System and method for collecting shooting exercise amount based on image recognition and backboard induction
By adopting image recognition and induction technology in the basketball training system, combined with the Karman filtering and dynamic adjustment module, the problem of tracking the inaccurate tracing of basketball in the existing technology and the inadequate training equipment adjustment And personalized training support.
Patent Information
- Application Number
- CN202411429302.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-14
- Publication Date
- 2025-01-24
AI Technical Summary
The prior art cannot track basketball dynamics stably and accurately under variable lighting conditions, and the accuracy and stability of feature point matching are insufficient, especially when the background is complex or the basketball movement speed is fast, and it is prone to errors. It is not possible to adjust the training equipment in real time. To adapt to athletes’ immediate performance, lack of scientific data support, and no integrated environmental monitoring affects training results and athlete performance.
Adopt a system based on image recognition and induction rebounds, through multi -layered image recognition algorithms and improved scale unchanging feature change algorithms, combined with high dynamic range imaging technology, extract feature points and estimate the basketball trajectory. And inductive data, generate personalized training suggestions, and dynamically adjust the setting of training equipment to adapt to the real -time performance of athletes.
It achieves accurate tracking of basketball position and trajectory under changing lighting conditions, improves the accuracy and stability of feature point matching, provides real-time training feedback and personalized training suggestions, enhances the scientific nature and effect of training, and ensures The settings of training equipment are always optimized to support optimal training results.
Smart Images
Figure CN119359758A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of image processing technology, and in particular to a system and method for collecting shooting motion based on image recognition and sensing backboard. Background Art
[0002] Basketball training and performance analysis is a highly specialized field that involves monitoring and improving everything from basic skill training to advanced competitive performance. With advances in technology, especially in image processing and sensor technology, more and more solutions are being proposed to improve training effectiveness and athlete performance. However, despite the many technologies and methods available, many basketball training systems still face challenges in improving accuracy, real-time feedback, and personalized training.
[0003] In traditional training, coaches usually rely on observation to guide athletes, which is limited in accuracy and consistency and may not be sufficient when analyzing athletes' performance in different environmental conditions. With the scientificization of competitive sports, the demand for higher accuracy and real-time analysis tools has increased. The limitations of existing technologies include the difficulty of tracking basketball movements in dynamic and complex environments, and the inadequacy in providing real-time, actionable feedback.
[0004] The existing technology has at least the following technical problems: the existing technology cannot stably and accurately track the dynamics of the basketball under changing lighting conditions, performs poorly in terms of the accuracy and stability of feature point matching, especially when the background is complex or the basketball moves at a fast speed, is prone to errors, fails to adjust the training equipment in real time to adapt to the athlete's immediate performance, and lacks scientific data support; and the existing technology often does not integrate environmental monitoring, which affects the training effect and the athlete's performance. Summary of the invention
[0005] The present invention provides a system and method for collecting shooting motion based on image recognition and sensing backboard to solve the technical problems that the prior art cannot stably and accurately track the dynamics of a basketball under variable lighting conditions, performs poorly in terms of accuracy and stability of feature point matching, is prone to errors especially when the background is complex or the basketball moves at a fast speed, fails to adjust training equipment in real time to adapt to the athlete's immediate performance, lacks scientific data support; and the prior art often does not integrate environmental monitoring, which affects the training effect and the athlete's performance.
