A tennis motion capture system

Abstract

The application relates to the technical field of motion capture, in particular to a tennis motion capture system, which comprises a motion sensor module, a multi-view synchronous camera module, a data fusion processing module and a real-time feedback module; the motion sensor module collects motion parameters of target joints of a tennis player; the multi-view synchronous camera module acquires spatial displacement data of target skeletal joint nodes when the player hits a ball; the data fusion processing module fuses the motion parameters and the spatial displacement data to generate a three-dimensional skeletal animation model; the real-time feedback module extracts a shoulder-hip torsion angle difference parameter from the three-dimensional skeletal animation model, outputs a directional vibration prompt signal through a tactile feedback device after the hitting action is completed, and outputs an overlapping comparison graph and a gradient heat map of a swing trajectory deviation by using a display terminal. The application can solve the problems of difficult precise capture of action details and provision of quantitative data feedback, overcome defects such as incomplete data collection, insufficient precision or poor adaptability of existing motion capture equipment, and provide strong support for precise optimization of a hitting action of a player.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of motion capture technology, and in particular to a tennis motion capture system. BACKGROUND

[0002] In tennis training and competition, accurate motion analysis is crucial for athletes to improve their technical level. The traditional method of relying solely on the naked eye observation and experience guidance of the coach cannot capture the key details such as the body posture, motion amplitude, and speed of the athlete at the moment of hitting the ball in detail and comprehensively, and cannot provide precise quantitative data feedback to the athlete, which is not conducive to the precise optimization of the hitting motion. Although there is simple video playback to assist in analysis, the data is not accurate and lacks real-time performance. Some existing motion capture devices have problems such as insufficient comprehensive data collection, insufficient accuracy, or difficulty in adapting to complex tennis motions.

[0003] Therefore, there is an urgent need to provide a technical solution to solve the above problems. SUMMARY

[0004] To solve the above technical problems, the present application provides a tennis motion capture system.

[0005] In a first aspect, the present application provides a tennis motion capture system, comprising the following technical solutions:

[0006] A motion sensor module comprising an inertial measurement unit fixed to each target joint of a tennis player, for real-time acquisition of motion parameters of each target joint of the tennis player;

[0007] A multi-view synchronous camera module comprising stereo vision cameras arranged at the midpoint of the baseline and the extension line of the net centerline of the tennis court, a plurality of stereo vision cameras forming a three-dimensional capture area covering the hitting motion of the tennis player, for synchronously acquiring spatial displacement data of a plurality of target skeletal joint nodes of the tennis player when hitting the ball;

[0008] A data fusion processing module for fusing the motion parameters of each target joint and the spatial displacement data of each target skeletal joint node to generate a three-dimensional skeletal animation model containing a time sequence;

[0009] A real-time feedback module for extracting a shoulder and hip twist angle difference parameter from the three-dimensional skeletal animation model, outputting a direction-identifiable vibration prompt signal through a tactile feedback device worn on the non-patella wrist of the tennis player when a preset time threshold is reached after the hitting motion ends, and synchronously outputting an overlapping comparison graph of the three-dimensional skeletal animation model and a pre-stored standard motion model on a display terminal deployed on the tennis court, and projecting a gradient heat map of the swing trajectory deviation on the surface of the three-dimensional skeletal animation model.

[0010] Further, all target joints include: wrist, elbow, shoulder, hip, knee and ankle; and the motion parameters of each target joint include: angular velocity data and acceleration data of the corresponding target joint.

[0011] Further, the number of the stereo vision cameras is three, which are respectively arranged at the midpoint of the baseline of the tennis court and the position of the extension line of the center line of the net, and the frame rate of each stereo vision camera is not less than 200 fps.

[0012] Further, the data fusion processing module is specifically used for:

[0013] The timing alignment is performed on all angular velocity data and all spatial displacement data, the pose drift error of the installation position of each inertial measurement unit is calculated based on each acceleration data, and the dynamic compensation is performed on all angular velocity data through a Kalman filter;

[0014] The shoulder-hip-knee dynamic link model is established according to the biomechanical constraint conditions, and the biomechanical constraint conditions include the adjacent joint rotation angle range limitation and the limb length constant condition;

[0015] The shoulder-hip-knee dynamic link model and all spatial displacement data are input into the inverse kinematics algorithm, the joint rotation angle optimal solution is iteratively solved, and the three-dimensional skeletal animation model containing the time sequence is generated.

