A Shooting Positioning Method and System Based on Shooting Curve Analysis and Posture Matching

Abstract

This invention proposes a method and system for determining the shooting point based on shooting curve analysis and posture matching, relating to the field of computer vision. The method involves preprocessing acquired shooting videos to obtain a video image sequence; extracting a complete binary image of the moving object from the video image sequence to identify the moving target; constructing the basketball's trajectory based on the basketball's center point within the moving target and determining the starting point of the shooting trajectory; and then, by reverse-engineering the trajectory based on the basketball's trajectory and the starting point of the shooting trajectory, determining the shooter and their posture; finally, determining the shooter's shooting point based on the shooter's posture. This invention extracts the basketball's trajectory and the athlete's hand posture from a static image sequence. Based on this information, it estimates the initial point by reverse-engineering the trajectory, achieving automatic detection of the shooter's shot point. This helps coaches and athletes develop tactical strategies and simulate training directions, improving the team's winning rate.

Description

Technical Field

[0001] This invention belongs to the field of computer vision, and in particular relates to a shooting positioning method and system based on shooting curve analysis and pose matching. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] In basketball games, players review and analyze the game after scoring. Although the human eye can directly see the trajectory of the basketball and the position of the shooter in the video, the human brain cannot collect a large amount of motion information and cannot accurately locate the shooter's shooting position. Therefore, it cannot meet the need to further improve the team's shooting accuracy by analyzing game information.

[0004] Existing methods for analyzing shooting points include those that use methods similar to reversing video playback, as well as those that employ single methods such as deep learning or single formulas. These methods all suffer from low accuracy or low efficiency. Therefore, seeking a more efficient and accurate method for determining shooting points has become a worthy research topic. Summary of the Invention

[0005] To overcome the shortcomings of the prior art, this invention provides a shooting positioning method and system based on shooting curve analysis and posture matching. It extracts the basketball trajectory and the athlete's hand posture from a static image sequence. Based on these two pieces of information, it estimates the initial position by reverse-engineering the trajectory, thereby achieving automatic detection of the shooter's shot point. This helps coaches and athletes formulate tactical strategies and simulate training directions, improving the team's winning rate.

[0006] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions:

[0007] The first aspect of this invention provides a shooting positioning method based on shooting curve analysis and posture matching;

[0008] Shooting positioning methods based on shot curve analysis and pose matching include:

[0009] The acquired basketball shooting videos are preprocessed to obtain a video image sequence;

[0010] Extract the complete binary image of the moving object from the video image sequence, determine the moving target, construct the basketball trajectory based on the center point of the basketball in the moving target, and determine the starting point of the shooting trajectory.

[0011] Based on the basketball's trajectory and the starting point of the shot, the trajectory is deduced in reverse to determine the shooter and their position.

[0012] Based on the shooter's posture, determine the shooter's shooting point.

[0013] Furthermore, the preprocessing involves dividing the basketball shooting video captured by the imaging device into a video image sequence frame by frame, and performing image color space conversion, image denoising, and image enhancement processing on each frame in the sequence.

[0014] Furthermore, a moving target detection algorithm combining frame difference and background difference methods is used to determine moving targets, specifically including:

[0015] The motion regions were extracted from the video image sequence using both frame difference and background difference methods.

[0016] The binary images of the moving regions extracted by the two methods are subjected to a logical OR operation, and the results of the two detection algorithms are combined to obtain a combined binary image of the moving target.

[0017] Mathematical morphology operations are used to remove noise and fill holes in the synthesized binary image of the moving target to obtain the final binary image of the moving target.

[0018] Furthermore, the motion region is extracted using the background subtraction method, specifically:

[0019] The background model is obtained through background modeling methods;

[0020] The difference between each frame of the video image sequence and the background model is calculated.

[0021] The obtained difference image is binarized pixel by pixel according to the set pixel threshold to obtain the binary image of the moving region.

[0022] The motion region is extracted using the frame difference method, specifically:

[0023] Subtract the pixel values ​​of two adjacent frames or two images separated by several frames;

[0024] Thresholding is applied to the subtracted images to extract the moving regions.

[0025] Furthermore, based on the shape characteristics of the basketball, the Hough circle detection method is used to mark the basketball, and then the centroid detection method is used to obtain the center point of the basketball.

[0026] Arrange the center points of the basketball in the video image sequence according to the time sequence in a two-dimensional rectangular coordinate system, and connect the arranged center points of the basketball to obtain the trajectory of the basketball during the shooting process.

