Billiard ball suit recognition method and system

Abstract

The present invention provides a billiard ball suit recognition method and system. The method uses a high-speed rotating target tracking camera group to capture a status image of each billiard ball during a billiard game, and uses mobile target focus tracking to track and capture the billiard ball rotating at high speed after being hit. According to the correspondence between the billiard ball suit and the billiard ball format, a real-time tracking rule for each billiard ball suit is automatically generated during the corresponding billiard ball format game. According to the real-time tracking rule for each billiard ball suit, the high-speed rotating target tracking camera group is intelligently controlled to automatically capture the image of the billiard ball rotating at high speed after being hit, and multiple continuous images of the billiard ball rotating at high speed after being hit are obtained. The billiard ball suit in each billiard game process status image and the billiard ball suit in multiple continuous images of the billiard ball rotating at high speed are identified, and the suits of the remaining billiard balls on the table, the suits of the billiard balls in the pocket, and the suits of the billiard balls causing changes in the game score are intelligently determined.

Description

Technical Field

[0001] The present invention relates to the technical field of image recognition motion intelligent control, and more particularly to a billiard ball suit recognition method and system. Background Art

[0002] Billiard ball suit recognition is very critical to the development of intelligent and automatic control technology of billiard game systems. Various billiard balls have different suits. During billiard games, the ball moves after hitting the ball, especially high-speed movement and high-speed rotation, which can easily make suit recognition difficult or even incorrect. During billiard games, multiple billiard ball targets of various standards often have different states of movement and stillness, different movement speeds, continuous collision movement and stillness conversion, and edge collision and rebound movement, which makes suit recognition more complicated. Specific problems include: how to capture the state image of each billiard game process during the billiard game and track the high-speed movement and rotation of the billiard ball after hitting the ball, how to generate real-time tracking rules for each billiard game process corresponding to the billiard standard game, how to track and capture the image of the high-speed movement and rotation of the billiard ball after hitting the ball, how to identify the suit of the remaining billiard balls in the pocket on the table and the changed billiard suit, etc. These problems remain to be solved. Therefore, it is necessary to propose a billiard ball suit recognition method and system to at least partially solve the problems existing in the existing technology. Summary of the Invention

[0003] A series of simplified concepts are introduced in the summary of the invention, which will be further explained in detail in the specific implementation method. The summary of the invention does not mean to attempt to limit the key features and necessary technical features of the technical solution for protection, nor does it mean to attempt to determine the scope of protection of the technical solution for protection.

[0004] To at least partially solve the above problems, the present invention provides a billiard ball suit recognition method, comprising:

[0005] S100, using a high-speed rotating moving target tracking camera group to capture images of each billiard ball during the billiard game, and using moving target focus tracking to track and capture the high-speed rotating billiard ball after it is hit;

[0006] S200, automatically generating real-time tracking rules for each billiard game process during a game of the corresponding billiard format according to the correspondence between billiard ball suits and billiard formats;

[0007] S300, based on the real-time tracking rules of each billiards game process, intelligently controls the high-speed rotating target tracking camera group, automatically tracks and captures images of the high-speed rotating billiard ball after the shot, and obtains multiple continuous images of the high-speed rotating billiard ball after the shot;

[0008] S400, identifying the billiard ball suit in each billiard game process state image and the billiard ball suit in multiple continuous images of high-speed rotating billiard balls, and intelligently determining the billiard ball suits remaining on the table, the billiard ball suits in the pockets, and the billiard ball suits that cause changes in the game score.

[0009] Preferably, S100 includes:

[0010] S101, setting up a high-speed rotating moving target tracking camera group at an unobstructed position above the billiard table;

[0011] S102, capturing images of each billiards game process state during the billiards game by a high-speed rotating moving target tracking camera group;

[0012] S103, using moving target focus tracking, tracking and photographing the billiard ball that rotates at high speed after being hit.

[0013] Preferably, S200 includes:

[0014] S201, according to the correspondence between billiard ball suits and billiard ball standards, selecting all billiard ball suit information of the corresponding standard;

[0015] S202, based on the information of all billiard balls of the corresponding standard and the corresponding billiard ball competition process, automatically generate real-time tracking rules for each billiard game process; automatically generating real-time tracking rules for each billiard game process includes: according to the information of all billiard balls of the corresponding standard, one by one corresponding to all rules of the billiard game process from the serve to the ball being pocketed or the change in the game score until the end of the game, automatically tracking each billiard ball according to each billiard ball suit of the cue ball, the scoring ball and the deciding ball, and automatically generating real-time tracking rules for each billiard game process.

[0016] Preferably, S300 includes:

[0017] S301, generating high-speed rotating motion target tracking rule information according to the real-time tracking rule of each billiards game process;

[0018] S302, intelligently controlling the high-speed rotating target tracking camera group according to the high-speed rotating target tracking rule information, automatically tracking and capturing images of the high-speed rotating billiard ball after the shot, and obtaining multiple continuous images of the high-speed rotating billiard ball after the shot.

[0019] Preferably, S400 includes:

[0020] S401, identifying the billiard ball suit in each billiard game process state image and the billiard ball suit in the multiple consecutive images of the billiard ball in high-speed motion and rotation based on each billiard game process state image and the multiple consecutive images of the billiard ball in high-speed motion and rotation;

[0021] S402, based on the billiard ball suits in each billiard game process status image and the billiard ball suits in multiple continuous images of high-speed rotating billiard balls, and according to the real-time tracking rules of each billiard game process, intelligently determine the billiard ball suits remaining on the table, the billiard ball suits in the pockets, and the billiard ball suits that cause changes in the game score.

[0022] The present invention provides a billiard ball suit recognition system, comprising:

[0023] The high-speed moving target tracking module uses a high-speed rotating moving target tracking camera group to capture the status image of each billiard ball during the billiard game, and uses moving target focus tracking to track and capture the high-speed rotating billiard ball after hitting the ball;

[0024] The billiards suit and rule information module automatically generates real-time tracking rules for each billiards game process according to the correspondence between billiards suits and billiards standards during the corresponding billiards standard game process;

[0025] The motion target tracking control module intelligently controls the high-speed rotating motion target tracking camera group according to the real-time tracking rules of each billiards game process, automatically tracks and captures images of the high-speed rotating billiard ball after the ball is hit, and obtains multiple continuous images of the high-speed rotating billiard ball after the ball is hit;

[0026] The billiard ball suit intelligent recognition module identifies the billiard ball suit in each billiard game process status image and the billiard ball suit in multiple continuous images of high-speed rotating billiard balls. It intelligently determines the suit of the remaining billiard balls on the table, the suit of the billiard balls in the pocket, and the suit of the billiard balls that cause changes in the game score.

[0027] Preferably, the high-speed moving target tracking module includes:

[0028] A high-speed rotating motion tracking camera unit is set up in an unobstructed position above the billiard table;

[0029] The billiards process status image shooting unit captures the status image of each billiards game process during the billiards game through a high-speed rotating motion target tracking camera group;

[0030] The high-speed rotating target tracking unit uses the moving target focus tracking to track and shoot the high-speed rotating billiard ball after hitting the ball.

[0031] Preferably, the billiards standard suit and rule information module includes:

[0032] The billiard ball suit and standard corresponding unit selects all billiard ball suit information of the corresponding standard according to the corresponding relationship between billiard ball suit and billiard standard;

[0033] The game process tracking rule unit automatically generates real-time tracking rules for each billiard game process according to the corresponding standard all billiard ball suit information and the corresponding billiard standard game process; the automatic generation of real-time tracking rules for each billiard game process includes: according to the corresponding standard all billiard ball suit information, one by one corresponding to all rules of the billiard game process from serving to billiard ball landing or game score change until the end of the game, automatically tracking each billiard ball according to each billiard ball suit of the cue ball, scoring ball and deciding ball, and automatically generating real-time tracking rules for each billiard game process.

[0034] Preferably, the moving target tracking control module includes:

[0035] The high-speed rotation motion tracking information unit automatically generates high-speed rotation motion target tracking rule information according to the real-time tracking rules of each billiard game process;

[0036] The high-speed billiards camera control unit intelligently controls the high-speed rotating target tracking camera group according to the high-speed rotating target tracking rule information, automatically tracks and shoots the image of the high-speed rotating billiards after hitting the ball, and obtains multiple continuous images of the high-speed rotating billiards after hitting the ball.

