Real-time analysis method for shot put action based on kinematics modeling

Abstract

The present application relates to shot analysis technical field, especially to a kind of shot throwing action real-time analysis method based on kinematics modeling.The action video information of user is acquired;Several key nodes of user are identified;The estimated landing site of shot is determined;The light distortion influence state is identified based on distortion influence quantity;When the current is strong light distortion influence state, each key node in action video information is corrected based on fixed point position offset vector;Whether the denoising standard for each intermediate node in determining each running track is corrected based on the drift oscillation of each key node;The distance between key nodes is corrected based on the dynamic elastic quantity of key node.When there is deviation between estimated landing site and actual landing site, the key node used to determine estimated landing site is corrected based on light condition, and the shot action analysis efficiency is improved.

Description

Technical Field

[0001] This invention relates to the field of shot put analysis technology, and in particular to a real-time analysis method for shot put throwing motion based on kinematic modeling. Background Technology

[0002] Against the backdrop of AI-driven sports development, using AI algorithms to calculate athletic performance and evaluate overall results has become a growing trend. For shot put, traditional methods of manually measuring distance and analyzing results are difficult to implement objectively, accurately, and efficiently.

[0003] Chinese Patent Publication No. CN119625055B discloses a method for analyzing the three-dimensional spatial trajectory and predicting the score of a shot put based on 2D images. The method includes establishing a 3D spatial curve model and proposing a projection cost function to calculate the difference between the 3D spatial curve and points on the 2D image plane. An optimization algorithm is then used to correct the 3D spatial curve based on this difference, ensuring that the 2D projection of the 3D spatial curve coincides with the shot put position across multiple frames. Furthermore, the method adaptively calculates the landing point position by fusing the first bounce curve and the rolling trajectory, and automatically calculates the physical distance and pixel relationship using the gravity formula to automatically calculate the actual landing physical distance. However, the above technical solution has the following problems: it does not consider the key nodes for determining the estimated landing point based on lighting conditions when there is a deviation between the estimated and actual landing points, thus affecting the efficiency of shot put motion analysis. Summary of the Invention

[0004] To address this issue, the present invention provides a real-time analysis method for shot put throwing motion based on kinematic modeling. This method overcomes the problem in existing technologies that do not consider the key nodes for determining the estimated landing point based on lighting conditions when there is a deviation between the estimated and actual landing points, thus affecting the efficiency of shot put motion analysis.

[0005] To achieve the above objectives, this invention provides a real-time analysis method for shot put throwing motion based on kinematic modeling, comprising: Obtain user's action video information; Identify several key nodes of the user, including joint nodes and several intermediate nodes located between the joint nodes to correct the joint nodes. Based on the trajectory of each key node in each frame of the motion video information, the estimated landing point of the shot put is determined. The distance deviation between the estimated landing point and the actual landing point is used to determine whether the throwing analysis is qualified, so that when the throwing analysis is found to be abnormal, the influence of illumination distortion can be identified based on the amount of distortion. When it is determined that the current state is affected by strong light distortion, the key nodes in the motion video information are corrected based on the fixed point offset vector. The noise reduction standard for each intermediate node is determined based on the drift and oscillation of each key node to determine whether correction is needed and to determine each running trajectory. Dynamic elasticity correction of the distance between key nodes based on key nodes.

[0006] Furthermore, the process of determining whether the throwing analysis is qualified based on the distance deviation between the estimated landing point and the actual landing point includes: When the distance deviation is less than or equal to the preset distance deviation, the current throw analysis is determined to be qualified, and the analysis of a single user is continued to be completed using the current analysis parameters; When the distance deviation is greater than the preset distance deviation, the current throwing analysis is determined to be abnormal, and the illumination distortion influence status is identified based on the distortion influence amount.

[0007] Furthermore, the process of identifying the illumination distortion effect state based on the distortion effect amount includes: When the distortion effect exceeds the preset distortion effect, the current state is determined to be under strong light distortion. When the distortion effect is less than or equal to the preset distortion effect, the current state is determined to be weak light distortion effect.

