 processing of the content also can detect the type of event, by detecting the relative position of the ball 520 relative to the body of the player. For example, changing the direction of the ball near the head of the player is likely to correspond to shock his head, while the change of direction of the ball near the feet of the player is likely to correspond to shock foot. Therefore workstation 10 analyzes the video, shot, for example, 20 cameras and 22.3, for detection of relative height of the ball 520 relatively Playground 30 and/or player, which were detected as closest to the ball, when the event occurred (as described above with reference to the stages s215-s225), using established technologies such as triangulation. As a rule, the correspondence between the relative height of the ball 520 on the player's and the event associated with that height, remain in the reference table. Therefore workstation selects the appropriate game event from the list 540 events in accordance with the position of the ball 520 relatively player, and data stored in the reference table.

In line with this, the game events may be associated with the players, essentially in real time without human intervention, which would lead to unnecessary delays when placing an annotation in your footage.

In addition, in the case when a workstation 10 processing contents automatically associate the specific type of event with a certain player, workstation 10 can set the playback mode to pause the video in such a way that the operator can check that event type is correctly associated with that player. Otherwise, the operator can select the correct player, noting, for example, a feature of the image corresponding to this player, and choosing the right game event from the list 540 events as described above.

It should be understood that the above-described technology for automatic detection of events, event type and associate the event with the player, in General, is limited to events in which the player comes in contact with the ball, for example, during the lockdown, hitting his head, kick etc. However, other gaming events, such as violation of the rules or jerk, not necessarily connected with the ball 520. Therefore, the above-described technology of manual selection event and automated detection and/or selection of events can be combined accordingly. In addition, it should be understood that, if a list of 540 event does not contain the requested event, then the operator can choose the type of event, using an appropriate input device, such as a keyboard, together with the corresponding graphical user interface, such as a dialog box. It should also be understood that the steps presented on fig.13 do not necessarily need to be performed in sequence and that, for example, stages s215-s225 can be performed in parallel with the stages of s200-s210.

Other tracking technologies ball 520 also can be used, which provides for operator the possibility to tag in locations where the ball is in contact with Playground 30 and, assuming the absence of torsion, wind or other movements that may affect the ball, he may mark the upper point of the parabola, on which the ball flies 520, being in air. These data can then be used as the data path for the ball and they can also be associated with the data in the registration record of events for assistance during the three-dimensional modeling of the match.

As you can understand, the various modifications can be made in the variants of the incarnation, described above, without going beyond the scope of the present invention, which is defined in the attached formula of the invention. For example, although indicative options incarnation were presented with reference to a football match, it should be understood that any other sporting event or entertainment event, such as a theatre performance, you can view using the technology disclosed above. In addition, other processors, in addition to the Cell processor, you can use to perform this technology. Processing in accordance with this technology can be implemented in the form of computer software product that contains statements that are executed by the processor, stored on data carrier, such as a floppy drive, optical drive, hard drive, EPROM, EEPROM, RAM, storage device type flash or any combination of these or other data carriers, or may be transferred using data signals over the network, such as Ethernet, wireless LAN, Internet, or using any combination of these or other networks.

Although the illustrative options for carrying out the invention have been described in detail here, with reference to the attached drawings, it should be understood that the invention is not limited to these exact variant of the incarnation, and that the various changes and modifications can be incarnated person skilled in the art, without going beyond the scope and nature of the invention, which are defined in the attached formula of the invention.

1. Way to generate the registration record of events for gaming events, associated with the physical object, present in a sporting event, method, comprising stages: track, within a sequence of video images, especially the images that match the corresponding physical objects in the sports event; display device, sequence video images, including the tracked features of the image; select tracked peculiarities of the image first feature displayed image, that corresponds to one of the physical objects to refer to this physical object as the selected physical object if you select allocate the first feature of the image; display when you select the first features of the image specified list of possible events in-game events, next to the selected selected first feature of the image and superimposed on an occurrence of the sequence of video images; choose game event from the display of a specified list of possible events of the game's events, to associate with the selected physical object in accordance with the choice of the user manually game event; associate the selected game event with the selected physical object to generate the registration record the events for this physical object.

2. The method according to claim 1, wherein: monitor by performing an analysis of the sequence of video images provisions in each video of the game object associated with an event to generate traffic data, which relate to the direction of motion game object in the sequence of video images; detect the occurrence of the event depending on the change of direction of a game object by parsing the data movement; and stamp the flag of one or more of the images, which were detected appearance of the game the event, to indicate that these video contain the emergence of in-game event.

3. The method of claim 2, which can detect the occurrence of in-game event, if the change in direction of the object in the game more than a specified threshold.

4. The method of claim 2, where: detected by the analysis of the sequence of videos of the provisions of the game the object relative to the selected physical object; and choose the game event of the specified list of events in accordance with the regulations of the game the object relative to the selected physical object.