[0016] Further, the data fusion processing module is further used for:

[0017] The real-time three-dimensional spatial coordinates and rotation angle data of the shoulder joint and the hip joint of the tennis player are extracted from the three-dimensional skeletal animation model;

[0018] In the acceleration phase of the ball hitting action of the tennis player, the first rotation angle value of the shoulder joint around the vertical trunk direction and the second rotation angle value of the hip joint around the vertical trunk direction are respectively calculated in real time, and the shoulder-hip torsion angle difference parameter is calculated in real time according to the difference between the first rotation angle value and the second rotation angle value.

[0019] Further, the real-time feedback module is specifically used for:

[0020] The pre-stored standard action model matched with the current ball hitting action type of the tennis player is acquired, and the three-dimensional skeletal animation model is normalized with the ball hitting action time axis of the pre-stored standard action model; wherein the pre-stored standard action model contains the time sequence motion data of each target joint in an ideal state.

[0021] Superimpose the joint track of the three-dimensional skeletal animation model and the joint track of the pre-stored standard action model in the same three-dimensional coordinate system, generate an overlapping comparison graph and output and display the overlapping comparison graph through the display terminal.

[0022] Further, the real-time feedback module is specifically used for:

[0023] Calculate the frame-by-frame Euclidean distance between the racquet swing trajectory curve of the tennis player and the standard trajectory curve corresponding to the pre-stored standard action model, and generate a deviation distance matrix;

[0024] According to the numerical distribution of the deviation distance matrix, map the gradient heat map representing the racquet swing trajectory deviation on the surface of the three-dimensional skeletal animation model.

[0025] In a second aspect, the application provides a tennis motion capture method, comprising the following technical solutions:

[0026] Real-time acquisition of the motion parameters of each target joint of the tennis player through an inertial measurement unit fixed to each target joint of the tennis player;

[0027] Synchronous acquisition of the spatial displacement data of a plurality of target skeletal joints of the tennis player when hitting the ball through a stereo vision camera arranged at the midpoint of the baseline and the position of the extended line of the net centerline of the tennis court;

[0028] Fusion of the motion parameters of each target joint and the spatial displacement data of each target skeletal joint to generate a three-dimensional skeletal animation model containing a time sequence;

[0029] According to the shoulder and hip twist angle difference parameters extracted from the three-dimensional skeletal animation model, when a preset time threshold is reached after the end of the hitting action, output a direction-identifiable vibration prompt signal through a tactile feedback device worn on the non-racquet hand wrist of the tennis player, and simultaneously output an overlapping comparison graph of the three-dimensional skeletal animation model and the pre-stored standard action model through a display terminal deployed on the tennis court, and project a gradient heat map of the racquet swing trajectory deviation on the surface of the three-dimensional skeletal animation model.

[0030] Further, all target joints include wrists, elbows, shoulders, hips, knees, and ankles; and the motion parameters of each target joint include angular velocity data and acceleration data of the corresponding target joint.

[0031] Further, the number of stereo vision cameras is three, and each stereo vision camera is arranged at the midpoint of the baseline and the position of the extended line of the net centerline of the tennis court, and the frame rate of each stereo vision camera is not less than 200 fps.

[0032] The application solves the problems that the action details cannot be accurately captured and quantitative data feedback cannot be provided by relying on the naked eye observation and experience guidance of the coach in the prior art, overcomes the defects of the existing part of motion capture devices such as incomplete data collection, insufficient precision or poor adaptability, and provides strong support for the accurate optimization of the ball hitting action of the player, and helps to improve the technical level of the player.

[0033] Other advantages, objects, and features of the application will be set forth in part in the following specification taken in conjunction with the accompanying drawings, and in part will become apparent to those skilled in the art from a consideration of the following specification and drawings, or can be learned from the practice of the application. BRIEF DESCRIPTION OF DRAWINGS

[0034] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of these drawings.

[0035] Figure 1 It is a structural schematic diagram of a tennis action capture system;

[0036] Figure 2 It is a flowchart of a tennis action capture method. DETAILED DESCRIPTION

[0037] The technical solutions in the embodiments of the present application will be described clearly and completely below. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0038] Figure 1 A structural schematic diagram of an embodiment of a tennis action capture system 100 provided by the present application is shown. As shown in the figure, Figure 1As shown, the system 100 comprises a motion sensor module 110, a multi-view synchronous camera module 120, a data fusion processing module 130 and a real-time feedback module 140.

[0039] The motion sensor module 110 comprises an inertial measurement unit fixed to each target joint of the tennis player, for real-time acquisition of the motion parameters of each target joint of the tennis player.