[0027] Furthermore, a parabolic function is fitted to the latter part of the basketball's trajectory, and the coordinates of the former part of the basketball's trajectory are successively substituted into the quadratic function in the order from back to front to calculate the error. When the error begins to exceed the set threshold, the point is regarded as the instant the basketball is thrown, that is, the starting point of the shooting trajectory.

[0028] Furthermore, the coordinates of key points on the human body are obtained using a human posture detection model. When the athlete's hand is higher than his head, he is identified as a shooting candidate. Among multiple candidates, the distance between the athlete's hand and the position of the basketball at the moment of release is calculated. The candidate corresponding to the hand coordinate with the smallest distance is identified as the shooting player.

[0029] The second aspect of the present invention provides a shooting accuracy system based on shooting curve analysis and posture matching.

[0030] A shooting accuracy system based on shooting curve analysis and posture matching includes a preprocessing module, a trajectory calculation module, a posture deduction module, and a position determination module.

[0031] The preprocessing module is configured to preprocess the acquired basketball shooting videos to obtain a video image sequence;

[0032] The trajectory calculation module is configured to: extract the complete binary image of the moving object from the video image sequence, determine the moving target, construct the basketball trajectory based on the center point of the basketball in the moving target, and determine the starting point of the shooting trajectory;

[0033] The pose deduction module is configured to: reverse-deduce the running trajectory based on the basketball's trajectory and the starting point of the shooting trajectory, and determine the shooter and the shooter's pose;

[0034] The point determination module is configured to determine the shooting point of the basketball player based on the shooting player's posture.

[0035] A third aspect of the present invention provides a computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the steps of the shooting positioning method based on shooting curve analysis and pose matching as described in the first aspect of the present invention.

[0036] The fourth aspect of the present invention provides an electronic device, including a memory, a processor, and a program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps in the shooting positioning method based on shooting curve analysis and pose matching as described in the first aspect of the present invention.

[0037] The above one or more technical solutions have the following beneficial effects:

[0038] The shooting point prediction method provided by this invention uses computer vision instead of human vision to extract the basketball trajectory and the athlete's hand posture from a static image sequence. Based on these two pieces of information, the initial point is estimated by reverse deducing the trajectory. This method uses computer technology to solve the problem of information analysis after a basketball game, and realizes the automatic detection of the shot's point of impact. It helps coaches and athletes to formulate tactical strategies and simulate training directions, thereby improving the team's winning rate.

[0039] This invention employs a multi-technology fusion method for fixed-point recognition, involving the integration of multiple methods such as video enhancement processing, background subtraction, Hough detection, and formula localization. Compared with other technologies that use a single method for localization, it achieves higher recognition accuracy.

[0040] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0041] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0042] Figure 1 This is a flowchart of the method in the first embodiment.

[0043] Figure 2 This is a system structure diagram of the second embodiment. Detailed Implementation

[0044] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0045] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of the invention; unless otherwise specified, 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 pertains.

[0046] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments of the present invention; as used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise; furthermore, it should be understood that when the terms “comprising” and / or “including” are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or combinations thereof.

[0047] Example 1

[0048] This embodiment discloses a shooting positioning method based on shooting curve analysis and posture matching;

[0049] like Figure 1 As shown, the shooting accuracy method based on shooting curve analysis and pose matching includes:

[0050] Step S1: Preprocess the acquired basketball shooting video to obtain a video image sequence;

[0051] The imaging device captures the shooting video and cuts out the dribbling and passing processes beforehand, leaving only the complete process of preparing to shoot, shooting, and making the shot.

[0052] The cropped basketball shooting video undergoes preprocessing, including image color space conversion, image denoising, and image enhancement. These operations improve image quality, making the moving basketball in the image easier to identify and increasing the accuracy of moving target detection.

[0053] The clipped basketball shooting video is divided into an image sequence of n frames. The n frames are then subjected to image color space conversion, image denoising, and image enhancement. These operations improve image quality, making the moving basketball in the image easier to identify and increasing the accuracy of moving target detection.

[0054] Step S2: Extract the complete binary image of the moving object from the video image sequence, determine the moving target, construct the basketball trajectory based on the center point of the basketball in the moving target, and determine the starting point of the shooting trajectory, specifically:

[0055] Step S201: Determine the moving target

[0056] To improve the accuracy of trajectory extraction, a moving target detection algorithm combining frame difference and background difference is adopted for target detection because the frame difference and background difference methods are complementary in performance: when using the frame difference method, the time interval between adjacent frames is small, and it is less affected by changes in scene lighting, which can make up for the background difference method's insensitivity to changes in scene environment; the background difference method can extract complete moving targets, which can solve the problem that the frame difference method cannot extract moving targets with complete outlines.

