A posture correction training method and device applied to human-centered intelligent manufacturing

Abstract

The present application provides a posture correction training method and device for human-centered intelligent manufacturing. The method provided by the present application includes: obtaining target joint point coordinates, inputting the target joint point coordinates into a human digital model to obtain the correspondence between the target joint point coordinates and human joints; the human digital model is used to match the target joint point coordinates and the real posture data of the human body; calculating according to the vector relationship of the target joint point coordinates to obtain the inter-joint angle and joint torsion angle of the target joint point coordinates; determining a target human risk score based on the correspondence, the inter-joint angle and the joint torsion angle; the target human risk score includes an expert evaluation score; when the target human risk score exceeds a preset risk threshold, a posture correction notification is sent using a composite message; wherein, the composite message sends reminders simultaneously based on multiple senses. The posture correction training method and device for human-centered intelligent manufacturing provided by the present application are used to accurately and reliably correct the posture of the human body.

Description

Technical Field

[0001] The present application relates to the technical field of posture correction, and particularly to a posture correction training method and device applied to human intelligent manufacturing. Background Art

[0002] Musculoskeletal diseases are a common type of health problem, including cervical spondylosis, lumbar disc herniation, arthritis, back pain, etc. These diseases are often related to poor postures and working environments. The ability of an operator to identify and optimize poor postures is a key factor in reducing musculoskeletal diseases, which is usually obtained through implicit experience accumulated over a long period of work or systematic training. Therefore, regardless of the age, skill level, and experience of the operator, conducting systematic posture correction training is of great significance for improving the awareness of identifying and optimizing poor postures.

[0003] Traditional posture correction training is guided by a coach or a physiotherapist, which is not only costly but also difficult to popularize. At the same time, since existing research has not elaborated on the quantization process of the reference posture in detail, there is no unified posture quantization standard. Further, the technology of using wearable devices to detect posture defects and provide vibration feedback not only has a single feedback method but also causes discomfort when worn for a long time, and cannot meet the actual usage requirements. Summary of the Invention

[0004] In view of this, the present application provides a posture correction training method and device applied to human intelligent manufacturing, which can accurately and reliably correct the posture of the human body.

[0005] Specifically, the present application is implemented through the following technical solutions:

[0006] The first aspect of the present application provides a posture correction training method applied to human intelligent manufacturing, and the method includes:

[0007] Obtain the coordinates of target joint points, and input the coordinates of the target joint points into a human digital model to obtain the corresponding relationship between the coordinates of the target joint points and human joints; the human digital model is used to match the coordinates of the target joint points and the real posture data of the human body;

[0008] Calculate according to the vector relationship of the coordinates of the target joint points to obtain the inter-joint angle and joint torsion angle of the coordinates of the target joint points;

[0009] Determine a target human risk score based on the corresponding relationship, the inter-joint angle, and the joint torsion angle; wherein, the target human risk score represents the error degree of the human joints; the target human risk score includes an expert evaluation score;

[0010] When the target human body risk score exceeds a preset risk threshold, a posture correction notice is sent using a composite message; wherein, the composite message sends a reminder based on multiple senses simultaneously.

[0011] The second aspect of this application provides a posture correction training device, which includes an acquisition module, a calculation module, a determination module, and a notification module; wherein,

[0012] The acquisition module is used to acquire target joint point coordinates, input the target joint point coordinates into a human body digital model, and obtain the corresponding relationship between the target joint point coordinates and human joints; the human body digital model is used to match the target joint point coordinates and the real posture data of the human body;

[0013] The calculation module calculates based on the vector relationship of the target joint point coordinates to obtain the inter-joint angle and joint torsion angle of the target joint point coordinates;

[0014] The determination module determines a target human body risk score based on the corresponding relationship, the inter-joint angle, and the joint torsion angle; wherein, the target human body risk score represents the error degree of the human joints; the target human body risk score includes an expert evaluation score;

[0015] The notification module is used to send a posture correction notice using a composite message when the target human body risk score exceeds a preset risk threshold; wherein, the composite message sends a reminder based on multiple senses simultaneously.

[0016] The posture correction training method and device provided in this application for human-centered manufacturing first acquire target joint point coordinates, input the target joint point coordinates into a human body digital model to obtain the corresponding relationship between the target joint point coordinates and human joints, then calculate based on the vector relationship of the target joint point coordinates to obtain the inter-joint angle and joint torsion angle of the target joint point coordinates, then determine a target human body risk score based on the corresponding relationship, the inter-joint angle, and the joint torsion angle, and finally send a posture correction notice using a composite message when the target human body risk score exceeds a preset risk threshold. In this way, through the human body digital model, each joint position and posture of the human body can be accurately captured and analyzed. Further, by calculating the inter-joint angle and joint torsion angle, the correctness and error degree of the posture can be quantified, and then combined with the corresponding relationship to determine the human body risk score, and the posture is judged through the human body risk score, and a reminder is given when the preset risk threshold is exceeded. In this way, the accuracy and credibility of the correction training are ensured. Further, by using a multi-sensory composite message reminder for reminder, attention can be attracted more effectively, so as to adjust the posture in time and improve the user experience. Description of the Drawings