[0040] The inertial measurement unit (IMU) refers to a micro-sensor device installed at the target joint of the tennis player, comprising a three-axis gyroscope and a three-axis accelerometer. Each target joint is fixed with an inertial measurement unit. All target joints include the wrist, elbow, shoulder, hip, knee and ankle. The motion parameters of each target joint include the angular velocity data (instantaneous angular velocity of joint rotation around three axes) and acceleration data (instantaneous acceleration of joint linear motion) of the corresponding target joint.

[0041] Specifically, each inertial measurement unit (IMU) acquires the motion parameters of the target joint of the tennis player at a preset sampling frequency The motion parameters of the target joint of the tennis player are acquired in real time, including: ① three-axis angular velocity components output by the three-axis gyroscope , , , unit: rad / s (rad / s); ② three-axis acceleration components output by the accelerometer , , , unit: m / s² (m / s²). ③ Add a precise timestamp to each frame of data , wherein is the sampling number. ④ The angular velocity data and acceleration data of each target joint of the tennis player with timestamp are sent to the data fusion processing module 130 through a wireless transmission protocol.

[0042] The multi-view synchronous camera module 120 comprises a stereo vision camera arranged at the midpoint of the bottom line on both sides of the tennis court and the position of the extension line of the center line of the net. Multiple stereo vision cameras form a three-dimensional capture area covering the tennis player's hitting action, for synchronously acquiring the spatial displacement data of multiple target skeletal joint nodes when the tennis player hits the ball.

[0043] The stereovision camera refers to a camera device with a binocular lens, and three-dimensional space coordinate measurement is realized through parallax calculation. The number of stereovision cameras is three by default, which are arranged at the midpoint of the baseline on the left side of the tennis court, the midpoint of the baseline on the right side of the tennis court, and the position of the extension line of the center line of the net (a point extending from the center of the net to the outside of the court by a preset distance).

[0044] The position of the extension line of the center line of the net specifically refers to a position extending 3-5 meters from the center of the net (0.914-meter-high net) to the outside of the court (to avoid blocking the movement of the tennis player). The three-dimensional capture area refers to a three-dimensional space formed by the intersection of the fields of view of the three stereovision cameras, covering the range of body movement of the tennis player when hitting the ball.

[0045] The target skeletal joint includes the bilateral joints of the shoulder joint, elbow joint, wrist joint, hip joint, knee joint, and ankle joint, a total of 12 (6 on the left side and 6 on the right side). The spatial displacement data refers to the real-time position change of the target skeletal joint in the three-dimensional space.

[0046] Specifically: ① For each target skeletal joint, the parallax of the target skeletal joint is calculated according to the binocular parallax principle: ; ② The depth of the target skeletal joint is reconstructed based on the camera intrinsic parameters (focal length , baseline distance ): ; ③ The pixel coordinates of the target skeletal joint are converted into three-dimensional coordinates in the world coordinate system: , ; is the camera optical center pixel coordinate. ④ For each target skeletal joint , the spatial displacement vector between adjacent frames is calculated: ; the same timestamp is added to all target skeletal joints , forming synchronous spatial displacement data and sending to the data fusion processing module 130 through a wireless transmission protocol.

[0047] The data fusion processing module 130 is used for fusing the motion parameters of each target joint and the spatial displacement data of each target skeletal joint to generate a three-dimensional skeletal animation model containing a time sequence.

[0048] The three-dimensional skeletal animation model refers to a dynamic model generated by fusing the motion parameters and the spatial displacement data. ​​​​​

[0049] In an alternative manner, the data fusion processing module 130 is specifically used for:

[0050] Timing alignment is performed on all angular velocity data and all spatial displacement data, and based on each acceleration data, the pose drift error of the installation position of each inertial measurement unit is calculated respectively, and all angular velocity data is dynamically compensated by a Kalman filter.

[0051] Specifically: ①Interpolation resampling is performed on all angular velocity data and all spatial displacement data, and timing alignment is performed, ensuring that the timestamps of all data points are aligned to a unified time sequence. ②Gravity component separation is performed on each acceleration data, and the installation position pose drift error of the inertial measurement unit is calculated through the low-frequency component of the acceleration data. ③The drift error after acceleration separation is determined according to the observation value of the defined state equation, and the drift error is estimated by a Kalman filter, and the calibrated angular velocity is output.

[0052] A shoulder-hip-knee dynamic link model is established according to biomechanical constraints, and the biomechanical constraints include adjacent joint rotation angle range limitation and limb length constant condition.