[0006] The system and method of collecting shooting motion based on image recognition and sensing backboard of the present invention specifically include the following technical solutions: The system for collecting shooting motion based on image recognition and sensing the backboard includes the following parts: Data acquisition module, wireless module, image recognition module, basketball tracking module, training suggestion module, dynamic adjustment module, feedback adjustment module; The data acquisition module is used to capture the dynamics of the basketball and collect raw data. The data acquisition module transmits the raw data through the wireless module; The wireless module is used to transmit the collected raw data to the image recognition module and the basketball tracking module; The image recognition module designs a multi-level image recognition algorithm to obtain a high dynamic range image, uses an improved scale-invariant feature transformation algorithm to extract feature points from the high dynamic range image, and the image recognition module is connected to the basketball tracking module; The basketball tracking module estimates the basketball motion trajectory based on the changes of feature points using the optical flow method, and uses the Kalman filter to integrate the basketball motion trajectory and the backboard sensing data in the original data to obtain the basketball state estimation; the basketball tracking module is connected to the training suggestion module; The training suggestion module generates a quality score of the personalized training suggestion based on the basketball state estimation, and the training suggestion module is connected to the dynamic adjustment module; The dynamic adjustment module adjusts the parameters of the basketball training by analyzing the quality score of the personalized training suggestion, generates new training parameters, and transmits the new training parameters to the feedback adjustment module; The feedback adjustment module collects parameters of the training device and dynamically adjusts the settings of the training device according to new training parameters.
[0007] The method for collecting shooting motion based on image recognition and sensing backboard includes the following steps: S1, capture the dynamics of the basketball, obtain the original data, and use the multi-level image recognition algorithm and the sensor backboard data fusion algorithm to obtain the basketball state estimation; S2. Based on the basketball state estimation, generate the quality score of the personalized training suggestion, further form new training parameters, and then calculate the training equipment adjustment amount, and dynamically adjust the settings of the training equipment; Applicable to the above-mentioned system for collecting shooting motion based on image recognition and sensing the backboard.
[0008] Preferably, the S1 specifically includes: In the process of realizing the multi-level image recognition algorithm, high dynamic range imaging technology is introduced to obtain high dynamic range images.
[0009] Preferably, the S1 specifically includes: An improved scale-invariant feature transformation algorithm is used to extract feature points from high dynamic range images and determine the basketball motion path.
[0010] Preferably, the S1 specifically includes: Based on the changes in feature points, the optical flow method is used to obtain image recognition data.
[0011] Preferably, the S1 specifically includes: In the process of implementing the inductive backboard data fusion algorithm, the backboard sensing data and the image recognition data in the original data are fused and recorded in the form of a force-time curve to obtain the sensed force.
[0012] Preferably, the S1 specifically includes: The observation vector is defined by combining the image recognition data and the sensed force. Based on the observation vector, a Kalman filter-based method is used for fusion to obtain the basketball state estimation.
[0013] Preferably, the S2 specifically includes: By analyzing the scores of the indicators in the quality score of the personalized training recommendation, the parameters of the basketball training are adjusted and new training parameters are generated.
[0014] Preferably, the S2 specifically includes: Based on the new training parameters, the target speed and target acceleration of the basketball are calculated, and then the adjustment amount of the training equipment is obtained.
[0015] The beneficial effects of the technical solution of the present invention are: 1. By combining high dynamic range imaging technology and an improved scale-invariant feature transformation algorithm, the system based on image recognition and sensing the backboard to collect shooting motion can accurately track the position and trajectory of the basketball under various lighting conditions, allowing coaches to analyze the performance of athletes based on precise data and adjust training plans in a targeted manner, thereby providing more personalized training support; 2. The inductive backboard in the system for collecting shooting motion based on image recognition and inductive backboard captures the position and force of the basketball contact, and combines image recognition technology to provide a comprehensive data set for real-time monitoring and analysis of shooting movements, which not only helps to adjust training strategies in real time, but also can provide instant feedback to athletes to help them understand and improve their techniques; 3. The data integrated through Kalman filtering technology not only provides accurate motion trajectories, but also allows the system to dynamically adjust the settings of the training equipment (such as shooting speed and angle) based on image recognition and sensing the backboard to collect shooting movement. This is done automatically according to the athlete's real-time performance and training needs, thereby ensuring that the settings of the training equipment are always optimized to support the best training results. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] Figure 1 This is a structural diagram of the system for collecting shooting motion based on image recognition and sensing the backboard according to the present invention; Figure 2This is a flow chart of the method for collecting shooting motion based on image recognition and sensing the backboard according to the present invention. DETAILED DESCRIPTION
[0017] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiment of the present invention, not all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0018] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
[0019] The specific scheme of the system and method for collecting shooting motion based on image recognition and sensing backboard provided by the present invention is described in detail below with reference to the accompanying drawings.
