WO2021036635A1 - 基于振动检测的数字孪生智能健康预测方法及装置 - Google Patents

基于振动检测的数字孪生智能健康预测方法及装置 Download PDF

Info

Publication number
WO2021036635A1
WO2021036635A1 PCT/CN2020/104822 CN2020104822W WO2021036635A1 WO 2021036635 A1 WO2021036635 A1 WO 2021036635A1 CN 2020104822 W CN2020104822 W CN 2020104822W WO 2021036635 A1 WO2021036635 A1 WO 2021036635A1
Authority
WO
WIPO (PCT)
Prior art keywords
force
user
vibration
video
mobile terminal
Prior art date
Application number
PCT/CN2020/104822
Other languages
English (en)
French (fr)
Inventor
高风波
Original Assignee
深圳市广宁股份有限公司
深圳市豪视智能科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳市广宁股份有限公司, 深圳市豪视智能科技有限公司 filed Critical 深圳市广宁股份有限公司
Publication of WO2021036635A1 publication Critical patent/WO2021036635A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/20—Movements or behaviour, e.g. gesture recognition
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00—Network arrangements or protocols for supporting network services or applications
    • H04L67/01—Protocols
    • H04L67/02—Protocols based on web technology, e.g. hypertext transfer protocol [HTTP]
    • H04L67/025—Protocols based on web technology, e.g. hypertext transfer protocol [HTTP] for remote control or remote monitoring of applications
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00—Network arrangements or protocols for supporting network services or applications
    • H04L67/01—Protocols
    • H04L67/10—Protocols in which an application is distributed across nodes in the network
    • H04L67/1097—Protocols in which an application is distributed across nodes in the network for distributed storage of data in networks, e.g. transport arrangements for network file system [NFS], storage area networks [SAN] or network attached storage [NAS]
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00—Network arrangements or protocols for supporting network services or applications
    • H04L67/01—Protocols
    • H04L67/12—Protocols specially adapted for proprietary or special-purpose networking environments, e.g. medical networks, sensor networks, networks in vehicles or remote metering networks

Definitions

  • This application relates to the field of Internet technology, and in particular to a digital twin intelligent health prediction method and device based on vibration detection.
  • Internet+ is a new format of Internet development under Innovation 2.0.
  • Internet+ means “Internet + various traditional industries", but this is not a simple addition of the two. Instead, it uses information and communication technologies and Internet platforms to deeply integrate the Internet with traditional industries and create a new development ecology. It represents a new social form, that is, to give full play to the optimization and integration role of the Internet in the allocation of social resources, to deeply integrate the innovative achievements of the Internet into economic and social domains, to enhance the innovation and productivity of the entire society, and to form A broader new form of economic development using the Internet as an infrastructure and realization tool.
  • the traditional health monitoring mechanism generally uses localized detection equipment. For example, when a user wears a bracelet during exercise, the bracelet can collect the user’s body data during exercise, and perform localized vibration detection and health prediction through the bracelet. However, it is necessary to purchase a bracelet with vibration detection function, and because the bracelet can only be worn on the hand, it cannot detect other parts of the body. The detection area is small, and it is difficult to satisfy the increasing number of users in various sports scenes. Health forecast demand.
  • the embodiment of the application provides a digital twin intelligent health prediction method and device based on vibration detection, which performs vibration detection on a user’s fitness action, and obtains the risk score of the user’s current action through the digital twin intelligent health prediction model, thereby realizing real-time monitoring of the user’s fitness When the current action is dangerous.
  • the data transmission process in the vibration detection method disclosed in the embodiments of the present application can be based on Internet + technology to form a distributed intelligent vibration detection system of local + cloud or server.
  • the local collection device can perform accurate and original vibration detection. Image collection and preprocessing.
  • the cloud or server can be based on the distributed data obtained, combined with various special health prediction models obtained through statistical analysis of big data technology, to predict the health status of the detected target, and realize the Internet and traditional health
  • the in-depth integration of the monitoring industry improves the intelligence and accuracy of health monitoring, and meets the needs for intelligent health prediction in an increasing number of various sports scenes.