[0017] Figure 1Flowchart of the first embodiment of the posture correction training method applied to human-centered intelligent manufacturing provided by this application;

[0018] Figure 2 Flowchart of the second embodiment of the posture correction training method applied to human-centered intelligent manufacturing provided by this application;

[0019] Figure 3 A hardware structure diagram of the posture correction training device where the posture correction training device of this application is located;

[0020] Figure 4 Schematic structural diagram of the first embodiment of the posture correction training device provided by this application. Detailed implementation manners

[0021] Here, the exemplary embodiments will be described in detail, and the examples are shown in the drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementation manners described in the following exemplary embodiments do not represent all implementation manners consistent with this application.

[0022] The terms used in this application are only for the purpose of describing specific embodiments and are not intended to limit this application. The singular forms "a", "the", and "said" used in this application are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0023] It should be understood that although terms such as first, second, and third may be used in this application to describe various information, such information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of this application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" as used herein may be interpreted as "when" or "while" or "in response to determining".

[0024] This application provides a posture correction training method and device applied to human-centered intelligent manufacturing to accurately and reliably correct the posture of the human body.

[0025] The posture correction training method and device provided by the present application for human-centered intelligent manufacturing first obtain the target joint point coordinates, input the target joint point coordinates into the human digital model to obtain the corresponding relationship between the target joint point coordinates and human joints, and then calculate based on the vector relationship of the target joint point coordinates to obtain the inter-joint angle and joint torsion angle of the target joint point coordinates. Then, based on the corresponding relationship, inter-joint angle, and joint torsion angle, the target human risk score is determined. Finally, when the target human risk score exceeds the preset risk threshold, a posture correction notification is sent using a composite message. In this way, through the human digital model, each joint position and posture of the human body can be accurately captured and analyzed. Further, by calculating the inter-joint angle and joint torsion angle, whether the posture is incorrect and the degree of error can be quantified. Then, combined with the corresponding relationship, the human risk score is determined. The posture is judged through the human risk score, and a reminder is given when the preset risk threshold is exceeded. In this way, the accuracy and credibility of the correction training are ensured. Further, by using multi-sensory composite message reminders, attention can be more effectively attracted, so that posture adjustment can be made in a timely manner, improving the user experience.

[0026] Specific embodiments are given below to introduce the technical solution of the present application in detail.

[0027] Figure 1 It is a flowchart of the first embodiment of the posture correction training method provided by the present application for human-centered intelligent manufacturing. Please refer to Figure 1 , the method provided in this embodiment may include:

[0028] S101. Obtain the target joint point coordinates, input the target joint point coordinates into the human digital model, and obtain the corresponding relationship between the target joint point coordinates and human joints; the human digital model is used to match the target joint point coordinates and the real posture data of the human body.

[0029] Specifically, the target joint point coordinates are the specific positions of each joint of the measured person in three-dimensional space. The posture of the measured person can be determined through the joints corresponding to the target joint point coordinates. When specifically implemented, the camera device for shooting the target joint point coordinates can be selected according to actual needs. For example, an ordinary camera or a depth camera can be selected to obtain the target joint point coordinates.

[0030] Further, the time series data collected can be determined as the target joint point coordinates of the bones of the measured person at a collection frequency of 30 frames.

[0031] Further, the coordinate system corresponding to the target joint point coordinates is determined based on the direction of the measured person. For example, in one embodiment, the three-axis directions are specified. When the measured person is facing the camera directly, the front of the person is the positive X direction, the upper side is the positive Y direction, and the left side is the positive Z direction.

[0032] It should be noted that the specific types of joints included in the target joint position coordinates are set according to actual needs, and are not limited in this embodiment. For example, in one embodiment, the joints corresponding to the target joint position coordinates may include the head, neck, left shoulder, left elbow, left wrist, left palm, left thumb, left middle finger tip, right shoulder, right elbow, right wrist, right palm, right thumb, right middle finger tip, upper spine, middle spine, lower spine, left hip, left knee, left ankle, left foot, right hip, right knee, right ankle, and right foot.

[0033] The following gives a specific embodiment for introducing the process of obtaining the target joint positions in detail:

[0034] (1) Collect the depth image data of the person to be measured.

[0035] Specifically, the depth image data refers to the image data containing the distance information of each pixel in the scene to the camera, and the depth image data can be obtained through a depth sensor or lidar. For example, in one embodiment, the depth image data of the operator is captured in real time by a binocular depth camera Kinect v2.

[0036] (2) Process the depth image data based on the human body bone point recognition algorithm to obtain the target joint point coordinates.