[0057] Background subtraction extracts moving objects by subtracting the current frame from the background frame. Specifically:

[0058] (1) Obtain the background model B(x,y) through background modeling;

[0059] (2) After establishing the background model, the difference between each frame image F(x,y) of the video image sequence and the background image B(x,y) is calculated.

[0060] (3) For the obtained difference image D(x,y), set a threshold T, and perform binarization processing on each pixel to obtain the binary image R(x,y) of the moving region; the binarization processing is as follows: when the pixel value of a point in the difference image is greater than T, it is considered to be a moving target pixel; otherwise, it is a background pixel. The formula is expressed as:

[0061] D(x,y)=|F(xy)-B(x,y)|

[0062]

[0063] The frame difference method is similar in principle to the background difference method, and the processing procedure is as follows:

[0064] (1) Subtract the pixel values ​​of two adjacent frames or two images separated by several frames;

[0065] (2) Threshold the images after subtraction to extract the motion regions. If the frames of the two subtracted images are frame k and frame (k+1) respectively, and their frame images are Fk(x,y) and Fk+1(x,y) respectively, the binarization threshold of the difference image is T, and the difference image is represented by D(x,y). The calculation process is as follows:

[0066]

[0067] After extracting the motion region from the video image sequence using the two methods described above, a logical "OR" operation is performed on the binary image of the motion region after threshold segmentation to obtain the moving target. At this point, the detected moving target still contains noise, holes, etc. Mathematical morphology operations are used to further process the image to remove noise and fill in the holes.

[0068] Erosion operations shrink the boundaries of objects inward, eliminating small and meaningless objects. Selecting large structuring elements can erode even the smallest connections between objects. Dilation, the dual operation of erosion, expands the boundaries of objects outward, merging background points in contact with the object into it. Dilation can also fill holes within objects. After erosion and dilation, a closing operation is performed on the moving targets extracted by frame difference and background subtraction. This process, first dilation then erosion, fills the holes inside the moving targets. Finally, the image is opened, first eroded then dilated, to eliminate small objects and improve the moving target detection results, yielding the final moving target detection outcome.

[0069] Step S202: Extract motion trajectory

[0070] The extracted complete binarized image of the moving target, including the player and the basketball, needs to be used to detect the basketball as the moving target.

[0071] Based on the shape characteristics of a basketball, specify the radius of the circle (r).max r min The detection threshold range (the minimum and maximum values ​​that the basketball may appear in the image) and the circular edge gradient range (the maximum deformation caused by a single frame exposure of the camera when the basketball is in motion) are used to exclude other similar circular interferences. The basketball is marked using the Hough circle detection method.

[0072] Based on the labeled basketball, the coordinates of the basketball's center point are obtained using centroid detection. Specifically, if the grayscale value at point (x,y) is G(x,y), and the size of the labeled basketball image is M×N, then the coordinates of the basketball's center point are:

[0073]

[0074] After the above processing, the identified basketball center point coordinates are arranged in a two-dimensional Cartesian coordinate system according to time sequence. Connecting the arranged points yields the basketball's trajectory during the shooting process. Although this trajectory does not include the basketball's movement during the initial dribbling and passing phases, it does include the trajectory during the pre-shooting phase. Further processing is needed to obtain the accurate coordinates of the basketball at the moment of release.

[0075] Step S203: Determine the starting point of the shooting trajectory

[0076] A parabolic function is fitted to the latter part of the basketball's trajectory. The coordinates of the former part of the trajectory are then substituted into a quadratic function in reverse order to calculate the error. When the error exceeds a set threshold, that point is considered the instant the basketball is released, i.e., the starting point of the shooting trajectory. Specifically:

[0077] In the n-frame binary images with the basketball's center point marked, we take the frame number k corresponding to the maximum value of the ordinate in the basketball's trajectory, i.e., the frame containing the highest point of the parabola. The nth frame is the image frame corresponding to when the basketball enters the hoop. We then take the coordinates of the basketball's center point from the kth to the nth frame, and use these k-n+1 coordinate points to fit the parabolic function y = ax. 2 Given the equation +bx+c, calculate the specific values ​​of a, b, and c to obtain the final parabolic trajectory function of the basketball: y = ax + bx + c. 2 +bx+c.

[0078] The center point x of the basketball in the first k-1 frames of the basketball's trajectory. m The coordinates (m = k-1, k-2, k-3, ..., 1) are substituted into the parabola function in reverse order to calculate y. When Δy = |y m When -y| < ε, the coordinates can be determined to be the center coordinates of the basketball after it has been thrown. When Δy = |y| < ε, the coordinates are considered to be the center coordinates of the basketball after it has been thrown. m -y|>ε and Δy=|y m+1 When -y|<ε, then (xm ,y m The coordinates of the basketball at the moment it is thrown can be regarded as the starting point of the shooting trajectory, and the corresponding m-th frame image at the moment of throwing is saved.