  • the first aspect of the present application provides a digital twin intelligent health prediction method based on vibration detection, the method is applied to a mobile terminal, the mobile terminal includes a camera, and the method includes:
  • the force-receiving part is within the shooting range of the mobile terminal, acquiring a vibration video of the force-receiving part according to the motion video;
  • the first force-receiving condition of the force-receiving part is obtained from the digital twin intelligent health prediction model, which stores the different force-receiving conditions of different body parts of the user Simulated vibration information under;
  • the first risk score of the current action of the user is determined.
  • a second aspect of the present application provides a digital twin intelligent health prediction device based on vibration detection, the device including:
  • the collection unit is used to collect the user's exercise video when the health prediction instruction is received;
  • An identification unit configured to identify the force-bearing part corresponding to the user's current action based on the user's motion video
  • a judging unit for judging whether the force-receiving part is within the shooting range of the mobile terminal
  • the first determining unit is configured to determine the first vibration information of the force-receiving part according to the vibration video of the force-receiving part;
  • the second acquiring unit acquires the first force-receiving condition of the force-receiving part from a digital twin intelligent health prediction model according to the first vibration information, and the digital-twin intelligent health prediction model stores different body parts of the user Simulated vibration information under different stress conditions;
  • a third aspect of the present application provides a mobile terminal.
  • the mobile terminal includes a processor, a memory, a communication interface, and one or more programs.
  • the one or more programs are stored in the memory and configured Executed by the processor, the program includes instructions for executing the steps in any of the above methods.
  • the fourth aspect of the present application provides a computer-readable storage medium that stores a computer program for electronic data exchange, wherein the computer program causes a computer to execute the part described in any of the above methods Or all steps.
  • the vibration detection-based digital twin intelligent health prediction method and device provided in this application, when a health prediction instruction is received, the user's exercise video is collected, and the force corresponding to the user's current action is identified based on the user's exercise video Part, obtain the vibration video of the stressed part according to the motion video, determine the first vibration information of the stressed part according to the vibration video of the stressed part, and obtain the first vibration information of the stressed part from the digital twin intelligent health prediction model according to the first vibration information.
  • the force status according to the first force status, determines the first risk score of the user's current action.
  • the mobile terminal processes the user’s exercise video to obtain the user’s current action risk score, which realizes real-time monitoring of the user’s current action risk during fitness.
  • the detection area is large, which can meet the user's health prediction needs in various sports scenes.
  • FIG. 1a is a schematic diagram of a digital twin intelligent health prediction system based on vibration detection according to an embodiment of the application;
  • FIG. 1b is a flowchart of a digital twin intelligent health prediction method based on vibration detection according to an embodiment of the application;
  • FIG. 2 is a schematic diagram of a digital twin intelligent health prediction device based on vibration detection provided by an embodiment of the application;
  • FIG. 3 is a schematic diagram of a mechanical model of a part provided by an embodiment of the application.
  • FIG. 4 is a schematic structural diagram of a mobile terminal provided by an embodiment of this application.
  • FIG. 5 is a schematic diagram of a user's elbow movement provided by an embodiment of the application.
  • the embodiment of the application provides a digital twin intelligent health prediction method and device based on vibration detection, which performs vibration detection on a user’s fitness action, and obtains the risk score of the user’s current action through the digital twin intelligent health prediction model, thereby realizing real-time monitoring of the user’s fitness When the current action is dangerous.
  • a mobile terminal collects the vibration video of the user's force-bearing part during the fitness process, and processes the vibration video of the force-bearing part.
  • the vibration information is obtained, and the risk score of the user's current action is obtained through the digital twin intelligent health prediction model.
  • the mobile terminal processes the user’s exercise video to obtain the user’s current action risk score, which realizes real-time monitoring of the user’s current action risk during fitness.
  • the detection area is large, which can meet the user's health prediction needs in various sports scenes.
  • FIG. 1a is a schematic diagram of a digital twin intelligent health prediction system based on vibration detection provided by an embodiment of the application.
  • the health prediction system includes a mobile terminal and a user.