[0037] Specifically, the human body bone point recognition algorithm can detect and recognize the joints of the human body through the information in the image. It should be noted that the specific human body bone point recognition algorithm can be determined according to actual needs, and is not limited in this embodiment. For example, the OpenPose algorithm can be selected as the human body bone point algorithm for processing the depth image data. Another example, in another embodiment, the MediaPipe algorithm can be selected as the human body bone point algorithm for processing the depth image data.

[0038] The posture correction training method provided in this embodiment for human-centered intelligent manufacturing first collects the depth image data of the person to be measured, and then processes the depth image data based on the human body bone point recognition algorithm to obtain the target joint point coordinates. In this way, the target joint point coordinates of the person to be measured can be accurately obtained and recognized, providing a reliable data basis for further posture correction.

[0039] Furthermore, the human digital model is a three-dimensional virtual model generated by a computer. The joint points included in the human digital model can accurately represent the real posture data of the human body, facilitating the calibration and configuration of the joints and bones of the human body.

[0040] In specific implementation, map the target joint point coordinates to the human digital model to obtain the corresponding relationship between each target joint point coordinate and the human joint.

[0041] S102. Calculate according to the vector relationship of the target joint point coordinates to obtain the inter-joint angle and joint twist angle of the target joint point coordinates.

[0042] Specifically, the inter-joint angle refers to the angle between two adjacent bone segments, and the inter-joint angle is determined by calculating the angle between the vectors of the two bone segments.

[0043] Furthermore, the joint twist angle represents the amount of torsion or rotation around the joint, and the rotation angle along the bone segment or joint axis is described by the joint twist angle.

[0044] In specific implementation, the three-dimensional coordinates of the joint can be obtained through the target joint point coordinates, and the corresponding inter-joint angle and joint twist angle are calculated through the coordinates. For example, in one embodiment, taking the inter-joint angles of the shoulder, elbow, and wrist as an example, calculate the vector v1 from the shoulder to the elbow and the vector v2 from the elbow to the wrist, and calculate based on v1 and v2 to obtain the inter-joint angle θ of the shoulder, elbow, and wrist.

[0045] S103. Determine the target human risk score based on the corresponding relationship, the inter-joint angle, and the joint twist angle; wherein, the target human risk score characterizes the error degree of the human joint; the target human risk score includes an expert evaluation score.

[0046] Specifically, the target human risk score is used to evaluate the correctness of the postures and potential health risks of the measured personnel, and is used to evaluate the human posture and joint health status. Through the target human risk score, a quantitative evaluation of the error degree of the human joint posture can be provided. It should be noted that the larger the value of the target human risk score, the more serious the error degree of the human joint.

[0047] In specific implementation, the target human risk score can be obtained based on the traditional human risk score evaluation method, or the target human risk score can be obtained based on the neural network human risk score evaluation method. In this embodiment, it is not limited thereto. For example, in one embodiment, the rapid whole-body assessment method can be adopted to analyze and process the corresponding relationship, the inter-joint angle, and the joint twist angle based on computer vision and kinematic principles, so as to obtain the target human risk score.

[0048] Furthermore, the target human score also includes an expert evaluation score. For example, in one embodiment, after calculating the target human risk score, the expert can perform operations such as correcting or confirming the target human risk score according to the actual situation. For another example, in another embodiment, the expert determines the score according to the actual situation, and after weighted operation, the target human risk score is jointly obtained with the calculated score.

[0049] S104. When the target human body risk score exceeds the preset risk threshold, a posture correction notice is sent using a composite message; wherein, the composite message sends reminders based on multiple senses simultaneously.

[0050] Specifically, the preset risk threshold is determined based on medical data and is a health risk value used to judge whether there are adverse problems with the human body posture. In other words, when the target human body risk score exceeds the preset risk threshold, it indicates that the current posture has a health risk and needs to be corrected.

[0051] In specific implementation, the specific value of the preset risk threshold is set according to actual needs and is not limited in this embodiment. For example, in one embodiment, the specific value of the preset risk threshold can be determined by combining historical posture data in a medical database.

[0052] Furthermore, the preset risk thresholds corresponding to different human joints are different. In specific implementation, they can be set according to the importance and health risks of different human joints in the posture. For example, in one embodiment, the preset health risk thresholds are: 2 points for the neck, 3 points for the torso, 2 points for the legs, 3 points for the upper arms, 2 points for the forearms, and 2 points for the wrists.

[0053] In this way, by setting different preset risk thresholds, the actual risks of each joint can be more accurately reflected, providing a more refined and personalized posture assessment and improving the accuracy and reliability of the overall assessment.

[0054] In this step, when the human body risk score exceeds the preset risk threshold, a composite message is generated. The composite message is used to convey a message to the person being measured to prompt the person being measured about the posture problem. It should be noted that the composite message sends reminders based on multiple senses simultaneously, and the specific multi-sensory reminder method can be selected according to actual needs and is not limited in this embodiment.

[0055] Optionally, the composite message includes at least a visual message, an auditory message, and a tactile message.