[0008] Furthermore, the process of correcting key nodes in motion video information based on fixed-point offset vectors includes: Obtain the standard coordinates of each pre-stored fixed point in the standard coordinate system; Identify the current coordinates of each fixed point in a single frame image; The offset vector of each fixed point is determined based on the difference between the current coordinates of each fixed point and the coordinates of the standard point. The global offset vector of a single frame image is determined based on the offset vector of each fixed point. The global offset vector is added to the coordinates of each key node to obtain the corrected key nodes.

[0009] Furthermore, the process of determining the drift oscillations at each key node includes, Determine the width of each user's area by identifying key nodes; For a single region, the difference between the maximum and minimum values ​​in each frame of the image is determined to obtain the region width change value; The drift oscillation is obtained by calculating the average value of the width change in each region.

[0010] Furthermore, the process of determining whether to correct the denoising standard based on the drift oscillation includes, When the drift oscillation amount is less than or equal to the preset drift oscillation amount, the current denoising standard is continuously used to process the motion video information; When the drift oscillation is greater than the preset drift oscillation, the noise reduction standard is corrected based on the drift oscillation.

[0011] Furthermore, based on the drift oscillation correction and denoising standard, where, The reduction in the cutoff frequency of a low-pass filter is positively correlated with the amount of drift oscillation.

[0012] Furthermore, the process of dynamically adjusting the distance between key nodes based on the elasticity of key nodes includes: The rate of change of distance between the sternal marker and the acromion marker is determined so that when the sequence of the rate of change of distance conforms to a preset sudden step change, the current state of force exertion is determined. The effective arm length of the upper arm is corrected based on the dynamic elasticity, and the compensation amount of the effective arm length of the upper arm is positively correlated with the dynamic elasticity.

[0013] Furthermore, the distortion effect is the average value of the brightness response values ​​in each frame of the image.

[0014] Furthermore, the preset sudden step change includes a distance change rate greater than a preset change threshold and a maximum distance change greater than a preset maximum change.

[0015] Compared with existing technologies, the beneficial effects of this invention are that it judges the qualification of the throw analysis based on the distance deviation between the estimated landing point and the actual landing point. If the deviation between the estimated and actual landing points is large, it indicates that there is an error in the analysis process, and further investigation is needed to determine the source of the error. The preset distance deviation characterizes the acceptable range of analysis error for the system. If the key node is mispositioned due to lighting, soft tissue oscillation, or recognition deviation, errors will occur in the release speed, release angle, and release height, which will ultimately be amplified into landing point deviation. Therefore, the analysis anomaly is identified at the result level through landing point deviation. The analysis results are closed-loop verified by the actual landing point, preventing the system from continuing to output erroneous analysis results when the key node recognition is already abnormal, improving the reliability of real-time analysis, and further improving the efficiency of shot put motion analysis.

[0016] Furthermore, under the influence of strong lighting distortion, key nodes are corrected based on the fixed-point offset vector to correct the overall coordinate shift caused by strong light. The fixed point is a stable point in the field. The standard point coordinates represent the theoretical position of the fixed point in the standard coordinate system. The current point coordinates represent the position of the fixed point identified in the current video frame. The fixed-point offset vector represents the deviation of the fixed point in the current image relative to the standard position. Strong lighting can cause local overexposure of fixed markers, causing the center of the point obtained by the image recognition algorithm to shift. If the fixed points show a consistent overall directional shift, the key human body nodes will be affected by similar image distortion. Therefore, the offset of the fixed points can be used to reverse-correct the coordinates of the key nodes. Mapping correction using the global offset vector under strong lighting reduces the impact of image distortion on the coordinates of the key nodes, improves the accuracy of the key node positions, and further reduces the errors in the prediction of release speed, joint angular velocity, and landing point. This further improves the efficiency of shot put motion analysis.