  • the mobile terminal can be a mobile phone, a tablet, or a notebook. With computers, mobile Internet devices, or other types of terminal devices with cameras, users can perform various fitness actions.
  • the mobile terminal includes a camera, and the mobile terminal can capture real-time actions of the user through the camera.
  • the mobile terminal 101 is a mobile phone
  • the mobile phone includes a rear camera.
  • the rear camera has a certain shooting range.
  • the fitness action of the user 102 is push-ups, in which:
  • the mobile terminal 101 is configured to collect the exercise video of the user 102 when receiving a health prediction instruction; identify the force-receiving part corresponding to the current action of the user 102 based on the motion video of the user 102; and determine whether the force-receiving part is Is located within the shooting range of the mobile terminal 101; if the force-bearing part is within the shooting range of the mobile terminal 101, obtain a vibration video of the force-bearing part according to the motion video; according to the force-bearing part Determine the first vibration information of the stressed part according to the vibration video; obtain the first force status of the stressed part from the digital twin smart health prediction model according to the first vibration information, and the digital twin smart health prediction
  • the model stores simulated vibration information of different body parts of the user 102 under different stress conditions; according to the first stress condition, the first risk score of the current action of the user 102 is determined.
  • the mobile terminal 101 when the mobile terminal 101 receives the health prediction instruction, it collects the exercise video of the user 102 through the camera, that is, the exercise video of the user 102 performing push-ups, and the mobile terminal 101 recognizes based on the video of the user 102 performing push-ups
  • the current action of the user 102 is a push-up, and then it is determined that the force-bearing part corresponding to the current action of the user 102 is an elbow joint.
  • the mobile terminal 101 determines that the elbow joint is within the shooting range, and then obtains a vibration video of the elbow joint.
  • the mobile terminal 101 Because the user 102 is performing a push-up When the elbow joints of both arms are involved, the mobile terminal 101 separately obtains the vibration videos of the elbow joints of both arms, and then judges the risk score of the current push-ups of the user 102. If the frequency of the push-ups of the user 102 is too high, then the elbow The vibration frequency of the joints will also be too high, so the risk score obtained is relatively high. The mobile terminal 101 can issue an alarm or prompt to the user 102 in time, prompting the user 102 to reduce the frequency or temporarily stop exercising.
  • the mobile terminal processes the user’s exercise video to obtain the user’s current action risk score, which realizes real-time monitoring of the user’s current action risk during fitness.
  • the detection area is large, which can meet the user's health prediction needs in various sports scenes.
  • Figure 1b is a flowchart of a digital twin smart health prediction method based on vibration detection provided by an embodiment of the application.
  • the digital twin smart health prediction method based on vibration detection provided in this embodiment of the application is applied to a mobile terminal ,
  • the mobile terminal includes a camera.
  • a digital twin intelligent health prediction method based on vibration detection provided by an embodiment of the present application may include:
  • the mobile terminal may be a mobile phone, a tablet computer, a notebook computer, a mobile Internet device, or other types of terminal devices with a camera.
  • the mobile phone interface displays the fitness strain prevention function entrance. After the user clicks to enter the fitness strain prevention function, the mobile phone turns on the camera to collect real-time fitness images for the user.
  • the current action of the user, and the force-receiving part corresponding to the action is identified. For example, when the user's current motion is push-ups, the force-receiving part is the arm, and when the user's current motion is sit-ups, the force-receiving part is the abdomen.
  • the mobile terminal is placed in a different position or the angle at which the mobile terminal photographs the user is different, there are cases where the mobile terminal does not photograph the user's force-receiving part, so it is necessary to determine whether the force-receiving part is within the shooting range of the mobile terminal.
  • the method for judging whether the stressed part is within the shooting range of the mobile terminal may be:
  • the force-receiving part is within the shooting range of the mobile terminal, obtain a vibration video of the force-receiving part according to the motion video.
  • the location of the force-receiving part is determined in the motion video, and then the vibration video of the force-receiving part is selected from the motion video.