[0053] Specifically: ①Define the joint chain structure of the shoulder-hip-knee dynamic link model: main joint chain: shoulder→hip→knee (left and right sides are symmetrical); secondary joint chain: shoulder joint→hip joint (torso torsion axis). ②For each target joint , the rotation angle is applied to the physiological constraint: ; for example, the shoulder joint external rotation angle , the hip joint internal rotation angle . ③The limb length constant condition is that the distance between adjacent target skeletal joint nodes (parent node) and (child node) is constrained: ; represents the pre-stored limb length, such as the shoulder-hip distance . ④Based on the above constraints, the shoulder-hip-knee dynamic link model is established: ; represents the hip joint rotation matrix, represents the initial direction vector of the leg.

[0054] The shoulder-hip-knee dynamic link model and all spatial displacement data are input into the inverse kinematics algorithm, and the optimal solution of the joint rotation angle is iteratively solved to generate a three-dimensional skeletal animation model containing a time sequence. Specifically: ①Construct a target function that minimizes the joint node position error: ;

[0055] wherein, To be solved target joint rotation angle, represent the coordinates of the visual capture of the first target skeletal joint, represent the constraint weight coefficient.

[0056] ②Calculate the Jacobian matrix of the objective function on the joint angle , and update the joint rotation angle by gradient descent iteration until convergence, get the optimized joint rotation angle, and substitute it into the shoulder-hip-knee dynamic linkage model to generate time series three-dimensional skeletal animation model frame by frame .

[0057] In an alternative way, the data fusion processing module 130 is also used to:

[0058] From the three-dimensional skeletal animation model, the real-time three-dimensional spatial coordinates and rotation angle data of the shoulder joint and hip joint of the tennis player are extracted.

[0059] Specifically: ①According to the three-dimensional skeletal animation model , the real-time three-dimensional spatial coordinates of the shoulder joint, the real-time three-dimensional spatial coordinates of the hip joint, the rotation angle data of the shoulder joint and the rotation angle data of the hip joint of the tennis player are extracted. ②Based on the shoulder-hip-knee dynamic linkage model, define the local coordinate system of the trunk. ③Convert the global rotation matrix of the shoulder joint and the hip joint to the local coordinate system of the trunk to obtain the final real-time three-dimensional spatial coordinates and rotation angle data of the shoulder joint and the hip joint.

[0060] In the acceleration phase of the tennis player's hitting action, the first rotation angle value of the shoulder joint around the vertical trunk direction and the second rotation angle value of the hip joint around the vertical trunk direction are calculated respectively in real time, and the shoulder-hip twist angle difference parameter is calculated in real time according to the difference between the first rotation angle value and the second rotation angle value.

[0061] Specifically: ①According to the change rate of the head speed in the three-dimensional skeletal animation model, determine the acceleration phase time window of the hitting action. ②In the acceleration phase of the hitting action, the first rotation angle value of the shoulder joint around the vertical trunk direction and the second rotation angle value of the hip joint around the vertical trunk direction are extracted. ③Real-time calculation of the shoulder-hip twist angle difference parameter.

[0062] The real-time feedback module 140 is configured to, when the difference between the shoulder and hip twist angles extracted from the three-dimensional skeletal animation model reaches a preset time threshold after the end of the hitting action, output a direction-identifiable vibration prompt signal through a tactile feedback device worn on the non-patella holding wrist of the tennis player, and synchronously output an overlapping comparison diagram of the three-dimensional skeletal animation model and a pre-stored standard action model on a display terminal deployed at the tennis court, and project a gradient heat map of the swing trajectory deviation on the surface of the three-dimensional skeletal animation model.

[0063] The difference between the shoulder and hip twist angles refers to the absolute difference between a first rotation angle value of the shoulder joint around the vertical trunk direction and a second rotation angle value of the hip joint around the vertical trunk direction during the acceleration phase of the hitting action. The preset time threshold refers to the upper limit of the time delay for triggering the tactile feedback after the end of the hitting action, and the value range is 0.3 seconds to 0.8 seconds (preferably 0.5 seconds).

[0064] The tactile feedback device refers to a wearable device worn on the non-patella holding wrist of the tennis player, which is internally provided with a multidirectional vibration motor. The tactile feedback device generates different vibration modes (left vibration, right vibration or continuous vibration) according to the deviation direction (such as left deviation or right deviation) of the difference between the shoulder and hip twist angles. The vibration prompt signal refers to the direction-identifiable vibration mode output by the tactile feedback device, including but not limited to: ① one-way short vibration (left vibration): indicating that the shoulder rotation lags behind; ② bidirectional alternating vibration (left-right alternating): indicating that the hip rotates excessively; ③ continuous vibration: indicating that the difference between the shoulder and hip twist angles exceeds the safety threshold.