[0037] Preferably, the billiard ball suit intelligent recognition module includes:

[0038] A billiard ball suit recognition unit for hitting a ball in a game, which recognizes the suit of the billiard ball in each billiard game process state image and the suit of the billiard ball in the multiple continuous images of the billiard ball in high-speed motion and rotation based on each billiard game process state image and the multiple continuous images of the billiard ball in high-speed motion and rotation;

[0039] The billiard suit intelligent judgment unit intelligently judges the suits of the remaining billiard balls on the table, the suits of the pocketed billiard balls, and the suits that cause changes in the game score based on the suits of the billiard balls in the status image of each billiard game process and the suits of the billiard balls in multiple continuous images of high-speed rotating billiard balls, and the real-time tracking rules of each billiard game process.

[0040] The beneficial effects of the above technical solution include:

[0041] The present invention provides a billiard ball suit recognition method and system. The method uses a high-speed rotating target tracking camera group to capture the status image of each billiard ball during a billiard game, and uses mobile target focus tracking to track and capture the billiard ball that is rotating at high speed after being hit. According to the corresponding relationship between the billiard ball suit and the billiard standard, a real-time tracking rule for each billiard game process is automatically generated during the corresponding billiard standard game process. According to the real-time tracking rule for each billiard game process, the high-speed rotating target tracking camera group is intelligently controlled to automatically track and capture the image of the billiard ball that is rotating at high speed after being hit, and obtain multiple continuous images of the billiard ball that is rotating at high speed after being hit. The billiard ball suit in each billiard game process status image and the billiard ball suit in the multiple continuous images of the high-speed rotating billiard ball are identified, and the billiard ball suit remaining on the table, the billiard ball suit that has been pocketed, and the billiard ball suit that causes a change in the game score are intelligently determined. The invention is beneficial to the improvement of the intelligence and automatic control of the billiard game system; it can identify the billiard game process of various standards and colors of billiard balls; it can solve the problem of the movement of the ball after hitting the ball, and can solve the problem that high-speed movement and high-speed rotation easily cause confusion and difficulty in identification of colors or even identification errors; it can significantly improve the recognition of different states of multiple billiard balls of various standards in the billiard game, tracking of different movement speeds, and more complex color recognition such as continuous impact movement and static conversion and edge impact rebound movement tracking; it can capture the state image of each billiard game process during the billiard game and track the high-speed movement and rotation of the billiard ball after hitting the ball; it can generate real-time tracking rules for each billiard game process corresponding to the billiard standard game; it can track and capture the image of the high-speed movement and rotation of the billiard ball after hitting the ball; it can significantly improve the accuracy of identifying the colors of the remaining pocket billiard balls on the table and the changing colors of the billiard balls.

[0042] The present invention provides a billiard ball suit identification method and system. Other advantages, objectives and features of the present invention will be partially reflected in the following description and will also be understood by those skilled in the art through research and practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0044] Figure 1 The figure is a diagram of an embodiment of a billiard ball suit recognition system of the present invention.

[0045] Figure 2 This is another embodiment of a billiard ball suit recognition system according to the present invention.

[0046] Figure 3 This is a diagram illustrating an embodiment of a billiard ball suit recognition method and system application according to the present invention. DETAILED DESCRIPTION

[0047] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments so that those skilled in the art can implement the invention with reference to the description. Figure 1-Figure 3 As shown, the present invention provides a billiard ball suit recognition method, comprising:

[0048] S100, using a high-speed rotating moving target tracking camera group to capture images of each billiard ball during the billiard game, and using moving target focus tracking to track and capture the high-speed rotating billiard ball after it is hit;

[0049] S200, automatically generating real-time tracking rules for each billiard game process during a game of the corresponding billiard format according to the correspondence between billiard ball suits and billiard formats;

[0050] S300, based on the real-time tracking rules of each billiards game process, intelligently controls the high-speed rotating target tracking camera group, automatically tracks and captures images of the high-speed rotating billiard ball after the shot, and obtains multiple continuous images of the high-speed rotating billiard ball after the shot;

[0051] S400, identifying the billiard ball suit in each billiard game process state image and the billiard ball suit in multiple continuous images of high-speed rotating billiard balls, and intelligently determining the billiard ball suits remaining on the table, the billiard ball suits in the pockets, and the billiard ball suits that cause changes in the game score.

[0052] The principle and effect of the above technical solution are as follows: the present invention provides a billiard ball suit recognition method, comprising: using a high-speed rotating target tracking camera group to capture each billiard ball game process status image during a billiard game, and using a moving target focus tracking to track and capture the high-speed rotating billiard ball after hitting the ball; according to the correspondence between the billiard ball suit and the billiard standard, in the corresponding billiard standard game process, automatically generating a real-time tracking rule for each billiard game process; according to the real-time tracking rule for each billiard game process, intelligently controlling the high-speed rotating target tracking camera group to automatically track and capture the high-speed rotating billiard ball image after hitting the ball, and obtaining multiple continuous images of the high-speed rotating billiard ball after hitting the ball; identifying the billiard ball suit in each billiard game process status image and the billiard ball suit in multiple continuous images of the high-speed rotating billiard ball, and intelligently determining the remaining billiard ball suits on the table, the billiard ball suits in the pockets, and the billiard ball suits that cause changes in the game score. Billiard ball suits; the present invention is conducive to the improvement of the intelligence and automatic control of billiard game systems; it can identify billiard balls of various standards and colors in various billiard game processes; it can solve the problem of ball movement after hitting the ball, and can solve the problem that high-speed movement and high-speed rotation easily cause confusion and difficulty in identifying colors or even recognition errors; it can significantly improve the recognition of different states of multiple billiard balls of various standards in billiard games, tracking different movement speeds, and more complex color recognition accuracy such as continuous impact movement and static conversion and edge impact rebound movement tracking; it can capture the state image of each billiard game process during the billiard game and track the high-speed movement and rotation of the billiard ball after hitting the ball; it can generate real-time tracking rules for each billiard game process corresponding to the billiard standard game; it can track and capture the image of the high-speed movement and rotation of the billiard ball after hitting the ball; it can significantly improve the accuracy of identifying the colors of the remaining billiard balls in the pocket on the table and the changing colors of the billiard balls;

[0053] The logic of the billiard ball suit recognition algorithm is as follows:

[0054] Step 1. Set the image cropping size: This helps to speed up recognition and reduce computational effort.

[0055] Step 2. Scaling to maintain a fixed size: This facilitates modification of cropping dimensions and reduces the impact on subsequent algorithm parameters.

[0056] Step 3. Gaussian blur processing: Select appropriate Gaussian blur kernel parameters to reduce the image light noise and brightness changes;

[0057] Step 4. Ball outline recognition: a. Add mask: set the shape and size of the mask to cover irrelevant areas in the image, reduce the amount of calculation and eliminate the image information in irrelevant areas more thoroughly; b. Color gamut conversion: convert the original BGR color gamut into LAB color gamut and HSV color gamut; LAB color gamut and HSV color gamut can better binarize the image to distinguish the ball outline; by fusing and comparing the results of LAB color gamut and HSV color gamut, the outline can be better obtained; c. Binarization image: empirically set the LAB color gamut and HSV color gamut to correspond to the upper and lower limits of the first optical definition, and use the upper and lower limits corresponding to the first optical definition. The image is binarized with the value of 0.05, and the color image is converted into a black and white image to obtain a first optical binarized image; d. Edge detection: Edge detection is performed on the first optical binarized image to find all contours in the image; e. Contour filtering: All contours in the image obtained based on edge detection are filtered and filtered, and a corresponding first minimum threshold is set according to the contour closure, contour length and contour area; contours smaller than the corresponding first minimum threshold are judged as noise points and deleted, thereby obtaining a final contour containing only the ball; f. Image fusion: The final ball contour after filtering is fused with the original image to obtain an image containing only pixels in the ball area;

