A method and apparatus for filtering ground staff

Abstract

The application provides a court staff filtering method, comprising: after a camera acquires an image of each frame, performing court boundary line identification and staff identification; filtering all staff whose foot points are outside the boundary line, or filtering staff outside the boundary line according to a staff frame height range. The application filters staff interference around the court, thereby improving the accuracy of capturing key players on the court.

Description

Technical Field

[0001] This application relates to the field of image filtering, and more particularly to a method for filtering people in a sports stadium, and also to a device for filtering people in a sports stadium. Background Technology

[0002] With the development of camera technology, the technology of shooting stadiums based on gimbals has become quite mature. The gimbal can freely rotate 360 ​​degrees, which will make the shooting of stadiums more stable.

[0003] Currently, the use of gimbals for shooting stadiums involves cameramen holding the camera equipment mounted on the gimbal or mounting the gimbal on a movable track for shooting. The gimbal reduces camera shake, greatly ensuring the quality of the footage.

[0004] In existing technologies, image tracking capabilities have been added to the gimbal, which can automatically adjust the tracking based on the actual conditions of the court. However, this method often suffers from reduced tracking ability or even tracking the wrong target due to insufficient image tracking capabilities. Summary of the Invention

[0005] The purpose of this application is to overcome the shortcomings of insufficient gimbal tracking capability in the prior art, and to provide a method for filtering people in a sports stadium, and also relates to a device for filtering people in a sports stadium.

[0006] This application provides a method for filtering stadium personnel, including:

[0007] After the camera captures each frame of the image, it performs boundary line recognition and personnel recognition on the court.

[0008] Filter all personnel whose footpoints are outside the boundary line, or filter personnel outside the boundary line based on the height range of the personnel frame.

[0009] Optionally, the filtering footpoint includes all personnel outside the boundary line, including:

[0010] Define the person's foot point as the bottom midpoint of the recognition box;

[0011] If the footpoint is outside the back boundary line, it is filtered; if the footpoint is inside the boundary line, it is not filtered.

[0012] Optionally, it also includes: if the boundary line is not identified, then filter out people outside the boundary line based on the height range of the person frame.

[0013] Optional, including:

[0014] When starting to film, a general height range for people inside the stadium is preset to filter out people outside the boundary line;

[0015] If one boundary line is identified, the height of the frame closest to that boundary line will be recorded. If this frame height is greater than or less than a preset range, the frame height range of the personnel will be updated in real time.

[0016] Optionally, if the boundary line of a certain side is not identified in the images of the following frames, the person is filtered out by extending the corresponding side border of the athlete.

[0017] This application also provides a stadium personnel filtering device, comprising:

[0018] The recognition module performs boundary line recognition and personnel recognition after the camera acquires each frame of image;

[0019] The filtering module filters out all personnel whose footpoints are outside the boundary line, or filters out personnel outside the boundary line based on the height range of the personnel frame.

[0020] Optionally, the filtering unit includes:

[0021] Foot point unit: Define the foot point of the person as the bottom midpoint of the recognition box;

[0022] The first filtering unit filters out nodes that are outside the back boundary line and does not filter out nodes that are inside the boundary line.

[0023] Optionally, the filtering module also includes:

[0024] If the second filtering unit does not detect the boundary line, it filters out people outside the boundary line based on the height range of the person frame.

[0025] Optionally, the second filtering unit includes:

[0026] When starting to film, a general height range for people inside the stadium is preset to filter out people outside the boundary line;

[0027] If one boundary line is identified, the height of the frame closest to that boundary line will be recorded. If this frame height is greater than or less than a preset range, the frame height range of the personnel will be updated in real time.

[0028] Optionally, the second filtering unit further includes: if the boundary line of a certain side is not identified in the images of the following frames, then the person is filtered by extending the corresponding side border of the athlete as the boundary line.

[0029] The advantages and beneficial effects of this application are as follows:

[0030] This application provides a method for filtering players on a sports field, including: recognizing the field boundary lines and players after the camera acquires each frame of image; filtering all players whose footholds are outside the boundary lines, or filtering players outside the boundary lines based on the height range of their player frames. This application improves the accuracy of capturing key players on the field by filtering out interference from players around the field. Attached Figure Description

[0031] Figure 1 This is a schematic diagram of the personnel filtering process in this application.