[0017] Furthermore, a denoising standard for correcting intermediate node trajectories is determined based on the drift oscillation magnitude of key nodes to address the key node drift problem caused by high-frequency soft tissue oscillations during the explosive force phase of the shot put. The drift oscillation magnitude is used to quantify the high-frequency width changes of the user's body surface area during the throwing process. The region width can be determined by paired key nodes and may include shoulder width, chest width, hip width, and upper arm region width. The region width change value is used to characterize the amplitude of width fluctuation of a certain region in consecutive frames. The larger the drift oscillation magnitude, the more obvious the deformation and shaking of the human body surface. The explosive rotation and braking during a shot put throw cause high-frequency oscillations in the soft tissues of the torso, shoulders, and upper arms. Visual perception captures the body surface location, not the center of bony joints; therefore, the rapid oscillation of the soft tissues manifests as high-frequency jitter in the key node coordinates. Intermediate nodes are located on the limb surface and are more susceptible to soft tissue oscillations; therefore, the denoising standard is adjusted for the intermediate node trajectory. The system adaptively adjusts the trajectory denoising intensity based on different users, force intensities, and soft tissue oscillation levels, reducing interference from high-frequency non-bone motion noise on motion analysis. This further improves the efficiency of shot put motion analysis.

[0018] Furthermore, dynamic elasticity correction of the distance between key nodes is used to address the shift in shoulder joint rotation center estimation caused by thoracic expansion and acromion abduction during the power exertion phase. Dynamic elasticity characterizes the degree of non-rigid displacement of key surface nodes relative to the bony joint center during the power exertion phase, corresponding to the apparent displacement of surface markers due to thoracic expansion, muscle bulging, and soft tissue deformation during power exertion. The change in distance between the sternal and acromion markers reflects the non-rigid deformation of the chest and shoulder region during the power exertion phase. Under normal circumstances, this distance does not increase dramatically in a short period; however, during the power exertion phase in shot putters, the increased intrathoracic pressure, thoracic expansion, and tense abduction of shoulder muscles lead to a rapid increase in the apparent distance between the sternal and acromion markers. Visual recognition identifies surface markers or feature points, while the shoulder joint rotation center is the motion center corresponding to the bony structure. When surface markers shift outward during the power generation phase, directly estimating the shoulder joint center using key surface nodes will cause the shoulder joint rotation center to shift forward or outward, affecting the calculation of the effective arm length, shoulder joint angular velocity, and release velocity contribution. By identifying the sudden increase in non-rigid deformation in the chest and shoulder region during the power generation phase, compensation is made for the shoulder joint rotation center and the effective arm length, reducing the impact of surface deformation on kinematic analysis results and improving the accuracy of shoulder joint motion parameters and shot put release parameters. This further improves the efficiency of shot put motion analysis. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating the steps of the real-time analysis method for shot put throwing motion based on kinematic modeling, as described in an embodiment of the present invention. Figure 2 This is a logic diagram for determining whether a throw analysis is qualified based on distance deviation, according to an embodiment of the present invention. Detailed Implementation

[0020] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0021] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0022] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0023] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0024] Please see Figure 1 and Figure 2 The diagrams shown are a flowchart of the steps in the real-time analysis method for shot put throwing motion based on kinematic modeling according to an embodiment of the present invention, and a logic determination diagram for determining whether the throwing analysis is qualified based on distance deviation. The real-time analysis method for shot put throwing motion based on kinematic modeling according to an embodiment of the present invention includes: S1, acquire the user's motion video information; S2, identify several key nodes of the user, including joint nodes and several intermediate nodes located between each joint node to correct the joint nodes. S3, based on the running trajectory of each key node in each frame of the motion video information, combined with the calibration parameters of the camera device and the coordinate calibration relationship of the throwing site, the estimated landing point of the shot put is determined; S4. Determine whether the throwing analysis is qualified based on the distance deviation between the estimated landing point and the actual landing point, so as to identify the illumination distortion influence state based on the distortion influence amount when the throwing analysis is found to be abnormal. S5, when it is determined that the current state is affected by strong light distortion, correct each key node in the motion video information based on the fixed point offset vector; S6, based on the drift and oscillation of each key node, determine whether to correct the noise reduction standard for each intermediate node during the process of determining each running trajectory; among them, low-pass filtering is used to denoise the intermediate node trajectory, and the cutoff frequency of the low-pass filter is adjusted according to the drift and oscillation to reduce the drift of intermediate nodes caused by the soft tissue oscillation of the user's torso. S7, based on the dynamic elasticity of key nodes, corrects the distance between key nodes.