[0056] Specifically, the visual message reminds the person being measured visually. In specific implementation, for example, in one embodiment, different videos can be pre-recorded for different joints, and the person being measured wears a mixed reality glasses device during the training process. When the target human body risk score of a certain joint exceeds the preset risk threshold, the corresponding video will be displayed on the mixed reality glasses device, enabling the person being measured to receive visual messages about adjusting the posture and movements in real time, thus achieving the reminder function.

[0057] Further, the auditory message is used to remind the person to be measured audibly. Specifically, in one implementation, for example, different audio can be pre-recorded for different joints, and the person to be measured wears a wireless headphone device during the training process. When the target human risk score of a certain joint exceeds the preset risk threshold, the corresponding audio is played on the wireless headphone device, so that the person to be measured can receive the auditory message about adjusting the posture and movement in real time, thus achieving the reminder function.

[0058] Further, the tactile message is used to remind the person to be measured tactilely. Specifically, in one implementation, for example, a tactile device including a single-chip microcomputer and a micro vibration motor is preset, and the tactile device is fixed to each joint through a strap. When the target human risk score of a certain joint exceeds the preset risk threshold, the control signal of the tactile device is packetized at the host computer end and transmitted to the single-chip microcomputer through serial communication. The data is unpacked at the single-chip microcomputer end, and then the micro vibration motor corresponding to the joint is driven to vibrate, thus achieving the reminder function.

[0059] Further, one or more reminder methods can be used simultaneously to form a composite message to remind the person to be measured.

[0060] The method provided in this embodiment uses a composite message that combines multiple methods based on vision, audition, and touch for reminder, which can be applied to various environments and scenarios, without being limited by a single sensory feedback. At the same time, the user can choose the most suitable reminder method according to personal preferences and usage environments, reducing the interference of a single reminder method to the user and improving the user experience and satisfaction.

[0061] Further, the posture correction notice can include the specific part that needs to be corrected and the specific correction method. When the person to be measured receives the posture correction notice, the posture can be corrected according to the information in the posture correction notice.

[0062] The posture correction training method provided by this application for human-centered intelligent manufacturing first obtains the coordinates of target joint points, inputs the coordinates of the target joint points into the human digital model to obtain the corresponding relationship between the coordinates of the target joint points and human joints, and then calculates based on the vector relationship of the coordinates of the target joint points to obtain the inter-joint angle and joint torsion angle of the coordinates of the target joint points. Then, based on the corresponding relationship, the inter-joint angle, and the joint torsion angle, the target human risk score is determined. Finally, when the target human risk score exceeds the preset risk threshold, a posture correction notification is sent using a composite message. In this way, through the human digital model, each joint position and posture of the human body can be accurately captured and analyzed. Further, by calculating the inter-joint angle and joint torsion angle, whether the posture is incorrect and the degree of error can be quantified. Then, combined with the corresponding relationship, the human risk score is determined. The posture is judged through the human risk score, and a reminder is given when the preset risk threshold is exceeded. In this way, the accuracy and credibility of the correction training are ensured. Further, by using a multi-sensory composite message reminder, attention can be more effectively attracted, so that the posture can be adjusted in time, improving the user experience.

[0063] Optionally, when the target human risk score is less than the preset risk threshold, the method further includes:

[0064] Return to the process of obtaining the coordinates of the target joint points and recalculate the target human risk score.

[0065] Specifically, when the target human risk score is less than the preset risk threshold, it means that the current posture is within the safe range and no posture correction is required. At this time, return to the step of obtaining the joint point coordinates again to continuously monitor the user's posture.

[0066] The method provided in this embodiment can continuously monitor and dynamically adjust by returning to obtain the joint point coordinates again and recalculating the risk score, ensuring that the user always maintains a healthy posture and reducing unnecessary interference, guaranteeing the reliability of the posture correction.

[0067] Optionally, the method for constructing the human digital model includes:

[0068] (1) Obtain the standard joint point coordinates in the standard posture and calculate the posture data of the standard joint point coordinates.

[0069] Specifically, the posture representing a healthy and ideal posture for human training can determine the standard posture through a medical database. It can be understood that the coordinates of the joint points in the standard posture are the standard joint point coordinates.

[0070] In this step, based on the collected standard joint point coordinates, the posture data of each joint point is calculated, and the posture data of each joint point can be characterized by the inter-joint angle and the joint torsion angle.

[0071] (2) Correct the posture data and construct the human digital model based on the corrected posture data.

[0072] Specifically, correct the posture data to eliminate measurement errors and noise, and ensure the accuracy and reliability of the posture data. In specific implementation, after correcting the posture data, it can be standardized to conform to a unified scale and format, which is convenient for subsequent model construction.

[0073] Optionally, the human digital model is created and deployed on the Unity3D platform; the correcting the posture data and constructing the human digital model based on the corrected posture data includes:

[0074] Use sphere elements to represent the posture data, and draw the connection lines of the standard joint point coordinates through the line renderer parameters.