[0065] The display terminal refers to an interactive screen or an augmented reality (AR) device deployed at the side of the tennis court. The display terminal is used to display the overlapping comparison diagram of the three-dimensional skeletal animation model and the pre-stored standard action model, and project the gradient heat map of the swing trajectory deviation on the surface of the three-dimensional skeletal animation model. The pre-stored standard action model refers to an idealized hitting action model stored in a database, which includes joint angles, joint trajectory and time sequence data obtained by professional athletes or biomechanical analysis.

[0066] The overlapping comparison diagram refers to a visual image generated by superimposing the joint trajectory (represented by a blue curve) of the three-dimensional skeletal animation model and the joint trajectory (represented by a red curve) of the pre-stored standard action model in the same three-dimensional coordinate system. The gradient heat map refers to a color mapping map generated based on the frame-by-frame Euclidean distance between the swing trajectory curve and the standard trajectory curve, which is used to quantitatively display the action deviation.

[0067] In an optional manner, the real-time feedback module 140 is specifically configured to:

[0068] acquire the pre-stored standard motion model matching the current hitting action type of the tennis player, and normalize the three-dimensional skeletal animation model with the hitting action time axis of the pre-stored standard motion model.

[0069] Specifically: ①extract the current hitting action feature vector of the tennis player from the three-dimensional skeletal animation model, determine the current hitting action type, and match the corresponding pre-stored standard motion model in the pre-stored standard motion model library. ②Divide the hitting action into acceleration period, hitting instant, and follow-up period, and perform nonlinear alignment of the time axis of the three-dimensional skeletal animation model and the standard model to ensure that the time proportions of the acceleration period, the hitting instant, and the follow-up period of the two models are consistent, and complete the normalization processing.

[0070] Superimpose the joint trajectory of the three-dimensional skeletal animation model and the joint trajectory of the pre-stored standard motion model in the same three-dimensional coordinate system, generate an overlapping comparison graph, and output and display it through the display terminal.

[0071] Among them, the joint trajectory of the three-dimensional skeletal animation model and the joint trajectory of the pre-stored standard motion model are displayed in different colors for easy superimposition and differentiation.

[0072] In an optional manner, the real-time feedback module 140 is specifically configured to:

[0073] Calculate the frame-by-frame Euclidean distance between the racquet trajectory curve of the tennis player and the corresponding standard trajectory curve of the pre-stored standard motion model, and generate a deviation distance matrix.

[0074] Among them, the trajectory point deviation distance between the racquet trajectory curve of the tennis player and the standard trajectory curve is calculated frame by frame using the Euclidean distance calculation formula, and finally a deviation distance matrix is generated.

[0075] According to the numerical distribution of the deviation distance matrix, the gradient heat map representing the racquet trajectory deviation is mapped on the surface of the three-dimensional skeletal animation model.

[0076] Among them, the numerical distribution of the deviation distance matrix is normalized and color-coded, and the color is rendered according to the position of the target skeletal joint on the surface of the three-dimensional skeletal animation model, and the gradient heat map representing the racquet trajectory deviation is mapped on the surface of the three-dimensional skeletal animation model.

[0077] The technical scheme of the embodiment acquires the motion parameters of the target joints of the tennis player and the spatial displacement data of the skeletal joints when the tennis player hits the ball through the motion sensor module and the multi-view synchronous camera module, respectively, the data fusion processing module fuses the two to generate a three-dimensional skeletal animation model containing a time sequence, and the real-time feedback module outputs a recognizable vibration prompt through the tactile feedback device when the difference between the shoulder and hip twist angles reaches a preset time threshold after the hitting action ends, and synchronously outputs an overlapping comparison graph of the three-dimensional skeletal animation model and a pre-stored standard action model and a gradient heat map of the swing trajectory deviation projected on the surface of the three-dimensional skeletal animation model. The technical scheme of the embodiment can solve the problem that the action details cannot be accurately captured and quantitative data feedback cannot be provided by relying on the naked eye observation and experience guidance of the coach in the prior art, overcome the defects of the prior art such as incomplete data collection, insufficient precision or poor adaptability, and provide strong support for the tennis player to accurately optimize the hitting action, which helps to improve the technical level of the tennis player.