[0020] See attached Figure 1 , which shows a system structure diagram of collecting shooting motion based on image recognition and sensing backboard provided by an embodiment of the present invention, the system includes the following parts: Data acquisition module, wireless module, image recognition module, basketball tracking module, training suggestion module, dynamic adjustment module, feedback adjustment module; The data acquisition module is used to capture the dynamics of the basketball and collect raw data including continuous video streams or images, backboard sensing data and environmental conditions. The data acquisition module transmits the raw data via a wireless module; The wireless module is used to transmit the collected raw data to the image recognition module and the basketball tracking module; The image recognition module designs a multi-level image recognition algorithm, combines multiple images with different exposure levels to increase the dynamic range of the image, obtains a high dynamic range image, and uses an improved scale-invariant feature transformation algorithm to extract feature points of the basketball in continuous frames from the high dynamic range image. The frame is the basic unit of the video stream. The image recognition module is connected to the basketball tracking module; The basketball tracking module estimates the basketball's motion trajectory using the optical flow method and uses Kalman filtering to integrate the basketball's motion trajectory and backboard sensing data to obtain a basketball state estimate; the basketball tracking module is connected to the training suggestion module; The training suggestion module generates a quality score of the personalized training suggestion based on the basketball state estimation, and the training suggestion module is connected to the dynamic adjustment module; The dynamic adjustment module dynamically adjusts the core parameters of basketball training by analyzing the quality scores of personalized training suggestions, generates new training parameters, and transmits the calculated new training parameters to the feedback adjustment module; The feedback adjustment module collects parameters of the training equipment in real time, and adjusts the settings of the training equipment, such as the settings of the basketball launcher, in real time according to the new training parameters, to ensure that the training environment matches the training plan.
[0021] See attached Figure 2 , which shows a flow chart of a method for collecting shooting motion based on image recognition and sensing backboard provided by an embodiment of the present invention, the method comprising the following steps: S1, capture the dynamics of the basketball, obtain the original data, and use the multi-level image recognition algorithm and the sensor backboard data fusion algorithm to obtain the basketball state estimation; The data acquisition module captures the dynamics of the basketball through cameras deployed in various corners and above the basketball court, collects a variety of raw data including continuous video streams or images, backboard sensing data and environmental conditions, and transmits them to the central processing unit in real time through a wireless module. The backboard sensing data records the position and force of the basketball contact in detail through the pressure sensor embedded in the backboard. The data acquisition module is designed with environmental sensors to monitor environmental conditions such as light.
[0022] All raw data is integrated and analyzed in the central processing unit, using a multi-level image recognition algorithm and a sensor backboard data fusion algorithm. By combining image recognition and sensing technology, real-time monitoring and analysis of basketball shooting actions can be achieved, including the precise measurement of the basketball's trajectory, speed, rotation, and backboard contact point.
[0023] The image recognition module designs a multi-level image recognition algorithm, which uses multi-stage image processing technology to accurately identify and track the position and trajectory of the basketball in real time, improve the image processing capabilities under different lighting conditions, and enhance the accuracy and stability of feature point matching.
[0024] Specifically, the multi-level image recognition algorithm uses high dynamic range imaging (HDR) technology to enhance the quality of images captured by the camera, especially under different lighting conditions; by combining multiple images with different exposure levels to increase the dynamic range of the image, thereby improving the visible details and contrast in the image. The implementation formula of the high dynamic range imaging technology is: , in, It is a high dynamic range image, which is the synthesis result of multiple images with different exposures, used to enhance the details and contrast of the image. Indicates The pixel values of the images represent the single image data captured under different exposure conditions. is the weight function, based on The weight is determined by the brightness of the pixel values of the image. is the number of images.