[0075] Specifically, the human digital model can include 25 joint points, which is convenient for matching with the joint position and pose data. The human digital model is created and deployed on the Unity3D platform, and the connection lines between the joint points are drawn by editing the LineRenderer parameters of the joint points.

[0076] The method provided in this embodiment constructs a human digital model through the posture data in the standard posture, can accurately analyze and monitor the human posture, provides a solid foundation for posture correction through the constructed human digital model, and ensures the effect of posture correction training.

[0077] Figure 2 It is the flowchart of the second embodiment of the posture correction training method applied to human intelligence manufacturing provided by this application. Please refer to Figure 2 , the method provided in this embodiment may include:

[0078] S201. Power on the depth camera, reminder device and related devices.

[0079] In specific implementation, the specific models of the depth camera, reminder device and related devices can be selected according to actual needs, and are not limited in this embodiment. For example, after steps such as power connection, device startup, system initialization, device calibration and function test, the depth camera, reminder device and related devices can be powered on.

[0080] It should be noted that the wireless transmission module is responsible for the communication between each device.

[0081] S202. Detect whether the depth camera captures the person to be measured, and detect whether the reminder device and the related devices are successfully connected.

[0082] In specific implementation, the depth image data can be analyzed through an image processing algorithm to identify and extract the human body contour and joint points. After the human body contour and joint points are extracted, it is determined that the measured person is captured.

[0083] Furthermore, it is possible to detect whether the reminder device responds as expected through a test signal. For example, in one embodiment, it is determined whether the reminder device and related devices are successfully connected by judging whether the display shows a test image, whether the speaker plays test audio, and whether the vibration device generates vibration.

[0084] It should be noted that if the reminder device and related devices are not successfully connected, the power-on process is repeated until the connection is successful.

[0085] S203. After the reminder device and the related devices are successfully connected, obtain the coordinates of the target joint points, input the coordinates of the target joint points into the human digital model, and obtain the corresponding relationship between the coordinates of the target joint points and the human joints; the human digital model is used to match the coordinates of the target joint points and the real pose data of the human body.

[0086] In specific implementation, the relevant explanations of this step can be referred to the description in the previous embodiments, and will not be elaborated here.

[0087] S204. Calculate the target distance between the coordinates of the corresponding target joint point pairs in two adjacent frames of the depth image data.

[0088] Specifically, the target joint point pair is two adjacent and connected nodes of the human body in one frame of the image, and the straight-line distance between these two nodes is the target distance.

[0089] In specific implementation, for example, in one embodiment, the depth images of two adjacent frames are the nth frame and the (n - 1)th frame respectively, and the target distance is calculated according to the coordinates of the target joint point pair:

[0090] ;

[0091] where L n is the target distance of the nth frame, x n , y n , z n are the three-axis coordinates of one joint point in the target joint point pair, and x n ´, y n ´, z n ´ are the three-axis coordinates of the other joint point in the target joint point pair.

[0092] S205. When the target distance is not equal to the preset error value, correct the coordinates of the target joint point pair in the depth image data.

[0093] Specifically, the specific value of the preset error value is set according to actual needs. In this embodiment, it is not limited herein.

[0094] In specific implementation, when the target distance is greater than the preset error value, the relationship between the target distance and the preset error value can be determined based on the following formula:

[0095] ;

[0096] wherein, L n is the target distance of the nth frame, L n-1 is the target distance of the (n - 1)th frame, and ξ is the preset error value.

[0097] Furthermore, when the target distance is greater than the preset error value, the coordinates of the target joint pair in the depth image data are corrected based on the following formula:

[0098] ;

[0099] where A n is the three-dimensional coordinate value of the joint after correction in the nth frame, A n-1 and A n-1-σ are the initial coordinate values of the joint in the (n - 1)th frame and the (n - 1 - σ)th frame, respectively.

[0100] Furthermore, if the difference between the target distance of the nth frame and the target distance of the (n - 1)th frame is less than the maximum allowable error value, the coordinate value of the joint is modified to be equal to the initial coordinate value.

[0101] The method provided in this embodiment corrects the joint coordinates according to the physiological and kinematic law that the distance between the joint points connected by the human joints does not change with human movement, and can eliminate the measurement inaccuracy caused by noise, interference, or device error, thereby improving the accuracy of the posture correction training.

[0102] S206. Calculate according to the vector relationship of the target joint coordinates to obtain the inter-joint angle and joint torsion angle of the target joint coordinates.

