A method for estimating the angular velocity of a table tennis ball

Abstract

The present invention relates to the technical field of intelligent control and discloses a method for estimating the rotational angular velocity of a table tennis ball. The method comprises the following steps: S1: obtaining two frames of camera images of a two-color table tennis ball in flight, taken before and after a net, separated by a time interval ΔT; S2: establishing a reference three-dimensional coordinate system for the two-color table tennis ball, and constructing a posture matrix R of the two-color table tennis ball in flight based on matrix rotation transformation; S3: processing the two-color table tennis ball in the two frames of camera images based on a geometric processing algorithm to obtain the corresponding actual posture matrices of the two-color table tennis ball in flight; and S4: calculating the magnitude and direction of the rotational angular velocity of the two-color table tennis ball based on the two actual posture matrices. By selecting a two-color table tennis ball, designing a reasonable camera layout, and employing an effective method for calculating the magnitude and direction of the rotational angular velocity, the present invention improves the accuracy of table tennis ball rotation estimation, saves costs, and increases practicality.

Description

Technical Field

[0001] The invention belongs to the technical field of intelligent control, and in particular relates to a method for estimating the rotational angular velocity of a table tennis ball. Background Art

[0002] Table tennis is my country's national sport, enjoying a broad mass base and widespread market application. Rotation is a key attribute of the sport, and quantitatively measuring the angular velocity of a table tennis ball during flight is crucial. This not only facilitates analysis of table tennis matches and the understanding and improvement of table tennis techniques and tactics, but can also be applied to table tennis sparring robots, improving their return success rates. Currently, several methods have been published for table tennis ball rotation estimation:

[0003] (1) Patent CN110941795A uses the trajectory of a table tennis ball in the air for estimation. The problem with this method is that the position information of the table tennis ball fluctuates to a certain extent. The influence of the rotation speed on the trajectory of the table tennis ball is easily submerged in these errors, resulting in a large error in the result, making it impossible to achieve accurate estimation.

[0004] (2) Patent CN103019024A discloses a method of tracking a table tennis ball using a telephoto camera on a two-degree-of-freedom pan-tilt platform, and realizing real-time observation and analysis of the rotation information of the table tennis ball by identifying and locating the trademark on the table tennis ball. The problem with this method is that the system is complex and the cost is high. At the same time, the table tennis ball trademark is irregular, difficult to identify and easily blocked, resulting in estimation failure.

[0005] (3) Patent CN112245893A discloses a method and device for measuring the rotation speed of a table tennis ball. The method mainly protects a method of increasing the image frame rate to about 3000fps by reducing the image range. The method has strict requirements on the environment and equipment, is costly, and is not practical. At the same time, the method for calculating the main direction angle is not disclosed. Utility Model Content

[0006] The present invention provides a method for estimating the rotational angular velocity of a table tennis ball. By selecting a two-color table tennis ball, designing a reasonable camera layout, and adopting an effective method for calculating the magnitude and direction of the rotational angular velocity, the method improves the accuracy of table tennis ball rotation estimation, saves costs, and increases practicality. This method solves the technical problem that the current calculation of the rotational angular velocity of a table tennis ball requires a high-frame-rate acquisition camera, which is expensive.

[0007] The present invention can be achieved through the following technical solutions:

[0008] A method for estimating the rotational angular velocity of a table tennis ball comprises the following steps:

[0009] S1, obtaining two frames of camera images before and after the flying two-color ping-pong ball passes through the net, separated by a time interval of ΔT;

[0010] S2. Establish a reference three-dimensional coordinate system for the two-color ping-pong ball, and construct a posture matrix R of the two-color ping-pong ball in flight based on matrix rotation transformation;

[0011] S3. Processing the two-color ping-pong balls in the two frames of camera images respectively according to a geometric processing algorithm to obtain corresponding actual posture matrices of the two-color ping-pong balls in flight;

[0012] S4. Calculate the magnitude and direction of the rotational angular velocity of the two-color ping-pong ball based on the two actual posture matrices.

[0013] Furthermore, the origin of the reference three-dimensional coordinate system is the center of the two-color ping-pong ball, the Z-axis direction is perpendicular to the color boundary equatorial plane of the two-color ping-pong ball, the color boundary equatorial plane is perpendicular to the table surface, the Y-axis direction is in the plane where the camera image is located, and the X-axis direction is determined by the right-hand principle.