[0058] Step 5. White outline recognition: Identify the white outline on the ball and obtain a binary image for size determination: a. Color gamut conversion: Convert the BGR image from step 4.f to the LAB color gamut and HSV color gamut again; the LAB color gamut and HSV color gamut are better binarized to distinguish the white area of the ball; b. Binarized image: Empirically set the new LAB color gamut and HSV color gamut to correspond to the upper and lower limits of the second optical definition, and use the corresponding upper and lower limits to binarize the image, turning the color image into a black and white image, to obtain the second optical binary image; c. Edge detection : Perform edge detection on the second optical binarization image to find all white contours in the image; d. Contour filtering: Filter and screen all white contours obtained based on edge detection, set the corresponding second minimum threshold based on the contour closure, contour length, and contour area; contours smaller than the corresponding second minimum threshold are judged as noise and deleted; thus, the final white area contour of the ball is obtained; e. Size identification: Determine the size of the ball based on the white area parameters; white area parameters include: number of white areas, area, arc length, aspect ratio of the approximate rectangle, and perimeter area ratio;

[0059] Step 6. Identification of color contours: Identify color contours on the ball to determine which ball it is: a. Color gamut conversion: Convert the BGR image in step 4.f to the LAB color gamut and HSV color gamut again; b. Binarization image: Empirically set new LAB color gamut and HSV color gamut corresponding to the upper and lower limits of the third optical definition, use the corresponding upper and lower limits of the third optical definition to binarize the image, and convert the color image into a black and white image to obtain the third optical binary image; c. Edge detection: Perform edge detection on the third optical binary image to find all color contours in the image; d. Contour filtering: Filter and screen all contours obtained based on the edge detection, set the corresponding third minimum threshold based on the contour closure, contour length, and contour area; contours smaller than the corresponding third minimum threshold are judged as noise and deleted; thus, the contour of the final ball with color area is obtained;

[0060] Step 7. Image fusion: Based on the first optical binarized image, the second optical binarized image, the third optical binarized image, and the final image containing only the ball outline, the final white ball outline, and the final ball outline with colored areas obtained in steps 4, 5, and 6, perform debinarization and image bit operations to obtain an image in which only the color area retains the original color and the other areas are all black.

[0061] Step 8. Color calculation: Perform edge detection on the image in step 7 to obtain color contour position information; perform mean calculation on the color within the contour range to obtain the average color of the identified target ball.

[0062] Step 9. Color Prediction: The color average value of the target ball identified in Step 8 is elastically increased in size. This reduces the recognition requirements and reduces the possibility of color errors due to light and angle changes resulting in the color average not being in the database. The color average value corresponds to the corresponding LAB value and HSV value. The color range of the balls learned in advance in the database is then compared. If the average value falls within the color range of the balls learned in advance in the database after the elastic increase, it indicates that the ball is likely a certain number in the database.

[0063] Step 10. Ball Number Determination: Based on the results of the big / small numbers and the color prediction, a logical judgment is made on the final result of the ball. If a clear judgment cannot be made, the most likely ball number and the truth of the result are returned; a true value indicates a correct result, and a false value indicates an uncertain result.

[0064] Step 11. Final Ball Number Determination: Since at least two cameras are used, each ball will provide two recognition results. Based on these results, a logical judgment is performed to determine the most likely ball number. The two recognition results include the ball number and the degree of authenticity. If the recognition results of the two images match, the result is output. If they do not match, the final result is logically judged. The result with the maximum degree of authenticity is selected through a loop.

[0065] Step 12. Learning about the color range of the ball in the database: Using the above algorithm, obtain the color mean. The maximum and minimum values of the mean values obtained from multiple angles of the ball are taken as the color range.

[0066] The camera is installed in different positions and the lighting environment is different, so two sets of algorithm parameters are used. This significantly improves the accuracy and stability of recognition, increases the threshold of recognition fault tolerance, and significantly improves the fault tolerance of billiard ball suit recognition.

[0067] In one embodiment, S100 includes:

[0068] S101, setting up a high-speed rotating moving target tracking camera group at an unobstructed position above the billiard table;

[0069] S102, capturing images of each billiards game process state during the billiards game by a high-speed rotating moving target tracking camera group;

[0070] S103, using moving target focus tracking, tracking and photographing the billiard ball that rotates at high speed after being hit.

[0071] The principle and effect of the above technical solution are as follows: a high-speed rotating target tracking camera group is set up in an unobstructed position above the billiard table; the high-speed rotating target tracking camera group is used to shoot the status image of each billiard ball during the billiard game; the moving target is used to focus and track the billiard ball that is rotating at high speed after hitting the ball; the high-speed rotating target tracking camera group includes: a target tracking compound camera, a target tracking motion control unit, a high-speed rotating motion motor unit and a target tracking high-speed camera shadowless lamp; the target tracking motion control unit controls the high-speed rotating motion motor unit, and the high-speed rotating motion motor unit drives the target tracking compound camera to shoot the status image of each billiard ball during the billiard game; the high-speed camera shadowless lamp eliminates the shadow of the billiard ball and emits light of a set wavelength; the target tracking compound camera monitors the reflected light of the set wavelength emitted by the target tracking high-speed camera shadowless lamp, keeps focusing and tracking the center point of the billiard ball surface, and uses the moving target to focus and track the billiard ball that is rotating at high speed after hitting the ball.

[0072] In one embodiment, S200 includes:

[0073] S201, according to the correspondence between billiard ball suits and billiard ball standards, selecting all billiard ball suit information of the corresponding standard;

[0074] S202, based on the information of all billiard balls of the corresponding standard and the corresponding billiard ball competition process, automatically generate real-time tracking rules for each billiard game process; automatically generating real-time tracking rules for each billiard game process includes: according to the information of all billiard balls of the corresponding standard, one by one corresponding to all rules of the billiard game process from the serve to the ball being pocketed or the change in the game score until the end of the game, automatically tracking each billiard ball according to each billiard ball suit of the cue ball, the scoring ball and the deciding ball, and automatically generating real-time tracking rules for each billiard game process.

[0075] The principle and effect of the above technical solution are as follows: according to the correspondence between billiard ball suits and billiard standards, all billiard ball suit information of the corresponding standard is selected; according to all billiard ball suit information of the corresponding standard, during the corresponding billiard standard game process, real-time tracking rules for each billiard game process are automatically generated; the automatic generation of real-time tracking rules for each billiard game process includes: according to all billiard ball suit information of the corresponding standard, all rules of the billiard game process from serving to billiard ball landing or change in game score until the end of the game are matched one by one, each billiard ball is automatically tracked according to each billiard ball suit of the cue ball, scoring ball and deciding ball, and real-time tracking rules for each billiard game process are automatically generated.

[0076] In one embodiment, S300 includes:

[0077] S301, generating high-speed rotating motion target tracking rule information according to the real-time tracking rule of each billiards game process;

[0078] S302, intelligently controlling the high-speed rotating target tracking camera group according to the high-speed rotating target tracking rule information, automatically tracking and capturing images of the high-speed rotating billiard ball after the shot, and obtaining multiple continuous images of the high-speed rotating billiard ball after the shot.

[0079] The principle and effect of the above technical solution are as follows: according to the real-time tracking rules of each billiard game process, high-speed rotating motion target tracking rule information is generated; according to the high-speed rotating motion target tracking rule information, the high-speed rotating motion target tracking camera group is intelligently controlled to automatically track and shoot the image of the high-speed rotating billiard ball after hitting the ball, and multiple continuous images of the high-speed rotating billiard ball after hitting the ball are obtained; the high-speed rotating motion target tracking camera group is intelligently controlled to automatically track and shoot the image of the high-speed rotating billiard ball after hitting the ball, including: according to the high-speed rotating motion target tracking rule information, the focus point of the high-speed rotating motion target tracking camera always follows the rapid movement of the high-speed rotating billiard ball after hitting the ball; when the tracked target billiard ball falls into the pocket, the adjacent billiard target is automatically tracked until the end of the game and the billiard target tracking is terminated; before the start of each billiard game, all billiard balls on the table are scanned and shot to obtain the first image of all billiard balls on the table; when the first ball is hit, multiple images of the billiard ball during the first hit are continuously shot. the first shot is used to track the moving billiard balls, and the second shot is used to identify the billiard balls that fall into the pocket in the first shot or the billiard balls that cause the match score to change; before the second shot, when all the remaining billiard balls on the table are stationary, a second image of the remaining billiard balls on the table is taken; after the second shot, multiple images of the billiard balls during the second shot are taken continuously until all the billiard balls on the table are stationary; the second shot is used to track the moving billiard balls, and the second shot is used to identify the billiard balls that fall into the pocket in the second shot or the billiard balls that cause the match score to change; after the billiard balls are stationary again after the second shot, a third image of the remaining billiard balls on the table is taken; when a billiard ball falls into the pocket or causes the match score to change in each shot, the billiard ball that falls into the pocket or the billiard balls that cause the match score to change is marked according to the last image of the billiard ball that falls into the pocket or the billiard balls that cause the match score to change; reversely mark multiple images of the billiard balls during the shot, and reversely mark the images of the billiard balls that fall into the pocket or the billiard balls that cause the match score to change in turn.