  • the Lagrangian motion amplification method is used to process the vibration video of the stressed part to obtain a vibration amplified video, and the vibration amplified video is processed to obtain the first vibration information of the stressed part.
  • the first vibration information obtain the first force status of the force-bearing part from a digital twin intelligent health prediction model, and the digital twin intelligent health prediction model stores the different body parts of the user in different stress conditions. Simulated vibration information under force conditions.
  • digital twin technology as the core technology to realize the interactive integration between the physical world of manufacturing and the information world, is another technological trend besides artificial intelligence, machine learning, AR/VR, and blockchain.
  • Digital twin technology dynamically presents the past and current behaviors or processes of a physical entity in a digital form.
  • digital twin technology provides more real-time, efficient, and intelligent services in the practice of intelligent manufacturing concepts and goals.
  • the digital twin smart health prediction model runs synchronously with the mobile terminal, and the digital twin smart health prediction model simulates the movement of the human body, and can timely predict whether the moving parts of the human body are easily strained during the fitness process.
  • the digital twin intelligent health prediction method of vibration detection combines spatial graphics processing with digital twin technology to realize the combination of computer vision technology and artificial intelligence, and specifically determines human movement through vibration detection.
  • the digital twin intelligent health prediction model needs to be constructed in advance. For different users, the user's height, weight, gender, age, physical condition, and environment may be different, so when the digital twin intelligent health prediction model is constructed in advance , First need to obtain the user's human body parameters through the mobile terminal, input the user's human body parameters to generate a digital twin smart health prediction model. After the construction is completed, the digital twin smart health prediction model simulates the vibration information of the user's different body parts under different stress conditions , And store the simulated vibration information in the digital twin smart health prediction model.
  • different force-receiving parts of the user have different force-receiving ranges and force-receiving frequencies. If the force is too large or the force-receiving frequency is too high, it is likely to cause strain on the force-receiving part of the user or damage to the user's body. According to the first force status, the risk score of the user's current action can be determined.
  • the digital twin smart health prediction method based on vibration detection provided by the embodiments of the present application, when a health prediction instruction is received, the user's exercise video is collected, and the force-bearing part corresponding to the user's current action is identified based on the user's exercise video.
  • Obtain the vibration video of the force-receiving part according to the motion video determine the first vibration information of the force-receiving part according to the vibration video of the force-receiving part, and obtain the first force of the force-receiving part from the digital twin intelligent health prediction model according to the first vibration information Status, according to the first force status, determine the first risk score of the user's current action.
  • the mobile terminal processes the user’s exercise video to obtain the user’s current action risk score, which realizes real-time monitoring of the user’s current action risk during fitness.
  • the detection area is large, which can meet the user's health prediction needs in various sports scenes.
  • Another embodiment of the present application provides another digital twin smart health prediction method based on vibration detection, which may include:
  • the mobile terminal when the user is exercising, he can turn on the health prediction function of the mobile terminal.
  • the mobile terminal receives the health prediction instruction, it turns on the camera to collect the user's exercise video.
  • the mobile terminal may be a mobile phone, a tablet computer, a notebook computer, a mobile Internet device, or other types of terminal devices with a camera.
  • the user collects real-time images of the user through the mobile terminal, and recognizes the user's current movement based on the collected movement video of the user, so as to determine the force-bearing part corresponding to the user's current movement.
  • the mobile phone interface displays the fitness strain prevention function entrance. After the user clicks to enter the fitness strain prevention function, the mobile phone turns on the camera to collect real-time fitness images for the user. The current action of the user, and the force-receiving part corresponding to the action is identified.
  • the mobile terminal is placed in a different position or the angle at which the mobile terminal photographs the user is different, there are cases where the mobile terminal does not photograph the user's force-receiving part, so it is necessary to determine whether the force-receiving part is within the shooting range of the mobile terminal.
  • the method for judging whether the force-receiving part is within the shooting range of the mobile terminal may be: comparing the feature points of the force-receiving part with the feature points of each body part of the user in the image frame of the sports video, if If the comparison is successful, it means that the force-receiving part is located in the shooting range of the mobile terminal. If the comparison fails, it means that the force-receiving part is not located in the shooting range of the mobile terminal.