[0014] Assume that the two-color ping-pong ball rotates around its own X-axis by angle a, and then rotates around its own Y-axis by angle b. The posture matrix of the two-color ping-pong ball at this time is obtained by solving the matrix rotation transformation. for .

[0015] Furthermore, based on the geometric processing algorithm, the method for obtaining the actual posture matrix of the corresponding two-color ping-pong ball in flight includes:

[0016] Select the smaller color area of ​​the two-color ping-pong ball in the camera image to solve the minimum circumscribed rectangle of any angle, obtain the minimum circumscribed rectangle, and then calculate the angle between the minimum circumscribed rectangle and the vertical direction as angle a;

[0017] The area ratio of the two colors of the two-color ping-pong ball in the camera image is used to solve the problem. Let's assume that the area ratio of the two colors of the two-color ping-pong ball in the camera image is ,in, 、 They represent the areas corresponding to the two colors respectively. According to the area projection theorem, the constraint relationship between angle b and area ratio k is obtained as follows: , and then solve the equation to get the rotation angle around its own Y axis .

[0018] Further, the expressions of the two posture matrices are , , where i=1.2.3, 、 Respectively represent one of the three-dimensional components of the corresponding attitude matrix,

[0019] The direction of the angular velocity of the two-color ping-pong ball is the vector the direction indicated;

[0020] The magnitude of the rotational angular velocity of the two-color ping-pong ball .

[0021] Furthermore, the difference in hue between the two colors of the two-color table tennis ball in the HSV space is greater than 90°.

[0022] Furthermore, two cameras are used to collect two frames of camera images. First, one of the cameras is started to collect one frame of camera image, and after a time interval of ΔT, the other camera is started to collect another frame of camera image.

[0023] A device for estimating the angular velocity of a table tennis ball based on the method for estimating the angular velocity of a table tennis ball described above comprises an acquisition module, a data processing module and a display module.

[0024] The acquisition module includes two cameras arranged on one side of the table and close to the net, for capturing two frames of camera images before and after the two-color table tennis ball passes through the net, with an interval of ΔT;

[0025] The data processing module is used to construct a posture matrix of the two-color ping-pong ball in flight, and solve the actual posture matrices corresponding to the two-color ping-pong ball when the two frames of camera images were captured based on a geometric processing algorithm. Then, based on the two actual posture matrices, the angular velocity and direction of the two-color ping-pong ball are solved.

[0026] The display module is used to display the collection results of the collection module and the processing results of the data processing module.

[0027] Furthermore, one of the cameras is located on one side of the net, and the other camera is located on the other side of the net. The number of pixels contained in the two-color table tennis ball in the camera image accounts for 0.2 to 0.5 of the total number of pixels.

[0028] The beneficial technical effects of the present invention are as follows:

[0029] (1) Using two low-frame-rate cameras with hardware interval triggering to solve the high-frame-rate requirement of a single camera for high-speed rotation of table tennis balls, reducing the cost of acquisition equipment, thus having strong practicality;

[0030] (2) A two-color ping-pong ball is used for posture estimation. Without changing the overall structure of the ping-pong ball, the accuracy of posture detection is improved as much as possible by using color segmentation, thus reducing the computational cost.

[0031] (3) The geometric relationship calculation is used to obtain the two angle parameters in the posture matrix of the two-color table tennis ball. That is, the minimum circumscribed arbitrary angle rectangle of the smaller area in the two-color table tennis ball is used to complete the estimation of the rotation angle of the table tennis ball in one direction; the ratio of the two color areas in the two-color table tennis ball is used to complete the estimation of the rotation angle of the table tennis ball in the other direction; finally, the posture estimation triggered by the interval between the two cameras is used for differential processing to complete the quantitative estimation of the size and direction of the rotation angular velocity of the two-color table tennis ball. The calculation process is relatively simple, the calculation speed is fast, and the real-time performance is good. It can fully meet the control requirements of various current serving robots and can effectively expand the scope of application. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a schematic diagram of the overall process of the present invention;

[0033] Figure 2 This is a schematic diagram of the state in which two cameras of the present invention are set on a table;

[0034] Figure 3 Schematic diagram of the establishment of a reference three-dimensional coordinate system for the two-color table tennis ball of the present invention;