[0080] In one embodiment, S400 includes:

[0081] S401, identifying the billiard ball suit in each billiard game process state image and the billiard ball suit in the multiple consecutive images of the billiard ball in high-speed motion and rotation based on each billiard game process state image and the multiple consecutive images of the billiard ball in high-speed motion and rotation;

[0082] S402, based on the billiard ball suits in each billiard game process status image and the billiard ball suits in multiple continuous images of high-speed rotating billiard balls, and according to the real-time tracking rules of each billiard game process, intelligently determine the billiard ball suits remaining on the table, the billiard ball suits in the pockets, and the billiard ball suits that cause changes in the game score.

[0083] The principle and effect of the above technical solution are as follows: based on each billiard game process state image and multiple continuous images of high-speed rotating billiard balls, the billiard ball suit in each billiard game process state image and the billiard ball suit in multiple continuous images of high-speed rotating billiard balls are identified; based on the billiard ball suit in each billiard game process state image and the billiard ball suit in multiple continuous images of high-speed rotating billiard balls, according to the real-time tracking rules of each billiard game process, the billiard ball suit remaining on the table, the billiard ball suit in the pocket, and the billiard ball suit causing the change in the game score are intelligently determined; the intelligent determination of the billiard ball suit remaining on the table, the billiard ball suit in the pocket, and the billiard ball suit causing the change in the game score includes: based on the billiard ball suit in the billiard game process state image and the billiard ball suit in the high-speed rotating billiard balls According to each billiard game process status image and multiple continuous images of high-speed rotating billiard balls, the billiard ball suits in each billiard game process status image and the billiard ball suits in multiple continuous images of high-speed rotating billiard balls are respectively identified to obtain the remaining billiard ball suit information on the table, and the billiard ball suit information in the pockets or the billiard ball suit information that causes the game score to change is obtained; according to the real-time tracking rules of each billiard game process, it is determined whether the remaining billiard ball suit information on the table, the obtained billiard ball suit information in the pockets and the billiard ball suit information that causes the game score to change comply with the game rules; the remaining billiard ball suit information on the table, the billiard ball suit information in the pockets and the billiard ball suit information that causes the game score to change are intelligently determined.

[0084] The present invention provides a billiard ball suit recognition system, comprising:

[0085] The high-speed moving target tracking module uses a high-speed rotating moving target tracking camera group to capture the status image of each billiard ball during the billiard game, and uses moving target focus tracking to track and capture the high-speed rotating billiard ball after hitting the ball;

[0086] The billiards suit and rule information module automatically generates real-time tracking rules for each billiards game process according to the correspondence between billiards suits and billiards standards during the corresponding billiards standard game process;

[0087] The motion target tracking control module intelligently controls the high-speed rotating motion target tracking camera group according to the real-time tracking rules of each billiards game process, automatically tracks and captures images of the high-speed rotating billiard ball after the ball is hit, and obtains multiple continuous images of the high-speed rotating billiard ball after the ball is hit;

[0088] The billiard ball suit intelligent recognition module identifies the billiard ball suit in each billiard game process status image and the billiard ball suit in multiple continuous images of high-speed rotating billiard balls. It intelligently determines the suit of the remaining billiard balls on the table, the suit of the billiard balls in the pocket, and the suit of the billiard balls that cause changes in the game score.

[0089] The principle and effect of the above technical solution are as follows: the present invention provides a billiard suit recognition system, comprising: a high-speed moving target tracking module 1, which uses a high-speed rotating moving target tracking camera group to shoot each billiard game process state image during the billiard game, and uses mobile target focus tracking to track and shoot the high-speed rotating billiard ball after hitting the ball; a billiard standard suit rule information module 2, which automatically generates a real-time tracking rule for each billiard game process according to the correspondence between the billiard suit and the billiard standard during the corresponding billiard standard game process; a moving target tracking control module 3, which intelligently controls the high-speed rotating moving target tracking camera group according to the real-time tracking rule of each billiard game process, automatically tracks and shoots the image of the high-speed rotating billiard ball after hitting the ball, and obtains multiple continuous images of the high-speed rotating billiard ball after hitting the ball; a billiard suit intelligent recognition module 4, which identifies the billiard suit in each billiard game process state image and the billiard suit in multiple continuous images of the high-speed rotating billiard ball, and intelligently judges the billiard suit. The present invention can identify the suits of billiard balls remaining on the table, the suits of pocketed billiard balls, and the suits of billiard balls that cause changes in game scores; it is beneficial to the improvement of the intelligence and automatic control of billiard game systems; it can identify billiard balls of various standards and suits in various billiard game processes; it can solve the problem of the movement of the ball after hitting the ball, and can solve the problem that high-speed movement and high-speed rotation easily cause confusion and difficulty in identifying suits or even recognition errors; it can significantly improve the recognition of different states of motion and stillness of multiple billiard balls of various standards in the billiard game process, track different movement speeds, and more complex suit recognition accuracy such as continuous impact motion and stillness conversion and edge impact rebound movement tracking; it can capture the state image of each billiard game process during the billiard game and track the high-speed movement and rotation of the billiard ball after hitting the ball; it can generate real-time tracking rules for each billiard game process corresponding to the billiard standard game; it can track and capture the image of the high-speed movement and rotation of the billiard ball after hitting the ball; it can significantly improve the accuracy of identifying the suits of billiard balls remaining in the pockets on the table and the changed billiard suits;

[0090] The logic of the billiard ball suit recognition algorithm is as follows:

[0091] Step 1. Set the image cropping size: This helps to speed up recognition and reduce computational effort.

[0092] Step 2. Scaling to maintain a fixed size: This facilitates modification of cropping dimensions and reduces the impact on subsequent algorithm parameters.

[0093] Step 3. Gaussian blur processing: Select appropriate Gaussian blur kernel parameters to reduce the image light noise and brightness changes;

[0094] Step 4. Ball outline recognition: a. Add mask: set the shape and size of the mask to cover irrelevant areas in the image, reduce the amount of calculation and eliminate the image information in irrelevant areas more thoroughly; b. Color gamut conversion: convert the original BGR color gamut into LAB color gamut and HSV color gamut; LAB color gamut and HSV color gamut can better binarize the image to distinguish the ball outline; by fusing and comparing the results of LAB color gamut and HSV color gamut, the outline can be better obtained; c. Binarization image: empirically set the LAB color gamut and HSV color gamut to correspond to the upper and lower limits of the first optical definition, and use the upper and lower limits corresponding to the first optical definition. The image is binarized with the value of 0.05, and the color image is converted into a black and white image to obtain a first optical binarized image; d. Edge detection: Edge detection is performed on the first optical binarized image to find all contours in the image; e. Contour filtering: All contours in the image obtained based on edge detection are filtered and filtered, and a corresponding first minimum threshold is set according to the contour closure, contour length and contour area; contours smaller than the corresponding first minimum threshold are judged as noise points and deleted, thereby obtaining a final contour containing only the ball; f. Image fusion: The final ball contour after filtering is fused with the original image to obtain an image containing only pixels in the ball area;

[0095] Step 5. White outline recognition: Identify the white outline on the ball and obtain a binary image for size determination: a. Color gamut conversion: Convert the BGR image from step 4.f to the LAB color gamut and HSV color gamut again; the LAB color gamut and HSV color gamut are better binarized to distinguish the white area of the ball; b. Binarized image: Empirically set the new LAB color gamut and HSV color gamut to correspond to the upper and lower limits of the second optical definition, and use the corresponding upper and lower limits to binarize the image, turning the color image into a black and white image, to obtain the second optical binary image; c. Edge detection : Perform edge detection on the second optical binarization image to find all white contours in the image; d. Contour filtering: Filter and screen all white contours obtained based on edge detection, set the corresponding second minimum threshold based on the contour closure, contour length, and contour area; contours smaller than the corresponding second minimum threshold are judged as noise and deleted; thus, the final white area contour of the ball is obtained; e. Size identification: Determine the size of the ball based on the white area parameters; white area parameters include: number of white areas, area, arc length, aspect ratio of the approximate rectangle, and perimeter area ratio;