  • the force-receiving part is within the shooting range of the mobile terminal, obtain a vibration video of the force-receiving part according to the motion video.
  • the location of the force-receiving part is determined in the motion video, and then the vibration video of the force-receiving part is selected from the motion video.
  • the vibration video of the stressed part contains the movement process of the stressed part. This movement process is very small and needs to be enlarged for subsequent extraction of vibration information.
  • the small motion can be zoomed in by tracking the motion trajectory and clustering of the target feature points in the video.
  • the method of calibrating the image frames of the vibration video of the stressed part to obtain multiple movement feature points of the temperature may be: selecting at least one frame of the image of the vibration video of the stressed part; The at least one frame of image determines the reference feature points in the vibration video of the force-bearing part that are relatively static during the video capture process; each frame of the image included in the at least one frame of image is intercepted according to N different circle centers.
  • N is an integer greater than 3
  • the target circular partition contains relative motion feature points
  • the relative motion feature point is A motion feature point that performs relative motion with respect to the reference feature point
  • intercept the target circular partition according to a preset window to obtain multiple intercepted partitions.
  • the size and shape of the preset window are determined according to the muscle shape of the force-bearing part ;Sequentially obtain the relative motion feature points whose motion distance is in the preset range from the multiple interception partitions, and accumulate the values of the obtained relative motion feature points, the preset range is the vibration of the force-bearing part in the normal force state Amplitude range: When the accumulated value is not less than the preset value, it is determined that the acquired relative motion feature points are the stable multiple motion feature points.
  • the Lagrangian zoom method to amplify the vibration video, it is first necessary to obtain a number of stable motion feature points in the vibration video, that is, the points of small motion, so as to be compatible with the static points (background points) and violent points in the vibration video.
  • the moving points are distinguished.
  • vibration video in addition to the vibration image, it also includes some relatively static background images.
  • the supporting bar is a relatively static background image
  • the ground is a relatively static background image.
  • the method of processing a plurality of stable motion feature points to obtain the first vibration magnified video may be: tracking the plurality of motion feature points to obtain the Trajectory vector; clustering the trajectory vectors of the multiple motion feature points using a clustering algorithm to obtain the K-type motion layer; obtain the target motion layer that needs to be amplified from the K-type motion layer; in the target motion layer The offset distance of the motion feature point is multiplied by a magnification to obtain an enlarged motion layer; the enlarged motion layer is rendered to obtain the first vibration enlarged video.
  • the trajectory vector uses numerical values to describe the motion direction, motion distance, and brightness change of the motion feature point; then clustering algorithm is used for multiple motion feature points
  • the trajectory vector of the trajectory vector is clustered to obtain the K-type motion layer, and the K-type motion layer is divided according to the correlation and similarity of the trajectory vector, so that different motion layers can contain different types of motion, so as to select the smallest of the K-type motion layers.
  • the motion layer corresponding to the motion is zoomed in to obtain the zoomed motion layer.
  • the image frame corresponding to the target video includes some blank areas due to the enlargement of the motion layer, it is necessary to perform rendering to fill the image frame.
  • F A frame image of a Fourier transform, a Fourier b conjugate signal is transformed frame image, in addition to the formula of the product related to the lower side of the two-mode signals of the Fourier transform. R crosses the cross power spectrum of the calculation result of this step.
  • the cross cross power spectrum After the cross cross power spectrum is obtained, it contains frequency domain noise, so it can be filtered to improve the signal-to-noise ratio, so as to improve the accuracy of the subsequent extracted vibration information.
  • the method further includes:
  • the correlation peak is a frequency domain signal; determine the filtering strategy corresponding to each correlation peak according to the corresponding position of each correlation peak in the vibration video and the frequency band of the correlation peak; The corresponding filtering strategy performs filtering processing on each correlation peak.
  • the cross cross power spectrum reflects the vibration information in the frequency domain, and the vibration information needs to be analyzed in the time domain, and the inverse Fourier transform (or inverse Fourier transform) needs to be performed.