[0032] Figure 2 This is the first schematic diagram of the filtered images of stadium personnel in this application.

[0033] Figure 3 This is a second schematic diagram of the filtered images of stadium personnel in this application.

[0034] Figure 4 This is a schematic diagram of the personnel filtering device for the stadium in this application. Detailed Implementation

[0035] The present application will be further described below with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand and implement the present application.

[0036] The following are examples of specific implementation processes provided to illustrate the technical solutions to be protected in this application. However, this application may also be implemented in other ways different from those described herein. Those skilled in the art can implement this application by different technical means under the guidance of the concept of this application. Therefore, this application is not limited to the specific embodiments below.

[0037] This application provides a method for filtering players on a sports field, including: recognizing the field boundary lines and players after the camera acquires each frame of image; filtering all players whose footholds are outside the boundary lines, or filtering players outside the boundary lines based on the height range of their player frames. This application improves the accuracy of capturing key players on the field by filtering out interference from players around the field.

[0038] Please refer to Figure 1 As shown, in scenarios where a gimbal camera is used to visually follow players, this patent can be used to filter out interference from people around the field, thereby improving the accuracy of capturing key players on the field.

[0039] In some ball sports courts, there are usually boundary lines to define the area of ​​play. From the perspective of a visual camera, the boundary lines can be divided into four positions: the left boundary line, the front boundary line, the right boundary line, and the back boundary line.

[0040] After the camera acquires each frame of image, it will perform court boundary line recognition and personnel recognition. In this application, the foot point of the personnel is defined as the bottom midpoint of the recognition box.

[0041] Specifically, the personnel filtering process on the court is as follows: after obtaining an image frame, the court boundary line and personnel are identified. Once the boundary line in a certain direction is identified, all personnel whose foot points are outside the boundary line can be filtered immediately. If the boundary line is not identified, filtering is required by the "relative size method".

[0042] The filtering can employ image recognition methods to identify the court boundaries and personnel. However, this image recognition approach may fail to identify the court edges.

[0043] Specifically, if the camera is facing the stadium, the image captured by the camera will show the front and back boundary lines. If the camera is facing the stands, it will also include at least one left boundary line or a boundary line.

[0044] The "relative size method" includes filtering people based on the principle that objects appear larger when closer and smaller when farther away in camera images, and the height of the people frame can reflect this.

[0045] When you first start filming, you can preset a general height range for the people frame inside the stadium to filter out people outside the boundary line.

[0046] As the filming continues, if a front or back boundary line is identified, the height of the person frame closest to that boundary line will be recorded. If the height of this person frame is greater than or less than a preset range, the height range of the person frame will be updated in real time, so that the height range of the person frame can be adjusted adaptively.

[0047] Furthermore, among the personnel that need to be filtered, the personnel tracker can lock onto the athlete closest to the boundary line. If the boundary line on a certain side is not identified in the following frames, the corresponding side border of that athlete is used as the extended boundary line to filter the personnel.

[0048] Please refer to Figure 2 As shown, Figure 2 (a) is a frame image acquired by the gimbal camera. Figure 2 In (b), the two solid lines represent the front and back boundary lines of the court after boundary line identification. This allows for quick filtering of personnel outside the boundary lines.

[0049] Specifically, according to the pixel coordinate system, the front boundary line and the back boundary line can be represented as:

[0050]

[0051] In the formula k f kb Let b represent the slopes of the front and back boundary lines, respectively. f and b b Let these represent the intercepts of the front and back boundary lines, respectively. Then, the pixels between the front and back boundary lines can be represented as:

[0052] k b x+b b ≤y≤k f x+b f (2)

[0053] Based on the intersection position of the two lines, the range of values ​​for x can be divided into the following three cases:

[0054]

[0055]

[0056] In the formula, x0 represents the x-coordinate of the intersection point of the two lines, p w This indicates the pixel width of the image.