[0078] Figure 2 A flowchart of an embodiment of a tennis action capture method provided by the application is shown. As shown in the figure, Figure 2 the method comprises the following steps:

[0079] S1, acquiring the motion parameters of each target joint of the tennis player in real time through an inertial measurement unit fixed to each target joint of the tennis player;

[0080] S2, synchronously acquiring the spatial displacement data of a plurality of target skeletal joints of the tennis player when the tennis player hits the ball by using a stereo vision camera arranged at the positions of the middle points of the two sidelines and the extension line of the center line of the net of the tennis court;

[0081] S3, fusing the motion parameters of each target joint and the spatial displacement data of each target skeletal joint to generate a three-dimensional skeletal animation model containing a time sequence;

[0082] S4, outputting a direction-recognizable vibration prompt signal through a tactile feedback device worn on the wrist of the non-paddle hand of the tennis player when the difference between the shoulder and hip twist angles reaches a preset time threshold after the hitting action ends, and synchronously outputting, by using a display terminal arranged on the tennis court, an overlapping comparison graph of the three-dimensional skeletal animation model and a pre-stored standard action model and a gradient heat map of the swing trajectory deviation projected on the surface of the three-dimensional skeletal animation model.

[0083] In an optional manner, all the target joints include the wrist, the elbow, the shoulder, the hip, the knee and the ankle, and the motion parameters of each target joint include the angular velocity data and the acceleration data of the corresponding target joint.

[0084] In an alternative way, the number of the stereoscopic vision cameras is three, which are respectively arranged at the midpoint of the baseline of the two sides of the tennis court and the position of the extension line of the center line of the net, and the frame rate of each stereoscopic vision camera is not less than 200 fps.

[0085] In an alternative way, S3 comprises:

[0086] The time sequence alignment is performed on all the angular velocity data and all the spatial displacement data, and the pose drift error of the installation position of each inertial measurement unit is calculated based on each acceleration data, and all the angular velocity data are dynamically compensated through a Kalman filter;

[0087] A shoulder-hip-knee dynamic link model is established according to biomechanical constraints, and the biomechanical constraints include adjacent joint rotation angle range limitation and limb length constant condition;

[0088] The shoulder-hip-knee dynamic link model and all the spatial displacement data are input into an inverse kinematics algorithm, and the optimal solution of the joint rotation angle is iteratively solved to generate a three-dimensional skeletal animation model containing time sequence.

[0089] In an alternative way, S3 further comprises:

[0090] Real-time three-dimensional spatial coordinates and rotation angle data of the shoulder joint and the hip joint of the tennis player are extracted from the three-dimensional skeletal animation model;

[0091] In the acceleration phase of the tennis player's hitting action, the first rotation angle value of the shoulder joint around the vertical trunk direction and the second rotation angle value of the hip joint around the vertical trunk direction are respectively calculated in real time, and the shoulder-hip twist angle difference parameter is calculated in real time according to the difference between the first rotation angle value and the second rotation angle value.

[0092] In an alternative way, the step of synchronously outputting the superimposed comparison graph of the three-dimensional skeletal animation model and the pre-stored standard action model by the display terminal arranged in the tennis court comprises:

[0093] The pre-stored standard action model matching the current hitting action type of the tennis player is obtained, and the three-dimensional skeletal animation model and the hitting action time axis of the pre-stored standard action model are normalized; wherein the pre-stored standard action model contains time sequence motion data of each target joint in an ideal state;

[0094] The joint node trajectory of the three-dimensional skeletal animation model and the joint node trajectory of the pre-stored standard action model are superimposed in the same three-dimensional coordinate system to generate a superimposed comparison graph and output and display through the display terminal.

[0095] In an alternative manner, the step of projecting the gradient heat map of the swing trajectory deviation on the surface of the three-dimensional skeletal animation model comprises:

[0096] calculating the frame-by-frame Euclidean distance between the swing trajectory curve of the tennis player and the standard trajectory curve corresponding to the pre-stored standard motion model, to generate a deviation distance matrix;

[0097] According to the numerical distribution of the deviation distance matrix, the gradient heat map representing the swing trajectory deviation is mapped on the surface of the three-dimensional skeletal animation model.

[0098] The technical scheme of the embodiment can solve the problem that the traditional method of relying on the naked eye observation and experience guidance of the coach cannot accurately capture the action details and provide quantitative data feedback, and overcome the defects of the existing motion capture devices, such as incomplete data collection, insufficient precision, or poor adaptability, thereby providing strong support for the accurate optimization of the swing action of the player and helping to improve the technical level of the player.

[0099] An electronic device according to an embodiment of the present application includes a memory, a processor, and a computer program stored on the memory and executable on the processor, and the processor implements any of the above tennis motion capture methods when executing the computer program. That is, an electronic device according to an embodiment of the present application can include but is not limited to a processor and a memory; the memory is configured to store a computer program; and the processor is configured to execute the tennis motion capture method shown in any of the embodiments of the present application by calling the computer program.