[0096] Step 6. Identification of color contours: Identify color contours on the ball to determine which ball it is: a. Color gamut conversion: Convert the BGR image in step 4.f to the LAB color gamut and HSV color gamut again; b. Binarization image: Empirically set new LAB color gamut and HSV color gamut corresponding to the upper and lower limits of the third optical definition, use the corresponding upper and lower limits of the third optical definition to binarize the image, and convert the color image into a black and white image to obtain the third optical binary image; c. Edge detection: Perform edge detection on the third optical binary image to find all color contours in the image; d. Contour filtering: Filter and screen all contours obtained based on the edge detection, set the corresponding third minimum threshold based on the contour closure, contour length, and contour area; contours smaller than the corresponding third minimum threshold are judged as noise and deleted; thus, the contour of the final ball with color area is obtained;

[0097] Step 7. Image fusion: Based on the first optical binarized image, the second optical binarized image, the third optical binarized image, and the final image containing only the ball outline, the final white ball outline, and the final ball outline with colored areas obtained in steps 4, 5, and 6, perform debinarization and image bit operations to obtain an image in which only the color area retains the original color and the other areas are all black.

[0098] Step 8. Color calculation: Perform edge detection on the image in step 7 to obtain color contour position information; perform mean calculation on the color within the contour range to obtain the average color of the identified target ball.

[0099] Step 9. Color Prediction: The color average value of the target ball identified in Step 8 is elastically increased in size. This reduces the recognition requirements and reduces the possibility of color errors due to light and angle changes resulting in the color average not being in the database. The color average value corresponds to the corresponding LAB value and HSV value. The color range of the balls learned in advance in the database is then compared. If the average value falls within the color range of the balls learned in advance in the database after the elastic increase, it indicates that the ball is likely a certain number in the database.

[0100] Step 10. Ball Number Determination: Based on the results of the big / small numbers and the color prediction, a logical judgment is made on the final result of the ball. If a clear judgment cannot be made, the most likely ball number and the truth of the result are returned; a true value indicates a correct result, and a false value indicates an uncertain result.

[0101] Step 11. Final Ball Number Determination: Since at least two cameras are used, each ball will provide two recognition results. Based on these results, a logical judgment is performed to determine the most likely ball number. The two recognition results include the ball number and the degree of authenticity. If the recognition results of the two images match, the result is output. If they do not match, the final result is logically judged. The result with the maximum degree of authenticity is selected through a loop.

[0102] Step 12. Learning about the color range of the ball in the database: Using the above algorithm, obtain the color mean. The maximum and minimum values of the mean values obtained from multiple angles of the ball are taken as the color range.

[0103] The camera is installed in different positions and the lighting environment is different, so two sets of algorithm parameters are used. This significantly improves the accuracy and stability of recognition, increases the threshold of recognition fault tolerance, and significantly improves the fault tolerance of billiard ball suit recognition.

[0104] In one embodiment, the high-speed moving target tracking module includes:

[0105] A high-speed rotating motion tracking camera unit is set up in an unobstructed position above the billiard table;

[0106] The billiards process status image shooting unit captures the status image of each billiards game process during the billiards game through a high-speed rotating motion target tracking camera group;

[0107] The high-speed rotating target tracking unit uses the moving target focus tracking to track and shoot the high-speed rotating billiard ball after hitting the ball.

[0108] The principle and effect of the above technical solution are as follows: the high-speed motion target tracking module includes: a high-speed rotating motion tracking camera unit, a high-speed rotating motion target tracking camera group is set in an unobstructed position above the billiard table; a billiard process state image shooting unit, which shoots each billiard process state image during the billiard game through the high-speed rotating motion target tracking camera group; the high-speed rotating motion target tracking unit uses a moving target to focus and track and shoot the high-speed rotating billiard ball after hitting the ball; the high-speed rotating motion target tracking camera group includes: a target tracking compound camera, a target tracking motion control unit, a high-speed rotating motion motor unit and a target tracking high-speed camera shadowless lamp; the target tracking motion control unit controls the high-speed rotating motion motor unit, and the high-speed rotating motion motor unit drives the target tracking compound camera to shoot each billiard process state image during the billiard game; the high-speed camera shadowless lamp eliminates the shadow of the billiard ball and emits a set wavelength of light; the target tracking compound camera monitors the reflected light of the set wavelength emitted by the target tracking high-speed camera shadowless lamp, keeps focusing and tracking the center point of the billiard ball surface, and uses a moving target to focus and track and shoot the high-speed rotating billiard ball after hitting the ball.

[0109] In one embodiment, the billiards standard, suit, and rule information module includes:

[0110] The billiard ball suit and standard corresponding unit selects all billiard ball suit information of the corresponding standard according to the corresponding relationship between billiard ball suit and billiard standard;

[0111] The game process tracking rule unit automatically generates real-time tracking rules for each billiard game process according to the corresponding standard all billiard ball suit information and the corresponding billiard standard game process; the automatic generation of real-time tracking rules for each billiard game process includes: according to the corresponding standard all billiard ball suit information, one by one corresponding to all rules of the billiard game process from serving to billiard ball landing or game score change until the end of the game, automatically tracking each billiard ball according to each billiard ball suit of the cue ball, scoring ball and deciding ball, and automatically generating real-time tracking rules for each billiard game process.

[0112] The principle and effect of the above technical solution are as follows: the billiards standard and suit rule information module includes: a billiards suit and suit corresponding unit, which selects all billiards suit information of the corresponding standard according to the correspondence between billiards suit and billiard standard; a game process tracking rule unit, which automatically generates real-time tracking rules for each billiards game process according to all billiards suit information of the corresponding standard and the corresponding billiards standard game process; the automatic generation of real-time tracking rules for each billiards game process includes: according to all billiards suit information of the corresponding standard, one by one corresponding to all rules of the billiards game process from serving to billiards ball landing or change in game score until the end of the game, automatically tracking each billiard ball according to each billiard suit of the cue ball, scoring ball and deciding ball, and automatically generating real-time tracking rules for each billiards game process.

[0113] In one embodiment, the moving target tracking control module includes:

[0114] The high-speed rotation motion tracking information unit automatically generates high-speed rotation motion target tracking rule information according to the real-time tracking rules of each billiard game process;

[0115] The high-speed billiards camera control unit intelligently controls the high-speed rotating target tracking camera group according to the high-speed rotating target tracking rule information, automatically tracks and shoots the image of the high-speed rotating billiards after hitting the ball, and obtains multiple continuous images of the high-speed rotating billiards after hitting the ball.

[0116] The principle and effect of the above technical solution are as follows: the motion target tracking control module includes: a high-speed rotating motion tracking information unit, which automatically generates high-speed rotating motion target tracking rule information according to the real-time tracking rules of each billiard game process; a high-speed motion billiard camera control unit, which intelligently controls the high-speed rotating motion target tracking camera group according to the high-speed rotating motion target tracking rule information, automatically tracks and shoots the image of the high-speed rotating billiard ball after hitting the ball, and obtains multiple continuous images of the high-speed rotating billiard ball after hitting the ball; the intelligent control of the high-speed rotating motion target tracking camera group, which automatically tracks and shoots the image of the high-speed rotating billiard ball after hitting the ball includes: according to the high-speed rotating motion target tracking rule information, the focus point of the high-speed rotating motion target tracking camera always follows the rapid movement of the high-speed rotating billiard ball after hitting the ball; when the tracked target billiard ball falls into the pocket, the adjacent billiard target is automatically tracked until the end of the game and the billiard target tracking is ended; before the start of each billiard game, all billiard balls on the table are scanned and photographed to obtain the first image of all billiard balls on the table; when During the first shot, multiple images of the billiard balls during the first shot are continuously captured until all the billiard balls on the table are stationary; the moving billiard balls during the first shot are tracked, and the billiard balls that fall into the pockets during the first shot or the billiard balls that cause a change in the match score are identified; before the second shot, when all the remaining billiard balls on the table are stationary, a second image of the remaining billiard balls on the table is captured; after the second shot, multiple images of the billiard balls during the second shot are continuously captured until all the billiard balls on the table are stationary; the moving billiard balls during the second shot are tracked, and the billiard balls that fall into the pockets during the second shot or the billiard balls that cause a change in the match score are identified; after the billiard balls come to rest again after the second shot, a third image of the remaining billiard balls on the table is captured; when a billiard ball falls into a pocket or causes a change in the match score during each shot, the billiard ball that falls into the pocket or the billiard balls that cause a change in the match score are marked based on the last image of the billiard ball that falls into the pocket or the billiard balls that cause a change in the match score are marked; multiple images of the billiard balls during the shot are reversely marked, and the images of the billiard balls that fall into the pocket or the billiard balls that cause a change in the match score are reversely marked in turn. In one embodiment, the billiard ball suit intelligent recognition module includes:

[0117] A billiard ball suit recognition unit for hitting a ball in a game, which recognizes the suit of the billiard ball in each billiard game process state image and the suit of the billiard ball in the multiple continuous images of the billiard ball in high-speed motion and rotation based on each billiard game process state image and the multiple continuous images of the billiard ball in high-speed motion and rotation;

[0118] The billiard suit intelligent judgment unit intelligently judges the suits of the remaining billiard balls on the table, the suits of the pocketed billiard balls, and the suits that cause changes in the game score based on the suits of the billiard balls in the status image of each billiard game process and the suits of the billiard balls in multiple continuous images of high-speed rotating billiard balls, and the real-time tracking rules of each billiard game process.

[0119] The principle and effect of the above technical solution are as follows: the billiard suit intelligent recognition module includes: a billiard suit recognition unit for hitting the ball in the game, which identifies the billiard suit in each billiard game process state image and the billiard suit in multiple continuous images of high-speed rotating billiards according to each billiard game process state image and multiple continuous images of high-speed rotating billiards; a billiard suit intelligent judgment unit, which intelligently judges the billiard suits of the remaining billiard balls on the table, the suits of the pocket billiard balls and the billiard suits that cause changes in the game score according to the billiard suits in each billiard game process state image and the billiard suits in multiple continuous images of high-speed rotating billiards and the real-time tracking rules of each billiard game process; intelligently judges the billiard suits of the remaining billiard balls on the table, the suits of the pocket billiard balls and the billiard suits that cause changes in the game score according to the billiard suits in each billiard game process state image and the billiard suits in multiple continuous images of high-speed rotating billiards; The suit of billiard balls and the suit of billiard balls that causes changes in game scores include: identifying the suit of billiard balls in each billiard game process state image and multiple continuous images of high-speed rotating billiard balls according to each billiard game process state image and multiple continuous images of high-speed rotating billiard balls, obtaining the suit information of billiard balls remaining on the table, and obtaining the suit information of pocketed billiard balls or the suit information of billiard balls that causes changes in game scores; determining the suit information of billiard balls remaining on the table according to real-time tracking rules of each billiard game process, and obtaining the suit information of pocketed billiard balls and the suit information of billiard balls that causes changes in game scores whether they comply with game rules; and intelligently determining the suit of billiard balls remaining on the table, the suit of pocketed billiard balls, and the suit of billiard balls that causes changes in game scores.

[0120] Although the embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the description and implementation methods. They can be fully applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily implemented. Therefore, without departing from the general concept defined by the claims and the scope of equivalents, the present invention is not limited to the specific details and illustrations shown and described herein.

Claims

1. A billiard ball suit recognition method, characterized in that: include: S100, using a high-speed rotating moving target tracking camera group to capture images of each billiard ball during the billiard game, and using moving target focus tracking to track and capture the high-speed rotating billiard ball after it is hit; S200, automatically generating real-time tracking rules for each billiard game process during a game of the corresponding billiard format according to the correspondence between billiard ball suits and billiard formats; S300, based on the real-time tracking rules of each billiards game process, intelligently controls the high-speed rotating target tracking camera group, automatically tracks and captures images of the high-speed rotating billiard ball after the shot, and obtains multiple continuous images of the high-speed rotating billiard ball after the shot; S400, identifying the billiard ball suits in each billiard game process state image and the billiard ball suits in multiple continuous images of high-speed rotating billiard balls, and intelligently determining the suits of the remaining billiard balls on the table, the suits of the pocketed billiard balls, and the suits of the billiard balls that caused the game score to change; S400 includes: S401, identifying the billiard ball suit in each billiard game process state image and the billiard ball suit in the multiple consecutive images of the billiard ball in high-speed motion and rotation based on each billiard game process state image and the multiple consecutive images of the billiard ball in high-speed motion and rotation; S402, based on the billiard ball suits in each billiard game process state image and the billiard ball suits in multiple continuous images of high-speed rotating billiard balls, and in accordance with the real-time tracking rules for each billiard game process, intelligently determine the suits of the remaining billiard balls on the table, the suits of the pocketed billiard balls, and the suits of the billiard balls that caused the game score to change; Intelligently determining the suits of billiard balls remaining on the table, the suits of pocketed billiard balls, and the suits of billiard balls causing changes in game scores includes: identifying the suits of billiard balls in each billiard game process state image and multiple continuous images of high-speed rotating billiard balls, respectively, based on each billiard game process state image and multiple continuous images of high-speed rotating billiard balls, obtaining the suit information of billiard balls remaining on the table, and obtaining the suit information of pocketed billiard balls or the suit information of billiard balls causing changes in game scores; determining the suit information of billiard balls remaining on the table, obtaining the suit information of pocketed billiard balls and the suit information of billiard balls causing changes in game scores according to the real-time tracking rules of each billiard game process, and whether the suit information of pocketed billiard balls and the suit information of billiard balls causing changes in game scores comply with the rules of the game; intelligently determining the suits of billiard balls remaining on the table, the suits of pocketed billiard balls, and the suits of billiard balls causing changes in game scores; S300 includes: S301, generating high-speed rotating motion target tracking rule information according to the real-time tracking rule of each billiards game process; S302, intelligently controlling a high-speed rotating target tracking camera group based on the high-speed rotating target tracking rule information to automatically track and capture images of the billiard ball rotating at high speed after the shot is hit, thereby acquiring multiple continuous images of the billiard ball rotating at high speed after the shot is hit; The intelligent control of the high-speed rotating target tracking camera group automatically tracks and shoots the image of the high-speed rotating billiard ball after the ball is hit, including: according to the high-speed rotating target tracking rule information, the focus point of the high-speed rotating target tracking camera always follows the rapid movement of the high-speed rotating billiard ball after the ball is hit; when the tracked target billiard ball falls into the pocket, the adjacent billiard target is automatically tracked until the billiard target tracking ends at the end of the game; before the start of each round of billiards, all billiard balls on the table are scanned and photographed to obtain the first image of all billiard balls on the table; when the ball is hit for the first time, multiple billiard ball images of the first hitting process are continuously shot until all billiard balls on the table are stationary; the billiard target moving during the first hitting process is tracked, and the billiard target that falls into the pocket for the first time or the billiard target that causes the change in the game score is identified; when the second Before the first shot, when all the remaining billiard balls on the table are stationary, a second image of the remaining billiard balls on the table is captured; after the second shot, multiple images of the billiard balls during the second shot are captured continuously until all the billiard balls on the table are stationary; the moving billiard balls during the second shot are tracked, and the billiard balls that are pocketed during the second shot or the billiard balls that cause a change in the match score are identified; after the billiard balls are stationary again during the second shot, a third image of the remaining billiard balls on the table is captured; when a billiard ball is pocketed or causes a change in the match score during each shot, the billiard balls that are pocketed or the billiard balls that cause a change in the match score are marked based on the last image of the billiard balls that are pocketed or the billiard balls that cause a change in the match score; multiple images of the billiard balls during the shot are reversely marked, and the images of the billiard balls that are pocketed or the billiard balls that cause a change in the match score are reversely marked in sequence; The logic of the billiard ball suit recognition algorithm is as follows: Ball outline recognition: a. Add mask: set the shape and size of the mask to cover irrelevant areas in the image, reduce the amount of calculation and eliminate the image information in irrelevant areas more thoroughly; b. Color gamut conversion: convert the original BGR color gamut into LAB color gamut and HSV color gamut; LAB color gamut and HSV color gamut can better binarize the image to distinguish the ball outline; by fusing and comparing the results of LAB color gamut and HSV color gamut, the outline can be better obtained; c. Binarization image: empirically set the LAB color gamut and HSV color gamut to correspond to the upper and lower limits of the first optical definition, and use the corresponding upper and lower limits of the first optical definition to Image binarization: converting a color image into a black and white image to obtain a first optical binarization image. d. Edge detection: performing edge detection on the first optical binarization image to find all contours in the image. e. Contour filtering: filtering all contours in the image obtained based on edge detection, setting a corresponding first minimum threshold based on the contour closure, contour length, and contour area. Contours smaller than the corresponding first minimum threshold are identified as noise and deleted, thereby obtaining a final contour containing only the ball. f. Image fusion: fusing the filtered final ball contour with the original image to obtain an image containing only pixels in the ball area. Identify the white contours on the ball and obtain a binary image for size determination: a. Color gamut conversion: Convert the BGR image in step 4.f to the LAB color gamut and HSV color gamut again; the LAB color gamut and HSV color gamut are better binarized to distinguish the white area of the ball; b. Binarized image: Empirically set the new LAB color gamut and HSV color gamut corresponding to the second optically defined upper and lower limits, use the corresponding second optically defined upper and lower limits to binarize the image, and convert the color image into a black and white image to obtain a second optical binary image; c. Edge detection: Perform edge detection on the second optical binary image to find all white contours in the image; d. Contour filtering: Filter and screen all white contours obtained based on edge detection, and set the corresponding second minimum threshold based on contour closure, contour length, and contour area; contours smaller than the corresponding second minimum threshold are judged as noise and deleted; thus, the final outline of the white area of the ball is obtained; e. Size identification: Determine the size of the ball based on the white area parameters; the white area parameters include: the number of white areas, area, arc length, aspect ratio of the approximate rectangle, and perimeter area ratio; Identify the colored contours on the ball to determine which ball it is: a. Color gamut conversion: Convert the BGR image in step 4.f to the LAB color gamut and HSV color gamut again; b. Binarize the image: Empirically set the new LAB color gamut and HSV color gamut corresponding to the upper and lower limits of the third optical definition, use the corresponding upper and lower limits of the third optical definition to binarize the image, and convert the color image into a black and white image to obtain the third optical binary image; c. Edge detection: Perform edge detection on the third optical binary image to find all color contours in the image; d. Contour filtering: Filter and screen all contours obtained based on the edge detection, set the corresponding third minimum threshold based on the contour closure, contour length, and contour area; Contours smaller than the corresponding third minimum threshold are judged as noise and deleted; thus, the contour of the colored area of the final ball is obtained; The obtained first optical binary image, the second optical binary image, the third optical binary image, the final image containing only the ball outline, the final white ball outline, and the final ball outline with a colored area are subjected to debinarization processing and image bit operation to obtain an image in which only the color area retains the original color and the other areas are all black; Color calculation: perform edge detection on the image fusion basis to obtain the color contour position information; perform mean calculation on the color within the contour range to obtain the average color of the identified target ball; Color prediction: The target ball's color average is identified and its range size is increased elastically. This reduces recognition requirements and reduces the possibility of color errors due to light and angle changes resulting in the color average not being in the database. The color average corresponds to the corresponding LAB value and HSV value. This is then compared to the color range of balls learned in the database. If the average falls within the color range of balls learned in the database after the elastic increase, it indicates that the ball is likely a certain number in the database. Ball number judgment: Based on the results of the big and small numbers and the prediction of the color; logically judge the final result of the ball; if the judgment cannot be made clearly, the most likely ball number and the truth of the result are returned; true indicates that the result is correct, and false indicates that the result is uncertain; Final ball number determination: Since at least two cameras are used, each ball will provide two recognition results. Based on these results, a logical judgment is made to determine the most likely ball number. The two recognition results include: the ball number and the result's authenticity. If the recognition results of the two images are consistent, the result is output; if they are inconsistent, the final result is logically judged. The result with the maximum authenticity is selected through a loop. Regarding the learning of the color range of the ball in the database: the color mean is obtained through the above algorithm, and the maximum and minimum values of the mean obtained from multiple angles of the ball are taken as the color range; The camera is installed in different positions and the lighting environment is different, so two sets of algorithm parameters are used. This significantly improves the accuracy and stability of recognition, increases the threshold of recognition fault tolerance, and significantly improves the fault tolerance of billiard ball suit recognition.