[0057] like Figure 2 As shown in (c), the white transparent area is the filtering area. Personnel near the boundary line are filtered based on their foot coordinates in the pixel coordinate system; that is, the foot coordinates must satisfy formulas (1)-(3), as follows: Figure 2 As shown in (d).

[0058] Behind the back boundary line, there is a group of spectators marked with personnel boxes. Since their feet are all outside the back boundary line, they need to be filtered out. However, in front of the front boundary line, there is a player whose feet are inside the front boundary line, therefore he should not be filtered out. Figure 2 As shown in (e).

[0059] like Figure 3 As shown, Figure 3 (a) is another frame image acquired by the gimbal camera. Figure 3 The two solid lines in (b) are the left and back boundary lines of the court after boundary line identification.

[0060] like Figure 3 As shown in (c), this allows for the rapid filtering of people outside the boundary line.

[0061] like Figure 3 As shown in (d), finally, people outside the front boundary line need to be filtered out, and spectators marked by the people box are also filtered out.

[0062] Figure 3 No front boundary line was identified in the middle, according to Figure 1 The flowchart is then filtered. Figure 3 (e) is a magnified view of a player in the field. Assuming that the height range of the player frame is constantly updated during the filming process, the height of the player frame closest to the front boundary line is the minimum value of the player frame height range. Then, the player's foot point is used as the horizontal line, and this line is used to replace the front boundary line to filter out players outside the front boundary line. When all the players' foot points are within the replaced front boundary line, players in the field are not filtered.

[0063] This application also provides a stadium personnel filtering device, including: an identification module 201 and a filtering module 202.

[0064] Please refer to Figure 4 As shown, the recognition module 201 performs court boundary line recognition and personnel recognition after the camera acquires images of each frame.

[0065] In scenarios where gimbal cameras are used to visually track players, this patent can be used to filter out interference from people around the field, thereby improving the accuracy of capturing key players on the field.

[0066] In some ball sports courts, there are usually boundary lines to define the area of ​​play. From the perspective of a visual camera, the boundary lines can be divided into four positions: the left boundary line, the front boundary line, the right boundary line, and the back boundary line.

[0067] After the camera acquires each frame of image, it will perform court boundary line recognition and personnel recognition. In this application, the foot point of the personnel is defined as the bottom midpoint of the recognition box.

[0068] The filtering module 202 filters all personnel whose footpoints are outside the boundary line, or filters personnel outside the boundary line based on the height range of the personnel frame.

[0069] Specifically, the filtering module 202 includes a first filtering unit. The field personnel filtering process is as follows: after obtaining a frame of image, the field boundary line and personnel recognition are performed. Once the boundary line in a certain direction is recognized, all personnel whose foot points are outside the boundary line can be filtered immediately. If the boundary line is not recognized, filtering is required by the "relative size method".

[0070] The filtering can employ image recognition methods to identify the court boundaries and personnel. However, this image recognition approach may fail to identify the court edges.

[0071] If the second filtering unit does not detect the boundary line, it filters out people outside the boundary line based on the height range of the person frame.

[0072] Specifically, if the camera is facing the stadium, the image captured by the camera will show the front and back boundary lines. If the camera is facing the stands, it will also include at least one left boundary line or a boundary line.

[0073] The "relative size method" includes filtering people based on the principle that objects appear larger when closer and smaller when farther away in camera images, and the height of the people frame can reflect this.

[0074] When you first start filming, you can preset a general height range for the people frame inside the stadium to filter out people outside the boundary line.

[0075] As the filming continues, if a front or back boundary line is identified, the height of the person frame closest to that boundary line will be recorded. If the height of this person frame is greater than or less than a preset range, the height range of the person frame will be updated in real time, so that the height range of the person frame can be adjusted adaptively.

[0076] Furthermore, among the personnel that need to be filtered, the personnel tracker can lock onto the athlete closest to the boundary line. If the boundary line on a certain side is not identified in the following frames, the corresponding side border of that athlete is used as the extended boundary line to filter the personnel.

[0077] Please refer to Figure 2 As shown, Figure 2 (a) is a frame image acquired by the gimbal camera. Figure 2 In (b), the two solid lines represent the front and back boundary lines of the court after boundary line identification. This allows for quick filtering of personnel outside the boundary lines.