[0025] The image recognition module also uses an improved scale-invariant feature transform (SIFT) algorithm to extract the feature points of the basketball in consecutive frames from the high dynamic range image, which is the basic unit of the video stream, and matches them to determine the movement path of the basketball in the video stream of the data acquisition module, thereby improving the accuracy of basketball position recognition and maintaining stability even in complex backgrounds or fast movements. The improved scale-invariant feature transform formula is: , in, is in scale The Laplace expression of the image under is used to detect feature points at the current scale. is the Gaussian blur kernel, used to blur the image in scale space, is the standard deviation of the kernel, representing the scale variation parameter, The choice of determines the feature extraction capability of the image at different scales. are the pixel coordinates of the high dynamic range image The pixel value of are the pixel coordinates of the background image The pixel value is used to reduce the background noise in dynamic scenes through background subtraction technology. are the image pixel coordinates.
[0026] The basketball tracking module uses the optical flow method to estimate the trajectory of the basketball, calculates the speed and direction of the basketball based on the changes in feature points, uses the second-order derivative to provide accurate speed estimation, obtains image recognition data, forms the trajectory of the basketball, and improves the tracking performance of the fast-moving basketball. The speed estimation formula is: , in, is the basketball velocity vector, They represent the velocity components of the basketball in the horizontal and vertical directions respectively, and The image is and The spatial gradient in the direction is used to capture the curvature changes at the edge of the image, which helps to track fast-moving objects. is the time gradient, used to smooth the changes in time, is time, and the gradient is calculated based on the change in pixel intensity between two consecutive frames to achieve an accurate estimation of the basketball velocity vector. is the square of the norm of the Laplace operator and is used to calculate the normalized velocity vector.
[0027] The sensor backboard data fusion algorithm provides a more comprehensive basketball shot analysis by fusing the backboard sensing data with the image recognition data. The built-in sensor of the backboard captures the contact position and force, records it in the form of a force-time curve, and obtains the sensed force.
[0028] , in, Indicates at time and location The force sensed at represents the amplitude, which represents the maximum value of the contact force; is the time decay factor, which indicates the rate at which the force decreases over time; is the spatial frequency, related to the distribution of contact points on the backboard; is the initial phase, indicating the initial time point when contact occurs; represents the amplitude of vibration, Represents the frequency of vibration. By describing the attenuation and periodic changes of the force, it can accurately capture the dynamic characteristics of the basketball when it hits the backboard.
[0029] The basketball tracking module uses Kalman filtering to combine information from image recognition data and sensed force to define an observation vector is a composite vector that includes speed and contact force to fully reflect the dynamics of each shot: , The Kalman filter-based method is used to fuse the image recognition data and the sensed force to improve the reliability and accuracy of the data and provide a scientific basis for training. The specific fusion formula is: , , in, Represents the basketball state estimate for the next time step, which represents the fusion of image recognition data, sensed force, and the current basketball state estimate , the best prediction of the ball's position and speed, is the state transition matrix, describing the evolution of the basketball state from one moment to the next; is the control input matrix, is the control input, which is used to introduce external control effects, such as environmental factors, training equipment adjustments, and the use of training auxiliary equipment; is the Kalman gain, which is used to optimize the accuracy of the estimation; is the observation matrix, which is used to map the basketball state to the observation space; is the estimation error covariance, which indicates the uncertainty of the current state estimate, is the transpose, is the observation noise covariance, which represents the statistical characteristics of the measurement error.
[0030] S2. Based on the basketball state estimation, generate the quality score of personalized training suggestions, further form new training parameters, and then calculate the adjustment amount of training equipment to dynamically adjust the settings of training equipment.