[0103] In some embodiments, the modified coordinate values are used to calculate all joint vectors and joint angles included in the kinematic model of the nth frame, and the calculation formula is as follows:

[0104] ;

[0105] wherein, is the joint vector, , , are the three-dimensional coordinate values of the child nodes after correction, , , is the three - dimensional coordinate value after the correction of the parent node, , , are the unit vectors of the x - axis, y - axis, and z - axis respectively;

[0106] ;

[0107] Among them, is the angle between joints, and are the two joint vectors for calculating the joint angle;

[0108] The calculation method of the joint twist angle includes:

[0109] Find the rotation angle according to the quaternions of two joint points, and the calculation formula of the twist angle is as follows:

[0110] ; ;

[0111] ;

[0112] ;

[0113] Among them, q, q´, Q, Q´ are the quaternions of the joint points;

[0114] Calculate the rotation matrix based on the following formula:

[0115] ;

[0116] Among them, Q, Q´ are the quaternions of the joint points, and R is the rotation matrix;

[0117] Calculate the rotation angle based on the following formula:

[0118] ;

[0119] Among them, q, q´, Q, Q´ are the quaternions of the joint points, R is the rotation matrix, and θ is the rotation angle.

[0120] It should be noted that when calculating other joint twist angles using the measured quaternion values, the calculation process is the same as above.

[0121] S207. Determine the target human body risk score based on the corresponding relationship, the angle between joints, and the joint twist angle; among them, the target human body risk score characterizes the error degree of the human joints; the target human body risk score includes the expert evaluation score.

[0122] In specific implementation, the relevant explanations of this step can refer to the description in the previous embodiments, and will not be elaborated here.

[0123] S208. When the target human body risk score exceeds a preset risk threshold, a posture correction notification is sent using a composite message; wherein, the composite message sends a reminder based on multiple senses simultaneously.

[0124] For specific implementation, the relevant explanations of this step can be referred to the descriptions in the previous embodiments and will not be elaborated here.

[0125] S209. Record each of the composite messages and obtain the degree of posture correction in each of the composite messages; wherein, the degree of posture correction represents the degree to which the composite message reminds the target human body posture to deviate.

[0126] Specifically, the degree of posture correction can quantify the gap between a person's current posture and the standard posture. In specific implementation, the degree of posture correction can be determined by the degree of the posture that the person to be measured needs to adjust in the composite message.

[0127] S210. Combine the number of the composite messages, the degree of posture correction, and relevant parameters to calculate the posture improvement effect of the target human body.

[0128] Specifically, the number of composite messages can reflect the frequency of the user's posture deviation from the standard; the degree of posture correction can determine the user's posture training habit; relevant parameters such as timestamps are used to judge the posture improvement effect for the effect judgment of this training.

[0129] The posture correction training method applied to humanoid manufacturing provided by this application can accurately capture and analyze the position and posture of each joint of the human body through a human digital model. Further, by calculating the angle between joints and the joint torsion angle, whether the posture is incorrect and the degree of error can be quantified, and then the human body risk score is determined in combination with the corresponding relationship. The posture is judged through the human body risk score, and a reminder is given when the preset risk threshold is exceeded. In this way, the accuracy and credibility of the correction training are ensured. Further, by using a multi-sensory composite message reminder for reminder, it can more effectively attract attention, so as to timely adjust the posture and improve the user experience.

[0130] Corresponding to the foregoing embodiment of a posture correction training method applied to humanoid manufacturing, this application also provides an embodiment of a posture correction training device.

[0131] An embodiment of the posture correction training device of the present application can be applied to a posture correction training device. The device embodiment can be implemented through software, or through hardware or a combination of software and hardware. Taking software implementation as an example, as a logically meaningful device, it is formed by the processor of the posture correction training device where it is located reading the corresponding computer program instructions in the non-volatile memory into the memory for operation. From the hardware level, as Figure 3 shown, it is a hardware structure diagram of the posture correction training device where the posture correction training device of the present application is located. In addition to Figure 3 the processor, memory, network interface, and non-volatile memory shown, the posture correction training device where the device is located in the embodiment usually includes other hardware according to the actual function of the posture correction training device, which will not be elaborated here.

[0132] Figure 4 It is a schematic structural diagram of Embodiment 1 of the posture correction training device provided by the present application. Please refer to Figure 4 , the device provided in this embodiment includes an acquisition module 410, a calculation module 420, a determination module 430, and a notification module 440; wherein,

[0133] The acquisition module 410 is configured to acquire the coordinates of target joint points, input the coordinates of the target joint points into a human digital model, and obtain the corresponding relationship between the coordinates of the target joint points and human joints; the human digital model is used to match the coordinates of the target joint points and the real posture data of the human body;

[0134] The calculation module 420 is configured to calculate according to the vector relationship of the coordinates of the target joint points to obtain the inter-joint angle and joint torsion angle of the coordinates of the target joint points;

[0135] The determination module 430 is configured to determine a target human body risk score based on the corresponding relationship, the inter-joint angle, and the joint torsion angle; wherein, the target human body risk score represents the error degree of the human joints; the target human body risk score includes an expert evaluation score;

[0136] The notification module 440 is configured to send a posture correction notification using a composite message when the target human body risk score exceeds a preset risk threshold; wherein, the composite message sends a reminder based on multiple senses simultaneously.