[0035] Figure 4 Schematic diagram of the coordinate system of the two-color table tennis ball of the present invention in a certain actual flying state;

[0036] Figure 5 Schematic diagram of the minimum circumscribed rectangle of any angle of the two-color table tennis ball of the present invention in a certain actual flight state;

[0037] Figure 6 This is a schematic diagram of two frames of camera images obtained by actually photographing a two-color table tennis ball used in the experiment of the present invention. DETAILED DESCRIPTION

[0038] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the following examples and drawings specifically illustrate the method for estimating the angular velocity of rotation of a table tennis ball of the present invention. It should be noted that the description of these implementation methods is used to help understand the present invention, but does not constitute a limitation of the present invention.

[0039] Quantitatively estimating the magnitude and direction of the angular velocity of a table tennis ball during flight has always been a difficult problem in the field of table tennis. In order not to affect the normal flight of the table tennis ball, conventional technology uses visual acquisition equipment to perform non-contact measurement of the rotation of the table tennis ball. However, the table tennis ball's rotation speed of up to 200 revolutions per second requires the frame rate of the visual acquisition equipment to be at least 400 frames, which increases the visual cost and reduces the field of view of the camera. Figure 1As shown, the present invention provides a method for estimating the rotational angular velocity of a table tennis ball. By selecting a two-color table tennis ball, designing a reasonable camera layout, and adopting an effective method for calculating the magnitude and direction of the rotational angular velocity, the method has lower cost, less computational complexity, and greater practical value while ensuring accuracy.

[0040] The details are as follows:

[0041] S1, obtaining two frames of camera images before and after the flying two-color ping-pong ball passes through the net, separated by a time interval of ΔT;

[0042] Considering the actual rotational speed of a flying ping-pong ball, if a single camera is used for capture, it must have a high acquisition frame rate—at least twice the ball's rotational speed. Therefore, the present invention employs two cameras to capture the rotational state of a flying ping-pong ball. Through hardware constraints, one camera is activated to capture a single frame of image. After a time interval ΔT, the other camera is activated to capture another frame. This allows the two cameras to maintain a low-frame acquisition state, allowing them to readily capture a two-color ping-pong ball passing through the net. For example, camera 1 is set to a ping-pong ball detection state at 100 fps to determine whether a ping-pong ball is in its field of view. If so, image acquisition begins. Then, hardware triggering is used to trigger camera 2 to capture images at a fixed time interval ΔT, such as 1 millisecond. This reduces the frame rate of the sampling cameras by tenfold, effectively lowering the frame rate requirement for the acquisition equipment and offering greater practicality.

[0043] At the same time, considering that the more pixels containing the two-color ping-pong balls in the collected camera image, the more conducive it is to subsequent data processing, therefore, Figure 2 As shown, two cameras are set on one side of the table and close to the net, one camera is in front of the net and the other is behind the net. Their positions ensure that the ratio of the number of pixels of the two-color table tennis ball observed in the camera image is between 2 / 10,000 and 5 / 10,000, without affecting the flight of the two-color table tennis ball in the table.

[0044] S2. Establish a reference three-dimensional coordinate system for the two-color ping-pong ball, and construct a posture matrix R of the two-color ping-pong ball in flight based on matrix rotation transformation;

[0045] To facilitate subsequent image processing, the two color areas on the two-color ping-pong ball must have a certain degree of distinction. Therefore, the difference in hue between the two colors in the HSV space must be greater than 90°. For example, a two-color ping-pong ball with half white and half orange on each side can be selected.

[0046] Establish the reference three-dimensional coordinate system where the two-color ping-pong ball is located, such as Figure 3As shown, its origin is at the center of the sphere, the Z-axis is perpendicular to the equatorial plane dividing the white and orange, the equatorial plane is perpendicular to the tabletop, the Y-axis is in the camera image plane of the ping-pong ball, and the X-axis is determined by the right-hand rule;

[0047] We might as well assume that the two-color ping-pong ball rotates around its own X axis by angle a, and then rotates around its own Y axis by angle b to obtain the following Figure 4 The posture shown in the figure can be solved by matrix rotation transformation to obtain the posture matrix of the two-color table tennis ball: Therefore, as long as the two angles a and b are obtained, the posture estimation of the two-color ping-pong ball can be completed.