[0025] Specifically, in S2, the method for correcting joint nodes through intermediate nodes includes: performing a consistency check on the position of the joint node based on the geometric constraint relationship between adjacent joint nodes and intermediate nodes; determining that a single joint node has an abnormal offset when the position change of the joint node in consecutive frames exceeds a preset motion threshold; for the abnormally offset joint node, obtaining the time-predicted position based on the motion continuity between consecutive frames, obtaining the geometrically inverse position based on the proportional relationship of the intermediate node on adjacent limbs, and obtaining the length-constrained position based on the standard length of adjacent limbs; and determining the corrected joint node position based on the time-predicted position, the geometrically inverse position, and the length-constrained position.

[0026] Specifically, in S3, the process of determining the estimated landing point of the shot put includes: based on the filtered key node trajectory, determining the frame where the wrist node velocity reaches a local peak and the elbow and shoulder joint angular velocities exceed the corresponding thresholds as the release frame; calculating the release velocity vector based on the coordinate changes of the wrist joint nodes near the release frame in the field coordinate system, and determining the release angle and horizontal velocity based on the release velocity vector; calculating the shot put flight time based on the release height, release velocity, and gravity projectile motion model, thereby obtaining the estimated landing point of the shot put.

[0027] Specifically, the process of determining whether a throw analysis is qualified based on the distance deviation between the estimated landing point and the actual landing point includes: When the distance deviation is less than or equal to the preset distance deviation, the current throw analysis is determined to be qualified, and the analysis of a single user is continued to be completed using the current analysis parameters; When the distance deviation is greater than the preset distance deviation, the current throwing analysis is determined to be abnormal, and the illumination distortion influence status is identified based on the distortion influence amount.

[0028] The preset distance deviation is determined based on the error distribution between the estimated landing point and the actual landing point in the calibrated throwing sample. The preset distance deviation can be the mean of the distance error between the estimated landing point and the actual landing point in the calibrated throwing sample. The calibrated throwing sample is a video sample of the throwing action collected under the condition that the camera device calibration, throwing site coordinate calibration and actual landing point marking have been completed. The calibrated throwing sample includes action video information, key node identification information, actual landing point information and estimated landing point information obtained based on the current parsing parameters.

[0029] Specifically, the process of identifying the state of illumination distortion based on the amount of distortion influence includes: The average value of the brightness response in each frame image is determined as the distortion effect. When the distortion effect exceeds the preset distortion effect, the current state is determined to be under strong light distortion. When the distortion effect is less than or equal to the preset distortion effect, the current state is determined to be weak light distortion effect.

[0030] In a single embodiment, the luminance response value is the average gray level of each pixel in a single frame of the image. The greater the distortion effect, the more abundant the ambient light at the time of image acquisition, and the greater the risk of overexposure.

[0031] Specifically, strong light shining on the human body surface, clothing, ground, or marked points produces high-intensity reflection, causing the corresponding pixels in the image sensor to receive too many photons, exceeding the charge capacity of the photosensitive unit, resulting in overexposure in local areas and loss of detail.

[0032] In a single embodiment, the preset distortion effect is the average distortion effect in the normal illumination calibration sample.

[0033] Specifically, the process of correcting key nodes in motion video information based on fixed-point offset vectors includes: Obtain the standard coordinates of each pre-stored fixed point in the standard coordinate system; Identify the current coordinates of each fixed point in a single frame image; The offset vector of each fixed point is determined based on the difference between the current coordinates of each fixed point and the coordinates of the standard point. The global offset vector of a single frame image is determined based on the offset vector of each fixed point. The global offset vector is added to the coordinates of each key node to obtain the corrected key nodes.

[0034] In a single embodiment, the process of correcting key nodes may further include fitting a spatial correction model based on the current point coordinates and standard point coordinates of each fixed point, and mapping and correcting the coordinates of each key node based on the spatial correction model to obtain the corrected key nodes.