  • the formula used to perform the inverse Fourier transform is:
  • R' is the cross cross power spectrum obtained after filtering
  • r is the vibration information of the pixel in the vibration video.
  • the digital twin intelligent health prediction model stores the simulated vibration information of different body parts of the user under different stress conditions.
  • the method before obtaining the first force status of the force-bearing part from the digital twin smart health prediction model according to the first vibration information, the method further includes: obtaining the user's body parameters.
  • the body parameters include age, gender, and height. Any combination of, weight, body fat rate, heart rate, and blood pressure; construct M parts mechanical models corresponding to the user's M body parts according to human body parameters, where M is a positive integer, M body parts and M parts
  • the mechanical models correspond one-to-one; the radial force is applied to the M designated positions in the M part mechanical models, among which, the M part mechanical models correspond to the M designated positions one-to-one, and the radial force is the simulation of M body parts Force; determine the M moving distances of the M parts of the mechanical model, of which the M parts of the mechanical model correspond to the M moving distances one-to-one; input the M moving distances into the preset dynamic algorithm for calculation to obtain the M parts M simulated vibration information of the mechanical model.
  • the part mechanics model is a virtual body part, that is, the user's body part is virtualized by adopting the structural part mechanics model. Further, a mechanical model of the part is constructed by three-dimensional scanning of the user's body part.
  • the radial force When the radial force is applied to the mechanical model of the part, the radial force needs to be calculated.
  • FIG. 3 is a schematic diagram of a mechanical model of a part according to an embodiment of the application.
  • the unit length is h
  • the entrance radius is R1
  • the exit radius is R2.
  • t be a unit vector acting on this axial element
  • n1 and n2 are the normal unit vectors of the entrance and exit, respectively.
  • the mechanical model is regarded as a cone or cylinder.
  • v1 and v2 be the average value of inlet and outlet speeds respectively.
  • the formula for calculating the radial force on the mechanical model of this part is as follows:
  • CAD computer-aided design
  • the preset dynamics algorithm is:
  • the simulated vibration information stored in the digital twin intelligent health prediction model includes vibration waveform diagrams, modal diagrams, and heat diagrams.
  • the human body contains many body parts, and it is too complicated to analyze the health status of all body parts. Therefore, a single mechanical model can be used to analyze the movement of each body part under different health conditions without considering the movement of these body parts. Instead of measuring the movement of various body parts under different conditions to simulate the energy produced by their interaction. Since the body parts involved in the movement of the human body are basically fixed, only the body parts that are mainly involved in the movement are analyzed, thereby reducing the complexity of modeling.
  • different force-receiving parts of the user have different force-receiving ranges and force-receiving frequencies. If the force is too large or the force-receiving frequency is too high, it is likely to cause strain on the force-receiving part of the user or damage to the user's body. According to the first force status, the risk score of the user's current action can be determined.
  • the first risk score of the user's current action after determining the first risk score of the user's current action according to the first force status, it further includes: when the first risk score is a muscle strain probability score, judging that the muscle strain may be possible Whether the sexual score exceeds the preset probability score; if the muscle strain probability score exceeds the preset probability score, the first warning message is issued, and the first warning message is used to prompt the user to adjust the exercise posture or stop the exercise to avoid the stressed part Strain; when the first risk score is the physical function score, judge whether the physical function score is lower than the preset score; if the physical function score is lower than the preset score, a second warning message is issued, and the second warning message is used to remind The user slows down the exercise frequency or stops exercising to avoid the user's physical injury.
  • the mobile terminal processes the user's exercise video to obtain the user's current movement risk score, which realizes real-time monitoring of the user's current movement risk during the exercise. And give warning to users.
  • the relevant part of the force-bearing part in the shooting range can be determined, or the wide-angle camera mode can be used to determine the relevant part of the force-bearing part in the shooting range. Position, so as to determine the vibration information of the force-bearing part based on the vibration information of the associated part.
  • the force-bearing part is within the shooting range of the mobile terminal, it is also necessary to identify the user’s clothing characteristics to determine whether the vibration video of the force-bearing part can be accurately obtained.