[0048] S3. Process the two-color ping-pong balls in the two camera images separately according to a geometric processing algorithm to obtain the corresponding actual posture matrix of the two-color ping-pong balls in flight;

[0049] To solve the angle a of rotation around its own X axis: use the area with less white and orange color of the two-color ping-pong ball in the camera image to solve the minimum circumscribed rectangle of any angle, such as Figure 5 As shown, the angle between the minimum circumscribed matrix obtained by solving and the vertical direction, i.e., the direction of gravity acceleration, is the rotation angle a of the two-color table tennis posture around its own X-axis;

[0050] To solve the angle b of rotation around its own Y axis: use the ratio of the white and orange areas of the ping-pong ball in the camera image to solve it. Let the ratio of the white and orange areas in the camera image be ,in 、 Respectively represent the areas corresponding to the two colors. Indicates the white area, is the area of ​​orange. According to the area projection theorem, the constraint relationship between the rotation b around its own Y axis and the area ratio k of white and orange in the camera image can be obtained as follows: , so by solving the equation, we can get the rotation angle around its own Y axis .

[0051] S4. Calculate the magnitude and direction of the rotational angular velocity of the two-color ping-pong ball based on the two actual posture matrices.

[0052] The expressions of the two posture matrices are , , where i=1.2.3, 、 Respectively represent one of the three-dimensional components of the corresponding attitude matrix,

[0053] The direction of the angular velocity of the two-color ping-pong ball is the vector the direction indicated;

[0054] The magnitude of the rotational angular velocity of the two-color ping-pong ball .

[0055] That is, after obtaining the attitude matrix corresponding to the two camera images, we can use any three-dimensional component in the attitude matrix to calculate the rotational angular velocity of the two-color ping-pong ball as follows:

[0056] (1) Calculate the pose estimation of the two-color ping-pong ball corresponding to the camera image obtained by camera 1 and , the corresponding Z-axis vector is ;

[0057] (2) and the pose estimation of the two-color ping pong ball corresponding to the camera image obtained by camera 2 and , the corresponding Z-axis vector is ;

[0058] (3) The direction of the angular velocity of the two-color table tennis ball is the vector the direction indicated;

[0059] The angular velocity of the two-color table tennis ball .

[0060] The present invention also provides a device for estimating the angular velocity of a table tennis ball's rotation based on the method for estimating the angular velocity of a table tennis ball's rotation described above. The device comprises an acquisition module, a data processing module, and a display module. The acquisition module comprises two cameras disposed on one side of a table near a net, for acquiring two frames of camera images of a two-color table tennis ball before and after it passes through the net, separated by a time interval ΔT. The data processing module is configured to construct a posture matrix of the two-color table tennis ball in flight, and to calculate the actual posture matrices corresponding to the two frames of camera images when the two frames of camera images are captured based on a geometric processing algorithm. The device then calculates the magnitude and direction of the angular velocity of the two-color table tennis ball based on the two actual posture matrices. The display module is configured to display the acquisition results of the acquisition module and the processing results of the data processing module.

[0061] Assume that a serving robot is used to launch a table tennis ball with a left-hand spin of 50 rpm during a test. The images captured by camera 1 and camera 2 are as follows: Figure 6 As shown, the interval between the two images is 1 millisecond, and the images of camera 1 and camera 2 are processed as follows:

[0062] The orange area (pixel sum) in the ping-pong ball area identified by camera 1 is smaller than the white area. The orange area is selected to form the smallest circumscribed rectangle of any angle, thereby obtaining the rotation angle of the ping-pong ball around its own X-axis. , the ratio of orange area to white area , so that the current rotation angle of the table tennis ball around its own Y axis can be calculated , so the Z-axis vector is obtained as ;

[0063] The white area (pixel sum) in the ping-pong ball area identified by camera 2 is smaller than the orange area. The white area is selected to form the smallest circumscribed rectangle of any angle, thereby obtaining the rotation angle of the ping-pong ball around its own X-axis. , the ratio of orange area to white area , so that the current rotation angle of the table tennis ball around its own Y axis can be calculated , so the Z-axis vector is obtained as ;

[0064] Thus, the direction vector of the rotation speed of the table tennis ball in this test can be obtained , the direction vector of the test rotation type is left-down rotation, and the speed of rotation is , which is not much different from the test rotation speed of 50rps (revolutions per second), achieving a correct solution.