[0137] The device of this embodiment can be used to execute the steps of the method embodiment shown in Figure 1 , and the specific implementation principle and implementation process are similar and will not be elaborated here.

[0138] Optionally, the acquisition module 410 is specifically configured to collect depth image data of the person to be measured;

[0139] The obtaining module 410 is further specifically configured to process the depth image data based on a human body bone point recognition algorithm to obtain the coordinates of the target joint points.

[0140] Optionally, the obtaining module 410 is further specifically configured to power on the depth camera, the reminder device, and related devices.

[0141] The obtaining module 410 is further specifically configured to detect whether the depth camera captures the person to be measured, and detect whether the reminder device and the related devices are successfully connected.

[0142] The obtaining module 410 is further specifically configured to repeat the power-on process when the reminder device and the related devices are not successfully connected.

[0143] Optionally, the determining module 430 is further configured to return to the process of obtaining the coordinates of the target joint points and recalculate the target human body risk score.

[0144] Optionally, the composite message at least includes a visual message, an auditory message, and a tactile message.

[0145] Optionally, the obtaining module 410 is further specifically configured to obtain the standard joint point coordinates in a standard posture and calculate the pose data of the standard joint point coordinates.

[0146] The obtaining module 410 is further specifically configured to correct the pose data and construct the human digital model based on the corrected pose data.

[0147] Optionally, the notification module 440 is further configured to record each composite message and obtain the degree of pose correction in each composite message; wherein, the degree of pose correction represents the degree to which the composite message reminds the deviation of the target human body pose.

[0148] The notification module 440 is further configured to calculate the pose improvement effect of the target human body in combination with the number of the composite messages, the degree of pose correction, and related parameters.

[0149] Optionally, the obtaining module 410 is further specifically configured to calculate the target distance between the coordinates of the corresponding target joint pairs in two adjacent frames of the depth image data.

[0150] The obtaining module 410 is further specifically configured to correct the coordinates of the target joint pairs in the depth image data when the target distance is not equal to a preset error value.

[0151] Optionally, the obtaining module 410 is further specifically configured to represent the pose data by spherical elements and draw the connection lines of the standard joint point coordinates through the line renderer parameters.

[0152] Please continue to refer to Figure 3 , this application also provides a posture correction training device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the steps of any of the methods provided in the first aspect of this application are implemented.

[0153] This application also provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the steps of any of the methods provided in this application are implemented.

[0154] For the specific implementation process of the functions and roles of each unit in the above device, please refer to the implementation process of the corresponding steps in the above method, which will not be elaborated here.

[0155] For the device embodiment, since it basically corresponds to the method embodiment, the relevant parts can be referred to the partial description of the method embodiment. The device embodiments described above are only illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this application. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0156] The above are only the preferred embodiments of this application and are not intended to limit this application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of this application shall be included within the scope of protection of this application.

Claims

1. A posture correction training method applied to human-oriented intelligent manufacturing, characterized in that: The method comprises: Acquire the target joint point coordinates, map the target joint point coordinates to the human body digital model, so as to obtain the corresponding relationship between each target joint point coordinate and the real joint of the human body, and construct the human body digital model through the posture data under the standard posture; the human body digital model is used to match the target joint point coordinates and the real posture data of the human body; The step of obtaining the target joint point coordinates includes: Collecting depth image data of the person being measured; Processing the depth image data based on a human skeleton point recognition algorithm to obtain the target joint point coordinates; After obtaining the target joint point coordinates, the method further includes: Calculating the target distance of the coordinates of the corresponding target joint point pairs in the depth image data of two adjacent frames; When the target distance is greater than the preset error value, the coordinates of the target joint point pair are corrected based on the following formula: ； Among them A n is the corrected three-dimensional coordinate value of the joint point in the nth frame, n is the number of frames of the current depth image data, n is a positive integer greater than 2, A n-1 and A n-1-σ are the initial coordinate values ​​of the joint point in the n-1th frame and the n-1-σth frame, respectively, σ is any frame number before the current depth image data, and σ is a positive integer greater than 1 and less than n; Calculate according to the vector relationship of the target joint point coordinates to obtain the joint angle and joint torsion angle of the target joint point coordinates; wherein the three-dimensional coordinates of the joint are obtained through the target joint point coordinates, and the corresponding joint angle and joint torsion angle are calculated through the coordinates; Determining a target human body risk score based on the corresponding relationship, the angle between the joints and the joint torsion angle; wherein the target human body risk score represents the degree of error of the human body joint; the target human body risk score includes an expert evaluation score; When the target human body risk score exceeds a preset risk threshold, a composite message is used to issue a posture correction notification; wherein the composite message is based on multi-sensory simultaneous reminders.