5. The method of claim 2, where: detect the relative distance between the object of the game and each of the physical objects to generate the data distances, which are related to the distance between the object of the game and each of the physical objects; analyze the data distances to detect, which of physical objects was closest to the game object, when were detected the occurrence of the event; and associate this game event with a physical object were detected, as was the closest to the game object, when there was a game event.

6. The method according to p.2, in which: a sequential sequence of video images; establish a pause in a sequential display of a sequence of video images on the video, which was marked with a flag, as containing the occurrence of the event; and display the first property image together with a specified list of events so that the user can choose the appropriate game event from the list of events to associate with the selected physical object.

7. The method according to claim 1, wherein every physical object is associated with a unique identity that uniquely identify the physical object.

8. Data storage media that contains the saved it readable computer instructions that, when loaded in your computer configure computer to complete method according to claim 1.

9. Device for generating the registration record of events for gaming events, associated with the physical object in the event contains: tracking device designed to track the sequence of videos of the features of the image that correspond to the corresponding physical objects in the sports event; the display device is intended for display video loop, including the tracked features of the image; selector features of the image, designed to select tracked peculiarities of the image of the first visible features of the image, that corresponds to one of physical objects, for refer to this physical object as the selected physical object, selecting the first feature of the image emit; selector-game event, designed to select the event from the display of a specified list of possible events in-game events to associate with the selected physical object in accordance with the choice of the user manually game event; the device of Association, intended to associate the selected event from the selected physical object to generate the registration record of events for this physical object; which, when you select the first features of the image of the specified list of possible events of the game's events reflect on the display device to selected selected first feature of the image and superimposed on an occurrence of the sequence of video images.

10. The device of claim 9, in which: tracking device during operation tracks, through analysis of the sequence of video images, the position of each video game object associated with an event to generate traffic data, which relate to the direction of motion of the object in the game in within a sequence of video images; and the device contains: detector designed to detect the occurrence of the event, depending on the change of direction of a game object, by means of analysis of traffic data; and device flag set, designed for flag is set for one or more of the images, which were detected the occurrence of in-game event to indicate that these video contain the emergence of in-game event.

12. The device according to paragraph 10, in which: tracking device during operation detected by the analysis of the sequence of video images, the position of the object in the game relative to the selected physical object; and selector-game event during operation selects game event of a specified list of events in accordance with the regulations of the game the object relative to the selected physical object.

13. The device according to paragraph 10, in which: tracking device during operation: detecting the relative distance between the object of the game and each of the physical objects for data generation distances, which are related to the distance between the object of the game and each of the physical objects; and analyses the data of distance for detection, which of physical objects was closest to the game object, when were detected appearance in-game event; and the device of Association, during operation, associates game event to the physical object, which were detected as was the closest to the game object, when there was a game event.

14. The device according to paragraph 10, containing: display for displaying the video loop; in which the device is made with the ability to: pause during the sequential display of a sequence of video images on the video, which was marked with a flag, as containing the occurrence of the event; and ensure the display on the display of the first features of the image together with a specified list of events so that the user can choose the appropriate game event from the list of events to associate with the selected physical object.

15. The device of claim 9, in which every physical object is associated with a unique identity that uniquely identify a physical object.

16. Graphical user interface designed to produce the registration record of events for gaming events, associated with physical objects, sports events, and subsequently event associated with the physical object that contains: features images in one frame of a sequence of frames that match the corresponding physical objects of the sport events; the device display intended for display of a sequence of video images, including the tracked features of the image; selector features of the image, designed to select from the displayed image feature of the first features of the image, that corresponds to one of physical objects to refer to this physical object selected as a physical object, selecting the first feature of the image emit; selector-game event, designed to select the event from the display of a specified list of events possible game events to associate with the selected physical object in accordance with the choice of the user manually game event; where the choice of the first features of the image specified list of events possible game events to display on the display device to selected selected first feature of the image and superimposed on an occurrence of the sequence of video images.

17. Device for generating the registration record of events for gaming events, associated with the physical object of a sporting event, containing: a tool for monitoring, within video loop, the characteristics of the image that correspond to the corresponding physical objects of the sport events; the tool to display the sequence of video images, including the tracked features of the image; the tool to select tracked peculiarities of the image of the first visible features of the image, that corresponds to one of the physical objects to refer to this physical object as the selected physical object, selecting the first feature of the image emit; tool to select a game events from the display of a specified list of possible events of the game's events, to associate with the selected physical object in accordance with the choice of the user manually game event; the tool to associate the selected event from the selected physical object, to generate the registration record of events for this physical object; which when selecting a first features of the image of the specified list of possible events of the game's events reflect on the display device to selected selected first feature of the image and a superimposed displayed the sequence of video images.