[0065] Although specific embodiments of the present invention are described above, those skilled in the art should understand that these are merely examples and that various changes or modifications may be made to these embodiments without departing from the principles and essence of the present invention. Therefore, the scope of protection of the present invention is limited by the appended claims.

Claims

1. A method for estimating the angular velocity of a table tennis ball, characterized in that The following steps are involved: S1, obtaining two frames of camera images before and after the flying two-color ping-pong ball passes through the net, separated by a time interval of ΔT; S2. Establish a reference three-dimensional coordinate system for the two-color ping-pong ball, and construct a posture matrix R of the two-color ping-pong ball in flight based on matrix rotation transformation; S3. Processing the two-color ping-pong balls in the two frames of camera images respectively according to a geometric processing algorithm to obtain corresponding actual posture matrices of the two-color ping-pong balls in flight; S4. Calculate the magnitude and direction of the rotational angular velocity of the two-color ping-pong ball based on the two actual posture matrices; According to the geometric processing algorithm, the method for obtaining the actual posture matrix of the corresponding two-color ping-pong ball in flight includes: Select the smaller color area of ​​the two-color ping-pong ball in the camera image to solve the minimum circumscribed rectangle of any angle, obtain the minimum circumscribed rectangle, and then calculate the angle between the minimum circumscribed rectangle and the vertical direction as angle a; The area ratio of the two colors of the two-color ping-pong ball in the camera image is used to solve the problem. Let's assume that the area ratio of the two colors of the two-color ping-pong ball in the camera image is ,in, 、 They represent the areas corresponding to the two colors respectively. According to the area projection theorem, the constraint relationship between angle b and area ratio k is obtained as follows: , and then solve the equation to get the rotation angle around its own Y axis .

2. The method for estimating the rotational angular velocity of a table tennis ball according to claim 1, wherein: The origin of the reference three-dimensional coordinate system is the center of the two-color ping-pong ball, the Z-axis direction is perpendicular to the color boundary equatorial plane of the two-color ping-pong ball, the color boundary equatorial plane is perpendicular to the table surface, the Y-axis direction is in the plane where the camera image is located, and the X-axis direction is determined by the right-hand principle. Assume that the two-color ping-pong ball rotates around its own X-axis by angle a, and then rotates around its own Y-axis by angle b. The posture matrix of the two-color ping-pong ball at this time is obtained by solving the matrix rotation transformation. for .

3. The method for estimating the rotational angular velocity of a table tennis ball according to claim 1, wherein: The expressions of the two posture matrices are , , where i=1.2.3, 、 Respectively represent one of the three-dimensional components of the corresponding attitude matrix, The direction of the angular velocity of the two-color ping-pong ball is the vector the direction indicated; The magnitude of the rotational angular velocity of the two-color ping-pong ball .

4. The method for estimating the rotational angular velocity of a table tennis ball according to claim 1, wherein: The difference in hue between the two colors of the two-color table tennis ball in the HSV space is greater than 90°.

5. The method for estimating the rotational angular velocity of a table tennis ball according to claim 1, wherein: Two cameras are used to collect two frames of camera images. First, one of the cameras is started to collect one frame of camera image. After a time interval of △T, the other camera is started to collect another frame of camera image.

6. A table tennis ball rotational angular velocity estimation device based on the table tennis ball rotational angular velocity estimation method according to claim 1, characterized in that: Including acquisition module, data processing module and display module, The acquisition module includes two cameras arranged on one side of the table and close to the net, for capturing two frames of camera images before and after the two-color table tennis ball passes through the net, with an interval of ΔT; The data processing module is used to construct a posture matrix of the two-color ping-pong ball in flight, and solve the actual posture matrices corresponding to the two-color ping-pong ball when the two frames of camera images were captured based on a geometric processing algorithm. Then, based on the two actual posture matrices, the angular velocity and direction of the two-color ping-pong ball are solved. The display module is used to display the collection results of the collection module and the processing results of the data processing module.

7. The device for estimating the rotational angular velocity of a table tennis ball according to claim 6, wherein: One of the cameras is located on one side of the net, and the other camera is located on the other side of the net. The number of pixels contained in the two-color table tennis ball in the camera image accounts for 0.2 to 0.5 of the total number of pixels.

Images