[0035] Specifically, the process of determining the drift oscillations at each key node includes, Determine the width of each user's area by identifying key nodes; For a single region, the difference between the maximum and minimum values ​​in each frame of the image is determined to obtain the region width change value; The drift oscillation is obtained by calculating the average value of the width change in each region.

[0036] The impact of high-frequency soft tissue oscillations on the throwing motion during the explosive spin phase of shot put throwing was analyzed by quantifying drift oscillations. A larger drift oscillation magnitude indicates more pronounced changes in the apparent width of certain areas of the user's body due to high-frequency soft tissue oscillations during the throwing process. A larger drift oscillation magnitude also indicates stronger high-frequency non-skeletal motion noise in the trajectory at key nodes.

[0037] Specifically, the process of determining whether to correct the denoising standard based on the amount of drift oscillation includes, When the drift oscillation amount is less than or equal to the preset drift oscillation amount, the current denoising standard is continuously used to process the motion video information; When the drift oscillation is greater than the preset drift oscillation, the noise reduction standard is corrected based on the drift oscillation.

[0038] In a single embodiment, the preset drift oscillation amount can be the average drift oscillation amount of the user in the calibrated action sample, in order to classify the impact of soft tissue oscillation.

[0039] Specifically, the noise reduction standard is based on drift oscillation correction, where... The reduction in the cutoff frequency of a low-pass filter is positively correlated with the amount of drift oscillation.

[0040] The greater the drift oscillation, the more pronounced the high-frequency noise, and the stronger the filtering should be, i.e., lowering the cutoff frequency. A lower cutoff frequency in a low-pass filter results in a stronger filter and a smoother trajectory.

[0041] Specifically, the process of dynamically adjusting the distance between key nodes based on the elasticity of key nodes includes: The rate of change of distance between the sternal marker and the acromion marker is determined so that when the sequence of the rate of change of distance conforms to a preset sudden step change, the current state of force exertion is determined. The effective arm length of the upper arm is corrected based on the dynamic elasticity, and the compensation amount of the effective arm length of the upper arm is positively correlated with the dynamic elasticity.

[0042] The greater the dynamic elasticity, the more significant the apparent increase in distance between the sternal landmark and the acromion landmark, corresponding to a greater amplitude of thoracic expansion and a greater degree of soft tissue deformation during the force exertion phase. The greater the effective arm length compensation, the more adequate the compensation for the more significant the estimated shift in the shoulder joint rotation center caused by thoracic expansion and non-rigid deformation of the soft tissue during the force exertion phase.

[0043] The sequence of distance change rates between sternal and acromion landmarks is determined based on the ratio of the change in distance between sternal and acromion landmark nodes in adjacent frames to the time interval between adjacent frames. Preset sudden increase step change includes distance change rate greater than preset change threshold and maximum distance change greater than preset maximum change.

[0044] The dynamic elasticity is the average of the maximum distance changes in each video frame that conforms to the preset sudden step change.

[0045] The preset change threshold is determined based on the distance change rate sequence during the non-force-generating phase of the calibration motion sample. The distance change rate sequence between the sternal marker node and the acromion marker node during the non-force-generating phase of the calibration motion sample is obtained, the mean of the distance change rate sequence is calculated, and this mean is determined as the preset change threshold.

[0046] The preset maximum change is determined based on the maximum distance change during the non-force-generating phase of the calibrated motion sample. The maximum change in distance between the sternal marker node and the acromion marker node relative to the baseline distance during the non-force-generating phase of the calibrated motion sample is obtained, and the average of this maximum change is determined as the preset maximum change. The preset change threshold and the preset maximum change are used to identify sudden, non-rigid deformation states in the chest and shoulder region during the force-generating phase; specifically, the preset change threshold is used to identify whether the distance change rate increases abnormally, and the preset maximum change is used to identify whether the distance change amplitude increases abnormally.