[0078] like Figure 2 As shown in (c), the white transparent area is the filtering area. People near the boundary line are filtered based on their foot coordinates in the pixel coordinate system. Figure 2 As shown in (d).

[0079] Behind the back boundary line, there is a group of spectators marked with personnel boxes. Since their feet are all outside the back boundary line, they need to be filtered out. However, in front of the front boundary line, there is a player whose feet are inside the front boundary line, therefore he should not be filtered out. Figure 2 As shown in (e).

[0080] like Figure 3 As shown, Figure 3 (a) is another frame image acquired by the gimbal camera. Figure 3 The two solid lines in (b) are the left and back boundary lines of the court after boundary line identification.

[0081] like Figure 3 As shown in (c), this allows for the rapid filtering of people outside the boundary line.

[0082] like Figure 3As shown in (d), finally, people outside the front boundary line need to be filtered out, and spectators marked by the people box are also filtered out.

[0083] Figure 3 No front boundary line was identified in the middle, according to Figure 1 The flowchart is then filtered. Figure 3 (e) is a magnified view of a player in the field. Assuming that the height range of the player frame is constantly updated during the filming process, the height of the player frame closest to the front boundary line is the minimum value of the player frame height range. Then, the player's foot point is used as the horizontal line, and this line is used to replace the front boundary line to filter out players outside the front boundary line. When all the players' foot points are within the replaced front boundary line, players in the field are not filtered.

[0084] In scenarios where gimbal cameras are used to visually track players, this patent can be used to filter out interference from people around the field, thereby improving the accuracy of capturing key players on the field.

[0085] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for filtering personnel in a sports stadium, characterized in that, include: After the camera captures each frame of the image, it performs boundary line recognition and personnel recognition on the court. When the court boundary line is detected, all personnel whose foot points are outside the boundary line are filtered out; when the court boundary line is not detected, personnel outside the boundary line are filtered out based on the height range of the personnel frame. At the beginning of the tracking shot, a general range of personnel frame heights within the court is preset to filter out personnel outside the boundary lines. As the tracking shot progresses, if a boundary line is identified, the frame height of the person closest to that boundary line is recorded. If this frame height is less than the preset range, the frame height of the nearest athlete is used as the minimum value of the personnel frame height range, and the personnel frame height range is updated in real time to make the personnel frame height range adaptively adjustable. If the boundary line of a certain side is not identified in the subsequent frames of the image, the straight line of the corresponding side border of the athlete closest to that side is used as the virtual boundary line to filter people.

2. The stadium personnel filtering method according to claim 1, characterized in that, The filter footpoints are for all personnel outside the boundary line, including: Define the person's foot point as the bottom midpoint of the recognition box; If the footpoint is outside the back boundary line, it is filtered; if the footpoint is inside the boundary line, it is not filtered.

3. A ball court personnel filtering device, characterized in that, include: The recognition module performs boundary line recognition and personnel recognition after the camera acquires each frame of image; The filtration module includes a first filtration unit and a second filtration unit; The first filtering unit is used to filter out all personnel whose footholds are outside the boundary line when the boundary line of the court is detected; The second filtering unit is used to filter people outside the boundary line based on the height range of the personnel frame when the boundary line of the court is not detected. Specifically, at the beginning of the follow-up shooting, a general personnel frame height range within the court is preset to filter people outside the boundary line. As the follow-up shooting progresses, if a boundary line on one side is detected, the frame height of the person closest to that boundary line is recorded. If this frame height is less than the preset range, the personnel frame height of the nearest athlete is used as the minimum value of the personnel frame height range, and the personnel frame height range is updated in real time so that the personnel frame height range is adaptively adjusted. If a boundary line on a certain side is not detected in the images of the following frames, the straight line of the corresponding side border of the athlete closest to that side is used as a virtual boundary line to filter people.

4. The stadium personnel filtering device according to claim 3, characterized in that, The filtering module further includes: The foot point unit defines the person's foot point as the bottom midpoint of the recognition box.

Images