2. A billiard ball suit recognition method according to claim 1, characterized in that: S100 includes: S101, setting up a high-speed rotating moving target tracking camera group at an unobstructed position above the billiard table; S102, capturing images of each billiards game process state during the billiards game by a high-speed rotating moving target tracking camera group; S103, using moving target focus tracking, tracking and photographing the billiard ball that rotates at high speed after being hit.

3. A billiard ball suit recognition method according to claim 1, characterized in that: S200 includes: S201, according to the correspondence between billiard ball suits and billiard ball standards, selecting all billiard ball suit information of the corresponding standard; S202, based on the information of all billiard balls of the corresponding standard and the corresponding billiard ball competition process, automatically generate real-time tracking rules for each billiard game process; automatically generating real-time tracking rules for each billiard game process includes: according to the information of all billiard balls of the corresponding standard, one by one corresponding to all rules of the billiard game process from the serve to the ball being pocketed or the change in the game score until the end of the game, automatically tracking each billiard ball according to each billiard ball suit of the cue ball, the scoring ball and the deciding ball, and automatically generating real-time tracking rules for each billiard game process.

4. A billiard ball suit recognition system, characterized in that: include: The high-speed moving target tracking module uses a high-speed rotating moving target tracking camera group to capture the status image of each billiard ball during the billiard game, and uses moving target focus tracking to track and capture the high-speed rotating billiard ball after hitting the ball; The billiards suit and color rule information module automatically generates real-time tracking rules for each billiards game process according to the corresponding relationship between billiards suits and billiards standards during the corresponding billiards standard game process; The motion target tracking control module intelligently controls the high-speed rotating motion target tracking camera group according to the real-time tracking rules of each billiards game process, automatically tracks and captures images of the high-speed rotating billiard ball after the ball is hit, and obtains multiple continuous images of the high-speed rotating billiard ball after the ball is hit; The intelligent billiard ball suit recognition module identifies the billiard ball suit in each billiard game process state image and the billiard ball suit in multiple continuous images of high-speed rotating billiard balls. It also intelligently determines the suit of the remaining billiard balls on the table, the suit of the pocketed billiard balls, and the suit of the billiard balls that caused the game score to change. Billiard ball suit intelligent recognition module, including: A billiard ball suit recognition unit for hitting a ball in a game, which recognizes the suit of the billiard ball in each billiard game process state image and the suit of the billiard ball in the multiple continuous images of the billiard ball in high-speed motion and rotation based on each billiard game process state image and the multiple continuous images of the billiard ball in high-speed motion and rotation; The intelligent billiard suit determination unit intelligently determines the suits of the remaining billiard balls on the table, the suits of the pocketed balls, and the suits that cause score changes based on the suits of the billiard balls in the status image of each billiard game process and the suits of the billiard balls in multiple continuous images of high-speed rotating billiard balls, and the real-time tracking rules of each billiard game process; Intelligently determining the suits of billiard balls remaining on the table, the suits of pocketed billiard balls, and the suits of billiard balls causing changes in game scores includes: identifying the suits of billiard balls in each billiard game process state image and multiple continuous images of high-speed rotating billiard balls, respectively, based on each billiard game process state image and multiple continuous images of high-speed rotating billiard balls, obtaining the suit information of billiard balls remaining on the table, and obtaining the suit information of pocketed billiard balls or the suit information of billiard balls causing changes in game scores; determining the suit information of billiard balls remaining on the table, obtaining the suit information of pocketed billiard balls and the suit information of billiard balls causing changes in game scores according to the real-time tracking rules of each billiard game process, and whether the suit information of pocketed billiard balls and the suit information of billiard balls causing changes in game scores comply with the rules of the game; intelligently determining the suits of billiard balls remaining on the table, the suits of pocketed billiard balls, and the suits of billiard balls causing changes in game scores; Moving target tracking control module, including: High-speed rotation motion tracking information unit automatically generates high-speed rotation motion target tracking rule information based on the real-time tracking rules of each billiards game process; The high-speed billiards camera control unit intelligently controls the high-speed rotating target tracking camera group according to the high-speed rotating target tracking rule information, automatically tracks and shoots the image of the high-speed rotating billiards after the ball is hit, and obtains multiple continuous images of the high-speed rotating billiards after the ball is hit; The intelligent control of the high-speed rotating target tracking camera group automatically tracks and shoots the image of the high-speed rotating billiard ball after the ball is hit, including: according to the high-speed rotating target tracking rule information, the focus point of the high-speed rotating target tracking camera always follows the rapid movement of the high-speed rotating billiard ball after the ball is hit; when the tracked target billiard ball falls into the pocket, the adjacent billiard target is automatically tracked until the billiard target tracking ends at the end of the game; before the start of each round of billiards, all billiard balls on the table are scanned and photographed to obtain the first image of all billiard balls on the table; when the ball is hit for the first time, multiple billiard ball images of the first hitting process are continuously shot until all billiard balls on the table are stationary; the billiard target moving during the first hitting process is tracked, and the billiard target that falls into the pocket for the first time or the billiard target that causes the change in the game score is identified; when the second Before the first shot, when all the remaining billiard balls on the table are stationary, a second image of the remaining billiard balls on the table is captured; after the second shot, multiple images of the billiard balls during the second shot are captured continuously until all the billiard balls on the table are stationary; the moving billiard balls during the second shot are tracked, and the billiard balls that are pocketed during the second shot or the billiard balls that cause a change in the match score are identified; after the billiard balls are stationary again during the second shot, a third image of the remaining billiard balls on the table is captured; when a billiard ball is pocketed or causes a change in the match score during each shot, the billiard balls that are pocketed or the billiard balls that cause a change in the match score are marked based on the last image of the billiard balls that are pocketed or the billiard balls that cause a change in the match score; multiple images of the billiard balls during the shot are reversely marked, and the images of the billiard balls that are pocketed or the billiard balls that cause a change in the match score are reversely marked in sequence; The logic of the billiard ball suit recognition