[0079] Step S3: Based on the basketball's trajectory and the starting point of the shooting trajectory, reverse the trajectory to determine the shooter and the shooter's position;

[0080] The system uses a human posture detection model to obtain the coordinates of key human points. When an athlete's hand is higher than their head, they are identified as a potential shooter. Among multiple candidates, the distance between the athlete's hand and the position of the basketball at the moment of release is calculated. The candidate with the smallest distance is identified as the shooter. Specifically:

[0081] Step S301: Pose Analysis

[0082] The m-th frame image corresponding to the instant the basketball is thrown is used as the input image for human pose analysis. Human pose detection models such as OpenPose are used to extract key points, obtaining the athlete's pose information, including: hand key point coordinates (a...). i ,b i ), Head key point coordinates (h i ,t i ), foot key point coordinates (f i ,g i ), (f' i ,g' i (These are the coordinates of the left and right feet), i is the athlete's number (i = 1, 2, 3...), and the key point coordinate information of each athlete is saved in sequence.

[0083] Step S302: Candidate Screening

[0084] When the number of identified athletes is greater than 1, further screening is needed to identify athletes with a throwing motion. Among the identified athletes, those with a raised hand gesture are selected as candidates, i.e., by comparing whether the ordinate of the athlete's hand key point is greater than the ordinate of the head key point. The coordinates of the athlete's hand key point are (a i ,b i The coordinates of the head key points are (h i ,t i When b i >t i At that time, the athlete was identified as a candidate, and their hand coordinates (a) were recorded. i ,b i When only one athlete is identified, they are directly identified as the shooter, and their left and right foot coordinates are recorded (f). i ,g i ), (f' i,g' i ).

[0085] Step S303: Determine the basketball shooter.

[0086] Among the shortlisted candidates, further evaluation of the specific shooting player is needed. This is achieved by comparing the distance Δd between the candidate's key hand points and the starting point of the shooting trajectory. At its minimum, the hand coordinates are (a i ,b i The player who shoots the basketball is the one who shoots the ball. At the same time, the coordinates of their left and right feet (f) are recorded. i ,g i ), (f' i ,g' i ).

[0087] Step S4: Determine the shooting point of the basketball player based on the shooting player's posture.

[0088] Based on the shooting posture of the basketball player, the shooting player has been determined in step S3. The average of the coordinates of the left and right feet of the shooting player is used to obtain the final shooting point coordinates.

[0089] Example 2

[0090] This embodiment discloses a shooting accuracy system based on shooting curve analysis and posture matching;

[0091] like Figure 2 As shown, the shooting accuracy system based on shooting curve analysis and posture matching includes a preprocessing module, a trajectory calculation module, a posture deduction module, and a point determination module.

[0092] The preprocessing module is configured to preprocess the acquired basketball shooting videos to obtain a video image sequence;

[0093] The trajectory calculation module is configured to: extract the complete binary image of the moving object from the video image sequence, determine the moving target, construct the basketball trajectory based on the center point of the basketball in the moving target, and determine the starting point of the shooting trajectory;

[0094] The pose deduction module is configured to: reverse-deduce the running trajectory based on the basketball's trajectory and the starting point of the shooting trajectory, and determine the shooter and the shooter's pose;

[0095] The point determination module is configured to determine the shooting point of the basketball player based on the shooting player's posture.

[0096] Example 3

[0097] The purpose of this embodiment is to provide a computer-readable storage medium.

[0098] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the shooting positioning method based on shooting curve analysis and pose matching as described in Embodiment 1 of this disclosure.

[0099] Example 4

[0100] The purpose of this embodiment is to provide an electronic device.

[0101] An electronic device includes a memory, a processor, and a program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps in the shooting positioning method based on shooting curve analysis and pose matching as described in Embodiment 1 of this disclosure.