[0047] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

[0048] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A real-time analysis method for shot put throwing motion based on kinematic modeling, characterized in that, include: Obtain user's action video information; Identify several key nodes of the user, including joint nodes and several intermediate nodes located between the joint nodes to correct the joint nodes. Based on the trajectory of each key node in each frame of the motion video information, the estimated landing point of the shot put is determined. The distance deviation between the estimated landing point and the actual landing point is used to determine whether the throwing analysis is qualified, so that when the throwing analysis is found to be abnormal, the influence of illumination distortion can be identified based on the amount of distortion. When it is determined that the current state is affected by strong light distortion, the key nodes in the motion video information are corrected based on the fixed point offset vector. The noise reduction standard for each intermediate node is determined based on the drift and oscillation of each key node to determine whether correction is needed and to determine each running trajectory. Dynamic elasticity correction of the distance between key nodes based on key nodes.

2. The real-time analysis method for shot put throwing motion based on kinematic modeling according to claim 1, characterized in that, The process of determining whether a throw analysis is qualified based on the distance deviation between the estimated landing point and the actual landing point includes: When the distance deviation is less than or equal to the preset distance deviation, the current throw analysis is determined to be qualified, and the analysis of a single user is continued to be completed using the current analysis parameters; When the distance deviation is greater than the preset distance deviation, the current throwing analysis is determined to be abnormal, and the illumination distortion influence status is identified based on the distortion influence amount.

3. The real-time analysis method for shot put throwing motion based on kinematic modeling according to claim 2, characterized in that, The process of identifying the state of illumination distortion based on the amount of distortion influence includes: When the distortion effect exceeds the preset distortion effect, the current state is determined to be under strong light distortion. When the distortion effect is less than or equal to the preset distortion effect, the current state is determined to be weak light distortion effect.

4. The real-time analysis method for shot put throwing motion based on kinematic modeling according to claim 3, characterized in that, The process of correcting key nodes in motion video information based on fixed-point offset vectors includes: Obtain the standard coordinates of each pre-stored fixed point in the standard coordinate system; Identify the current coordinates of each fixed point in a single frame image; The offset vector of each fixed point is determined based on the difference between the current coordinates of each fixed point and the coordinates of the standard point. The global offset vector of a single frame image is determined based on the offset vector of each fixed point. The global offset vector is added to the coordinates of each key node to obtain the corrected key nodes.

5. The real-time analysis method for shot put throwing motion based on kinematic modeling according to claim 4, characterized in that, The process of determining the drift oscillations at each key node includes, Determine the width of each user's area by identifying key nodes; For a single region, the difference between the maximum and minimum values ​​in each frame of the image is determined to obtain the region width change value; The drift oscillation is obtained by calculating the average value of the width change in each region.

6. The real-time analysis method for shot put throwing motion based on kinematic modeling according to claim 5, characterized in that, The process of determining whether to correct the noise reduction standard based on drift oscillation includes, When the drift oscillation amount is less than or equal to the preset drift oscillation amount, the current denoising standard is continuously used to process the motion video information; When the drift oscillation is greater than the preset drift oscillation, the noise reduction standard is corrected based on the drift oscillation.

7. The real-time analysis method for shot put throwing motion based on kinematic modeling according to claim 6, characterized in that, The noise reduction standard is based on drift oscillation correction, where... The reduction in the cutoff frequency of a low-pass filter is positively correlated with the amount of drift oscillation.

8. The real-time analysis method for shot put throwing motion based on kinematic modeling according to claim 7, characterized in that, The process of dynamically adjusting the distance between key nodes based on key nodes includes: The rate of change of distance between the sternal marker and the acromion marker is determined so that when the sequence of the rate of change of distance conforms to a preset sudden step change, the current state of force exertion is determined. The effective arm length of the upper arm is corrected based on the dynamic elasticity, and the compensation amount of the effective arm length of the upper arm is positively correlated with the dynamic elasticity.

9. The real-time analysis method for shot put throwing motion based on kinematic modeling according to claim 8, characterized in that, The distortion effect is the average value of the brightness response values ​​in each frame of the image.

10. The real-time analysis method for shot put throwing motion based on kinematic modeling according to claim 9, characterized in that, Preset sudden increase step change includes distance change rate greater than preset change threshold and maximum distance change greater than preset maximum change.

Images