algorithm is as follows: Ball outline recognition: a. Add mask: set the shape and size of the mask to cover irrelevant areas in the image, reduce the amount of calculation and eliminate the image information in irrelevant areas more thoroughly; b. Color gamut conversion: convert the original BGR color gamut into LAB color gamut and HSV color gamut; LAB color gamut and HSV color gamut can better binarize the image to distinguish the ball outline; by fusing and comparing the results of LAB color gamut and HSV color gamut, the outline can be better obtained; c. Binarization image: empirically set the LAB color gamut and HSV color gamut to correspond to the upper and lower limits of the first optical definition, and use the corresponding upper and lower limits of the first optical definition to Image binarization: converting a color image into a black and white image to obtain a first optical binarization image. d. Edge detection: performing edge detection on the first optical binarization image to find all contours in the image. e. Contour filtering: filtering all contours in the image obtained based on edge detection, setting a corresponding first minimum threshold based on the contour closure, contour length, and contour area. Contours smaller than the corresponding first minimum threshold are identified as noise and deleted, thereby obtaining a final contour containing only the ball. f. Image fusion: fusing the filtered final ball contour with the original image to obtain an image containing only pixels in the ball area. Identify the white contours on the ball and obtain a binary image for size determination: a. Color gamut conversion: Convert the BGR image in step 4.f to the LAB color gamut and HSV color gamut again; the LAB color gamut and HSV color gamut are better binarized to distinguish the white area of the ball; b. Binarized image: Empirically set the new LAB color gamut and HSV color gamut corresponding to the second optically defined upper and lower limits, use the corresponding second optically defined upper and lower limits to binarize the image, and convert the color image into a black and white image to obtain a second optical binary image; c. Edge detection: Perform edge detection on the second optical binary image to find all white contours in the image; d. Contour filtering: Filter and screen all white contours obtained based on edge detection, and set the corresponding second minimum threshold based on contour closure, contour length, and contour area; contours smaller than the corresponding second minimum threshold are judged as noise and deleted; thus, the final outline of the white area of the ball is obtained; e. Size identification: Determine the size of the ball based on the white area parameters; the white area parameters include: the number of white areas, area, arc length, aspect ratio of the approximate rectangle, and perimeter area ratio; Identify the colored contours on the ball to determine which ball it is: a. Color gamut conversion: Convert the BGR image in step 4.f to the LAB color gamut and HSV color gamut again; b. Binarize the image: Empirically set the new LAB color gamut and HSV color gamut corresponding to the upper and lower limits of the third optical definition, use the corresponding upper and lower limits of the third optical definition to binarize the image, and convert the color image into a black and white image to obtain the third optical binary image; c. Edge detection: Perform edge detection on the third optical binary image to find all color contours in the image; d. Contour filtering: Filter and screen all contours obtained based on the edge detection, set the corresponding third minimum threshold based on the contour closure, contour length, and contour area; Contours smaller than the corresponding third minimum threshold are judged as noise and deleted; thus, the contour of the colored area of the final ball is obtained; The obtained first optical binary image, the second optical binary image, the third optical binary image, the final image containing only the ball outline, the final white ball outline, and the final ball outline with a colored area are subjected to debinarization processing and image bit operation to obtain an image in which only the color area retains the original color and the other areas are all black; Color calculation: perform edge detection on the image fusion basis to obtain the color contour position information; perform mean calculation on the color within the contour range to obtain the average color of the identified target ball; Color prediction: The target ball's color average is identified and its range size is increased elastically. This reduces recognition requirements and reduces the possibility of color errors due to light and angle changes resulting in the color average not being in the database. The color average corresponds to the corresponding LAB value and HSV value. This is then compared to the color range of balls learned in the database. If the average falls within the color range of balls learned in the database after the elastic increase, it indicates that the ball is likely a certain number in the database. Ball number judgment: Based on the results of the big and small numbers and the prediction of the color; logically judge the final result of the ball; if the judgment cannot be made clearly, the most likely ball number and the truth of the result are returned; true indicates that the result is correct, and false indicates that the result is uncertain; Final ball number determination: Since at least two cameras are used, each ball will provide two recognition results. Based on these results, a logical judgment is made to determine the most likely ball number. The two recognition results include: the ball number and the result's authenticity. If the recognition results of the two images are consistent, the result is output; if they are inconsistent, the final result is logically judged. The result with the maximum authenticity is selected through a loop. Regarding the learning of the color range of the ball in the database: the color mean is obtained through the above algorithm, and the maximum and minimum values of the mean obtained from multiple angles of the ball are taken as the color range; The camera is installed in different positions and the lighting environment is different, so two sets of algorithm parameters are used. This significantly improves the accuracy and stability of recognition, increases the threshold of recognition fault tolerance, and significantly improves the fault tolerance of billiard ball suit recognition.

5. A billiard ball suit recognition system according to claim 4, characterized in that: High-speed moving target tracking module, including: A high-speed rotating motion tracking camera unit is set up in an unobstructed position above the billiard table; The billiards process status image shooting unit captures the status image of each billiards game process during the billiards game through a high-speed rotating motion target tracking camera group; The high-speed rotating target tracking unit uses the moving target focus tracking to track and shoot the high-speed rotating billiard ball after hitting the ball.

6. A billiard ball suit recognition system according to claim 4, characterized in that: Billiards standard suit rules information module, including: The billiard ball suit and standard corresponding unit selects all billiard ball suit information of the corresponding standard according to the corresponding relationship between billiard ball suit and billiard standard; The game process tracking rule unit automatically generates real-time tracking rules for each billiard game process according to the corresponding standard all billiard ball suit information and the corresponding billiard standard game process; the automatic generation of real-time tracking rules for each billiard game process includes: according to the corresponding standard all billiard ball suit information, one by one corresponding to all rules of the billiard game process from serving to billiard ball landing or game score change until the end of the game, automatically tracking each billiard ball according to each billiard ball suit of the cue ball, scoring ball and deciding ball, and automatically generating real-time tracking rules for each billiard game process.

Images