[0102] 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 shooting positioning method based on shooting curve analysis and posture matching, characterized in that, include: The acquired basketball shooting videos are preprocessed to obtain a video image sequence; Extract the complete binary image of the moving object from the video image sequence to determine the moving target. Based on the center point of the basketball in the moving target, construct the basketball's trajectory and determine the starting point of the shooting trajectory. Specifically: Based on the shape characteristics of the basketball, the Hough circle detection method is used to mark the basketball. Then, the centroid detection method is used to obtain the center point of the basketball. Specifically, if the gray value at point (x,y) is G(x,y), and the size of the marked image is M×N, then the coordinates of the center point of the basketball are: ； Arrange the center points of the basketball in the video image sequence according to the time sequence in a two-dimensional rectangular coordinate system, and connect the arranged center points of the basketball to obtain the trajectory of the basketball during the shooting process. The latter part of the basketball's trajectory is fitted with a parabolic function, and the coordinates of the former part of the basketball's trajectory are successively substituted into the quadratic function in the order from back to front to calculate the error. When the error begins to exceed the set threshold, the point is regarded as the instant the basketball is thrown, that is, the starting point of the shooting trajectory. Based on the basketball's trajectory and the starting point of the shot, the trajectory is deduced in reverse to determine the shooter and their position. The coordinates of key points of the human body are obtained by using a human posture detection model. When the athlete's hand is higher than his head, he is identified as a shooting candidate. Among multiple candidates, the distance between the athlete's hand and the position of the basketball at the moment of release is calculated. The candidate corresponding to the hand coordinate with the smallest distance is identified as the shooting player. Based on the shooter's posture, determine the shooter's shooting point.

2. The shooting positioning method based on shooting curve analysis and posture matching as described in claim 1, characterized in that, The preprocessing involves dividing the basketball shooting video captured by the imaging device into a video image sequence frame by frame, and performing image color space conversion, image denoising, and image enhancement processing on each frame in the sequence.

3. The shooting positioning method based on shooting curve analysis and posture matching as described in claim 1, characterized in that, A moving target detection algorithm combining frame difference and background difference methods is used to determine moving targets, specifically including: The motion regions were extracted from the video image sequence using both frame difference and background difference methods. The binary images of the moving regions extracted by the two methods are subjected to a logical OR operation, and the results of the two detection algorithms are combined to obtain a combined binary image of the moving target. Mathematical morphology operations are used to remove noise and fill holes in the synthesized binary image of the moving target to obtain the final binary image of the moving target.

4. The shooting positioning method based on shooting curve analysis and posture matching as described in claim 1, characterized in that, The motion region was extracted using the background subtraction method, specifically: The background model is obtained through background modeling methods; The difference between each frame of the video image sequence and the background model is calculated. The obtained difference image is binarized pixel by pixel according to the set pixel threshold to obtain the binary image of the moving region. The motion region is extracted using the frame difference method, specifically: Subtract the pixel values ​​of two adjacent frames or two images separated by several frames; Thresholding is applied to the subtracted images to extract the moving regions.

5. A shooting accuracy system based on shooting curve analysis and posture matching, characterized in that, It includes a preprocessing module, a trajectory calculation module, a pose deduction module, and a point determination module: The preprocessing module is configured to preprocess the acquired basketball shooting videos to obtain a video image sequence; The trajectory calculation module is configured to: extract the complete binary image of the moving object from the video image sequence, determine the moving target, construct the basketball's trajectory based on the basketball's center point in the moving target, and determine the starting point of the shooting trajectory, specifically: Based on the shape characteristics of the basketball, the Hough circle detection method is used to mark the basketball. Then, the centroid detection method is used to obtain the center point of the basketball. Specifically, if the gray value at point (x,y) is G(x,y), and the size of the marked image is M×N, then the coordinates of the center point of the basketball are: ； Arrange the center points of the basketball in the video image sequence according to the time sequence in a two-dimensional rectangular coordinate system, and connect the arranged center points of the basketball to obtain the trajectory of the basketball during the shooting process. The latter part of the basketball's trajectory is fitted with a parabolic function, and the coordinates of the former part of the basketball's trajectory are successively substituted into the quadratic function in the order from back to front to calculate the error. When the error begins to exceed the set threshold, the point is regarded as the instant the basketball is thrown, that is, the starting point of the shooting trajectory. The pose deduction module is configured to: reverse-deduce the running trajectory based on the basketball's trajectory and the starting point of the shooting trajectory, and determine the shooter and the shooter's pose; The coordinates of key points of the human body are obtained by using a human posture detection model. When the athlete's hand is higher than his head, he is identified as a shooting candidate. Among multiple candidates, the distance between the athlete's hand and the position of the basketball at the moment of release is calculated. The candidate corresponding to the hand coordinate with the smallest distance is identified as the shooting player. The point determination module is configured to determine the shooting point of the basketball player based on the shooting player's posture.

6. A computer-readable storage medium having a program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps in the shooting positioning method based on shooting curve analysis and pose matching as described in any one of claims 1-4.

7. An electronic device, comprising a memory, a processor, and a program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the shooting positioning method based on shooting curve analysis and pose matching as described in any one of claims 1-4.

Images