[0100] A computer readable storage medium according to an embodiment of the present application has a computer program stored thereon, and the computer program is executable on a processor to implement any of the above tennis motion capture methods.

[0101] Alternatively, the computer readable storage medium can be a read-only memory (ROM), a random access memory (RAM), a compact disc read-only memory (CD-ROM), a magnetic tape, a floppy disk, and an optical data storage device, etc.

[0102] In an example embodiment, a computer program product or computer program is also provided, which includes computer instructions stored in a computer readable storage medium. A processor of an electronic device reads the computer instructions from the computer readable storage medium, and the processor executes the computer instructions, so that the electronic device performs the above tennis motion capture method.

[0103] It should be understood that the flow and block diagrams in the drawings show the possible architectural, functional, and operational architectures of methods and computer program products according to various embodiments of the present application. In this regard, each block in the flow and block diagrams can represent a module, a segment, or a portion of code that comprises one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions noted in the blocks can occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and / or flowchart illustrations, and combinations thereof, can be implemented by a dedicated hardware-based system that performs the specified functions or operations, or combinations of hardware and software.

[0104] The computer readable storage medium of the embodiments of the present application can be, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of the computer readable storage medium can include, but are not limited to, an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the present application, the computer readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus, or device.

[0105] The computer readable storage medium described above carries one or more programs, which, when executed by the electronic device, cause the electronic device to perform the method shown in the above embodiments.

[0106] The above description is only the preferred embodiment of the present application and the explanation of the applied technical principles. It should be understood by those skilled in the art that the disclosed range of the present application is not limited to the technical solutions formed by the specific combination of the above technical features, and should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the disclosed concept. For example, the technical solutions formed by the mutual replacement of the above features and the technical features with similar functions disclosed in the present application (but not limited to) can be used.

[0107] It should be noted that the terms "first", "second" and the like in the specification and claims of the present application are used to distinguish similar objects, and represent a specific order or sequence. The order of use of similar objects can be interchanged in appropriate cases, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described.

[0108] Those skilled in the art know that the present application can be implemented as a system, a method or a computer program product, so the present application can be specifically implemented as follows: it can be a complete hardware, a complete software (including firmware, resident software, microcode, etc.), or a combination of hardware and software, which is generally referred to as "circuit", "module" or "system" in this paper. In addition, in some embodiments, the present application can also be implemented as a computer program product in one or more computer readable media, which contains computer readable program code.

[0109] Although the embodiments of the present application have been shown and described above, it should be understood that the above embodiments are exemplary and cannot be understood as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.

Claims

1. A tennis motion capture system characterized by, The system comprises: a motion sensor module comprising an inertial measurement unit fixed to each target joint of a tennis player, for collecting motion parameters of each target joint of the tennis player in real time; a multi-view synchronous camera module comprising stereoscopic vision cameras arranged at the midpoint of the baseline on both sides of the tennis court and the position of the extension line of the net centerline, a plurality of stereoscopic vision cameras forming a three-dimensional capture area covering the tennis player's hitting action, for synchronously acquiring spatial displacement data of a plurality of target skeletal joint nodes when the tennis player hits the ball; a data fusion processing module for fusing the motion parameters of each target joint and the spatial displacement data of each target skeletal joint node to generate a three-dimensional skeletal animation model containing a time sequence; a real-time feedback module for outputting a direction-identifiable vibration prompt signal through a tactile feedback device worn on the non-paddle-wielding wrist of the tennis player when a preset time threshold is reached after the hitting action is completed according to the shoulder-hip twist angle difference parameter extracted from the three-dimensional skeletal animation model, and synchronously outputting an overlapping comparison graph of the three-dimensional skeletal animation model and a pre-stored standard action model on a display terminal deployed on the tennis court, and projecting a gradient heat map of the swing trajectory deviation on the surface of the three-dimensional skeletal animation model; The data fusion processing module is specifically used for: time sequence alignment of all angular velocity data and all spatial displacement data, and based on each acceleration data, respectively calculating the pose drift error of the installation position of each inertial measurement unit, and dynamically compensating all angular velocity data through a Kalman filter; establishing a shoulder-hip-knee dynamic link model according to biomechanical constraints, the biomechanical constraints including adjacent joint rotation angle range limitation and limb length constant condition; inputting the shoulder-hip-knee dynamic link model and all spatial displacement data into an inverse kinematics algorithm to iteratively solve the optimal solution of the joint rotation angle and generate a three-dimensional skeletal animation model containing a time sequence; The data fusion processing module is further used for: extracting real-time three-dimensional spatial coordinates and rotation angle data of the shoulder joint and the hip joint of the tennis player from the three-dimensional skeletal animation model; in the acceleration phase of the tennis player's hitting action, respectively calculating a first rotation angle value of the shoulder joint around the vertical trunk direction and a second rotation angle value of the hip joint around the vertical trunk direction in real time, and calculating the shoulder-hip twist angle difference parameter in real time according to the difference between the first rotation angle value and the second rotation angle value.