[0031] The position, velocity, acceleration and rotation data can be directly extracted from the basketball state estimation. The contact force is transmitted by the basketball tracking module, which provides data support for analyzing the dynamics of the basketball when it hits the backboard. The training suggestion module generates a quality score for personalized training suggestions based on the basketball state estimation at the next time step to improve the athlete's skills and efficiency.
[0032] , in, is the quality score of the personalized training suggestions, is the shooting success rate, It is the speed of improving shooting skills. It is the skill level, which is a quantitative indicator. is the estimated height of the current shot extracted from the basketball state estimate, is the maximum shooting height, is the variance of the basketball state estimate at the next time step, is the velocity vector extracted from the basketball state estimate, is the magnitude of the velocity vector, is the acceleration vector extracted from the basketball state estimate, is the magnitude of the acceleration vector, is the weight coefficient, which is used to adjust the impact of success rate in the total score. is the weight coefficient, which is used to adjust the impact of the shooting skill improvement speed on the total score. is the weight coefficient used to adjust the impact of the shot height ratio on the total score. is the weight coefficient, which is used to adjust the influence of the variance of basketball state estimation on the total score. is the weight coefficient, which is used to adjust the influence of the velocity vector modulus in the total score. is the weight coefficient used to adjust the influence of the magnitude of the acceleration vector in the total score.
[0033] Dynamically adjust the module by parsing the quality score of personalized training suggestions , dynamically adjust the core parameters of basketball training, such as training intensity, key skills and recovery cycle. The quality score of personalized training suggestions comprehensively reflects the athlete's performance and training effect. It is calculated based on the precise measurement data of the trajectory, speed, rotation and backboard contact point of the basketball. The score of each indicator in the quality score of personalized training suggestions is analyzed to identify weaknesses and strengths in training.
[0034] A target score is preset based on the ideal training effect. The target score is obtained by expert experience. New training parameters are generated based on the gap between the quality score of the personalized training suggestion and the target score, including the duration, number of repetitions, intensity, technical difficulty, etc. of the training. By adjusting the new training parameters, the training plan can be more accurately targeted at the athlete's current needs and potential. The new training parameters are calculated using the following formula: , in, are the new training parameters, It is the current training parameter, which is set by the coach based on his professional experience and understanding of the athletes. is the sensitivity adjustment factor, which is used to control the speed and amplitude of parameter adjustment. is the target score, The function is used to convert the score difference Mapping to a smooth adjustment scale makes parameter adjustment smoother and more continuous.
[0035] The new training parameters will be calculated The feedback adjustment module is used to set the training equipment and method for the next training cycle. The feedback adjustment module uses the parameters collected in real time from the training equipment (including basketball speed and acceleration ) and new training parameters , adjust the settings of the training equipment, such as the shooting speed and angle of the basketball machine, to ensure that the training conditions adapt to the athlete's current state and new training parameters in real time.
[0036] Based on the new training parameters, calculate the target speed and target acceleration of the basketball, and then adjust the training equipment settings to match the target speed and target acceleration of the basketball. Use the following formula to calculate the required training equipment adjustments: , in, represents the adjustment amount of the training equipment, and is the weight factor, which adjusts the impact weight of the target deviation of the speed and acceleration of the basketball. is the Euclidean norm, According to the new training parameters Calculate the target speed of the basketball, According to the new training parameters Calculate the target acceleration of the basketball. The specific formula is: , , in, and They are the basic values of the speed and acceleration of the basketball. and They are the speed adjustment factor and the acceleration adjustment factor, and are the new training parameters The section for speed and acceleration adjustments.
[0037] Based on the calculated training equipment adjustment amount, the settings of the training equipment are dynamically adjusted, such as the shooting speed and angle of the basketball launcher. The adjusted training equipment settings will be applied in real time in the training environment to optimize the training effect and ensure that the training conditions are consistent with the athlete's needs and training plan.