2. The method according to claim 1, characterized in that The method for calculating the angle between the joints includes: Use the corrected coordinate values ​​to calculate all joint vectors and joint angles contained in the kinematic model of the nth frame. The calculation formula is as follows: ； in, is the joint vector, , , is the corrected three-dimensional coordinate value of the child node, , , is the corrected three-dimensional coordinate value of the mother node, , , are the unit vectors of the x-axis, y-axis, and z-axis respectively; ； Among them, the is the angle between joints, and stated The two joint vectors for calculating the joint angle; The joint torsion angle calculation method comprises: The rotation angle is calculated based on the quaternion of the two joint points. The calculation formula of the torsion angle is as follows: ； ； ； ； Wherein, q, q´, Q, and Q´ are the quaternions of the joint points, q1, q2, and q3 are the three imaginary parts of q, and q1´, q2´, and q3´ are the three imaginary parts of q´; The rotation matrix is ​​calculated based on the following formula: ； Wherein, the Q and Q´ are the quaternions of the joint points, and the R is the rotation matrix; The rotation angle is calculated based on the following formula: ； Among them, q, q´, Q, and Q´ are quaternions of the joint points, R is the rotation matrix, and θ is the rotation angle.

3. The method according to claim 1, characterized in that When the target human body risk score is less than a preset risk threshold, the method further includes: Return to the process of obtaining the target joint point coordinates, and recalculate the target human body risk score.

4. The method according to claim 1, characterized in that: The composite message includes at least a visual message, an auditory message and a tactile message.

5. The method according to claim 1, characterized in that The method for constructing the human body digital model comprises: Obtaining standard joint point coordinates under a standard posture, and calculating posture data of the standard joint point coordinates; The posture data is corrected, and the human body digital model is constructed based on the corrected posture data.

6. The method according to claim 1, characterized in that After sending a posture correction notification using a composite message when the target human body risk score exceeds a preset risk threshold, the method further includes: Recording each of the composite messages, and obtaining a posture correction degree in each of the composite messages; wherein the posture correction degree represents a degree to which the composite message reminds the target human body of a posture deviation; The posture improvement effect of the target human body is calculated by combining the number of the composite messages, the posture correction degree and related parameters.

7. The method according to claim 1, characterized in that After the depth image data is processed based on the human skeleton point recognition algorithm to obtain the coordinates of the target joint point, the method further includes: Calculate the target distance L of the coordinates of the corresponding target joint point pairs in the depth image data of two adjacent frames n ; ； Among them, L n is the target distance in the nth frame, x n ,y n 、z n is the three-axis coordinate of a joint point in the target joint point pair, x n ´,y n ´, z n ´ is the three-axis coordinate of another joint point in the target joint point pair; When the target distance is not equal to a preset error value, the coordinates of the target joint point pair in the depth image data are corrected.

8. The method according to claim 7, characterized in that The step of correcting the coordinates of the target joint point pair in the depth image data when the target distance is not equal to the preset error value includes: When the target distance difference is less than the maximum allowable error value, the coordinate value of the joint point is modified to be equal to the initial coordinate value.

9. A posture correction training device, characterized in that: The device includes an acquisition module, a calculation module, a determination module and a notification module; wherein, The acquisition module is used to acquire the target joint point coordinates, map the target joint point coordinates to the human body digital model to obtain the corresponding relationship between each target joint point coordinate and the real joint of the human body, build the human body digital model through the posture data under the standard posture, input the target joint point coordinates into the human body digital model, and obtain the corresponding relationship between the target joint point coordinates and the human body joints; the human body digital model is used to match the target joint point coordinates with the real posture data of the human body; The step of obtaining the target joint point coordinates includes: Collecting depth image data of the person being measured; Processing the depth image data based on a human skeleton point recognition algorithm to obtain the target joint point coordinates; The acquisition module is further used to calculate the target distance of the coordinates of the target joint point pair corresponding to the depth image data of two adjacent frames; when the target distance is greater than the preset error value, the coordinates of the target joint point pair are corrected based on the following formula: ； Among them A n is the corrected three-dimensional coordinate value of the joint point in the nth frame, n is the number of frames of the current depth image data, n is a positive integer greater than 2, A n-1 and A n-1-σ are the initial coordinate values ​​of the joint point in the n-1th frame and the n-1-σth frame, respectively, σ is any frame number before the current depth image data, and σ is a positive integer greater than 1 and less than n; The calculation module is used to calculate according to the vector relationship of the target joint point coordinates to obtain the joint angle and joint torsion angle of the target joint point coordinates; wherein the three-dimensional coordinates of the joint are obtained through the target joint point coordinates, and the corresponding joint angle and joint torsion angle are calculated through the coordinates; The determination module is used to determine a target human body risk score based on the corresponding relationship, the angle between the joints and the joint torsion angle; wherein the target human body risk score represents the error degree of the human body joint; and the target human body risk score includes an expert evaluation score; The notification module is used to send a posture correction notification using a composite message when the target human body risk score exceeds a preset risk threshold; wherein the composite message sends reminders based on multiple senses at the same time.

Images