2. The tennis motion capture system of claim 1, wherein, All target joints include: wrist, elbow, shoulder, hip, knee and ankle; the motion parameters of each target joint include: angular velocity data and acceleration data of the corresponding target joint.

3. The tennis motion capture system of claim 1, wherein, The number of stereoscopic vision cameras is three, respectively arranged at the midpoint of the baseline on both sides of the tennis court and the position of the extension line of the net centerline, and the frame rate of each stereoscopic vision camera is not less than 200 fps.

4. The tennis motion capture system of claim 1, wherein, The real-time feedback module is specifically used for: acquire the pre-stored standard motion model matching the current hitting action type of the tennis player, and normalize the three-dimensional skeletal animation model with a hitting action time axis of the pre-stored standard motion model; wherein the pre-stored standard motion model contains time sequence motion data of each target joint in an ideal state; superimpose the joint node track of the three-dimensional skeletal animation model and the joint node track of the pre-stored standard motion model in the same three-dimensional coordinate system to generate an overlapping comparison graph and output and display the overlapping comparison graph through the display terminal.

5. The tennis motion capture system of claim 4, wherein, The real-time feedback module is specifically used for: calculating frame-by-frame Euclidean distances between a racquet swing track curve of the tennis player and a standard track curve corresponding to the pre-stored standard motion model to generate a deviation distance matrix; according to a numerical distribution of the deviation distance matrix, mapping the gradient heat map representing racquet swing track deviation on a surface of the three-dimensional skeletal animation model.

6. A tennis motion capture method characterized by, The method comprises: real-time collection of motion parameters of each target joint of the tennis player through an inertial measurement unit fixed to each target joint of the tennis player; synchronous acquisition of spatial displacement data of multiple target skeletal joint nodes of the tennis player when hitting a ball through stereovision cameras arranged at positions of midpoints of two sidelines and a centerline extension of a net of a tennis court; fusion of the motion parameters of each target joint and the spatial displacement data of each target skeletal joint node to generate a three-dimensional skeletal animation model containing a time sequence; when a preset time threshold is reached after the hitting action ends, outputting a direction-identifiable vibration prompt signal through a tactile feedback device worn on a wrist of a non-racquet-holding hand of the tennis player, and synchronously outputting, through a display terminal deployed at the tennis court, an overlapping comparison graph of the three-dimensional skeletal animation model and a pre-stored standard motion model, and a gradient heat map of racquet swing track deviation projected on a surface of the three-dimensional skeletal animation model; The step of fusing the motion parameters of each target joint and the spatial displacement data of each target skeletal joint node to generate a three-dimensional skeletal animation model containing a time sequence comprises: time sequence alignment of all angular velocity data and all spatial displacement data, calculation of pose drift errors of installation positions of each inertial measurement unit based on each acceleration data, and dynamic compensation of all angular velocity data through a Kalman filter; establishment of a shoulder-hip-knee dynamic linkage model according to biomechanical constraints, wherein the biomechanical constraints include adjacent joint rotation angle range limitations and limb length constant conditions; input of the shoulder-hip-knee dynamic linkage model and all spatial displacement data into an inverse kinematics algorithm to iteratively solve joint rotation angle optimal solutions and generate a three-dimensional skeletal animation model containing a time sequence; further comprising: extraction of real-time three-dimensional space coordinates and rotation angle data of shoulder joints and hip joints of the tennis player from the three-dimensional skeletal animation model; In the acceleration phase of the tennis player's hitting action, the first rotation angle value of the shoulder joint around the vertical trunk direction and the second rotation angle value of the hip joint around the vertical trunk direction are calculated in real time, respectively, and the shoulder-hip torsion angle difference parameter is calculated in real time according to the difference between the first rotation angle value and the second rotation angle value.

7. The tennis motion capture method according to claim 6, wherein, All target joints include: wrist, elbow, shoulder, hip, knee and ankle; the motion parameters of each target joint include: angular velocity data and acceleration data of the corresponding target joint.

8. The tennis motion capture method according to claim 6, wherein, The number of the stereoscopic vision cameras is three, which are respectively arranged at the midpoint of the baseline of the two sides of the tennis court and the position of the extension line of the center line of the net, and the frame rate of each stereoscopic vision camera is not less than 200 fps.

Images