[0038] In summary, a system and method for collecting shooting motion based on image recognition and sensing the backboard has been completed.
[0039] The order of the embodiments of the invention is for description only and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0040] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.
[0041] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.
Claims
1. A system for collecting shooting motion based on image recognition and induction backboard, characterized in that: Includes the following sections: Data acquisition module, wireless module, image recognition module, basketball tracking module, training suggestion module, dynamic adjustment module, feedback adjustment module; The data acquisition module is used to capture the dynamics of the basketball and collect raw data. The data acquisition module transmits the raw data through the wireless module; The wireless module is used to transmit the original data to the image recognition module and the basketball tracking module; The image recognition module designs a multi-level image recognition algorithm to obtain a high dynamic range image, uses an improved scale-invariant feature transformation algorithm to extract feature points from the high dynamic range image, and the image recognition module is connected to the basketball tracking module; The basketball tracking module estimates the basketball's motion trajectory based on the changes in feature points using the optical flow method; and obtains a basketball state estimate by combining the basketball's motion trajectory with the backboard sensing data in the original data; the basketball tracking module is connected to the training suggestion module; The training suggestion module generates a quality score of the personalized training suggestion based on the basketball state estimation, and the training suggestion module is connected to the dynamic adjustment module; The dynamic adjustment module adjusts the parameters of the basketball training by analyzing the quality score of the personalized training suggestion, generates new training parameters, and transmits the new training parameters to the feedback adjustment module; The feedback adjustment module collects parameters of the training device and dynamically adjusts the settings of the training device according to new training parameters.
2. The method for collecting shooting motion based on image recognition and sensing the backboard is applied to the system for collecting shooting motion based on image recognition and sensing the backboard as claimed in claim 1, characterized in that: The following steps are involved: S1, capture the dynamics of the basketball, obtain the original data, and use the multi-level image recognition algorithm and the sensor backboard data fusion algorithm to obtain the basketball state estimation; S2. Based on the basketball state estimation, generate the quality score of personalized training suggestions, further form new training parameters, and then calculate the adjustment amount of training equipment to dynamically adjust the settings of training equipment.
3. The method for collecting shooting motion based on image recognition and sensing backboard as claimed in claim 2, characterized in that: The S1 specifically includes: In the process of realizing the multi-level image recognition algorithm, high dynamic range imaging technology is introduced to obtain high dynamic range images.
4. The method for collecting shooting motion based on image recognition and sensing backboard as claimed in claim 3, characterized in that: The S1 specifically includes: Feature points are extracted from high dynamic range images using an improved scale-invariant feature transform algorithm.
5. The method for collecting shooting motion based on image recognition and sensing backboard as claimed in claim 4, characterized in that: The S1 specifically includes: Based on the changes in feature points, the optical flow method is used to obtain image recognition data.
6. The method for collecting shooting motion based on image recognition and sensing backboard as claimed in claim 5, characterized in that: The S1 specifically includes: In the process of implementing the inductive backboard data fusion algorithm, the backboard sensing data and the image recognition data in the original data are fused and recorded in the form of a force-time curve to obtain the sensed force.
7. The method for collecting shooting motion based on image recognition and sensing backboard as claimed in claim 6, characterized in that: The S1 specifically includes: The observation vector is defined by combining the image recognition data and the sensed force. Based on the observation vector, a Kalman filter-based method is used for fusion to obtain the basketball state estimation.
8. The method for collecting shooting motion based on image recognition and sensing backboard as claimed in claim 2, characterized in that: The S2 specifically includes: By analyzing the scores of the indicators in the quality score of the personalized training recommendation, the parameters of the basketball training are adjusted and new training parameters are generated.
9. The method for collecting shooting motion based on image recognition and sensing backboard as claimed in claim 8, characterized in that: The S2 specifically includes: Based on the new training parameters, the target speed and target acceleration of the basketball are calculated, and then the adjustment amount of the training equipment is obtained.