  • the clothing feature covers the vibration characteristics of the force-receiving part, so the vibration video of the force-receiving part cannot be accurately obtained. It is necessary to re-determine the relevant part that is related to the force-receiving part and can accurately obtain the vibration video within the shooting range of the mobile terminal. The location of the related part, and the vibration video of the related part is obtained.
  • the mobile terminal obtains the exercise video of the user during the fitness process, confirms the force-receiving part, and performs different processing according to whether the force-receiving part is within the shooting range of the mobile terminal to obtain the vibration information of the force-receiving part. Compare the vibration information of the force-bearing part with the simulated vibration information of the force-bearing part stored in the digital twin intelligent health prediction model under the unstressed condition to determine the force-bearing condition of the force-bearing part, thereby determining the current action of the user Risk score, and the user is warned based on the risk score.
  • FIG. 5 is a schematic diagram of a user's elbow movement according to an embodiment of the application.
  • the user is doing elbow exercises. For example, when the user exercises the biceps, the elbow will be in a repeated process of bending and straightening. When the user's elbow is straightened, the elbow joint will be as shown in the figure.
  • the elbow joints vibrate regularly, so that the vibration video of the elbow joint can be obtained through the user's mobile phone and other terminals, so as to analyze the vibration video to determine whether the user's exercise is standard, and Whether it will cause elbow joint damage.
  • FIG. 2 is a schematic diagram of a digital twin intelligent health prediction device based on vibration detection according to another embodiment of the application.
  • another embodiment of the present application provides a digital twin intelligent health prediction device based on vibration detection, which may include:
  • the collection unit 201 is configured to collect the user's exercise video when the health prediction instruction is received;
  • the recognition unit 202 is configured to recognize the force-bearing part corresponding to the user's current action based on the user's motion video;
  • the determining unit 203 is configured to determine whether the force-receiving part is located within the shooting range of the mobile terminal;
  • the first acquiring unit 204 is configured to acquire a vibration video of the stressed part according to the motion video if the stressed part is within the shooting range of the mobile terminal;
  • the first determining unit 205 is configured to determine the first vibration information of the stressed part according to the vibration video of the stressed part;
  • the second acquiring unit 206 acquires the first force status of the stressed part from the digital twin smart health prediction model according to the first vibration information, and the digital twin smart health prediction model stores the different bodies of the user Simulated vibration information of parts under different stress conditions;
  • the second determining unit 207 is configured to determine the first risk score of the user's current action according to the first force status.
  • FIG. 4 is a schematic structural diagram of a mobile terminal provided by an embodiment of the application.
  • a mobile terminal provided by an embodiment of the present application may include:
  • the processor 401 such as a CPU.
  • the memory 402 optionally, the memory may be a high-speed RAM memory, or a stable memory, such as a disk memory.
  • the communication interface 403 is used to implement connection and communication between the processor 401 and the memory 402.
  • the structure of the mobile terminal shown in FIG. 4 does not constitute a limitation on the mobile terminal, and may include more or fewer components than shown in the figure, or a combination of certain components, or different component arrangements. .
  • the memory 402 may include an operating system, a network communication module, and a health prediction program.
  • the operating system is a program that manages and controls the hardware and software resources of a mobile terminal, a program that supports health prediction, and the operation of other software or programs.
  • the network communication module is used to implement communication between various components in the memory 402 and communication with other hardware and software in the mobile terminal.
  • the processor 401 is configured to execute the health prediction program stored in the memory 402, and implement the following steps:
  • the force-receiving part is within the shooting range of the mobile terminal, obtaining a vibration video of the force-receiving part according to the motion video;
  • the first force-receiving condition of the force-receiving part is obtained from the digital twin intelligent health prediction model, which stores the different force-receiving conditions of different body parts of the user Simulated vibration information under;
  • the first risk score of the current action of the user is determined.
  • Another embodiment of the present application provides a computer-readable storage medium, the computer-readable storage medium stores a computer program, and the computer program is executed by a processor to implement the following steps:
  • the force-receiving part is within the shooting range of the mobile terminal, acquiring a vibration video of the force-receiving part according to the motion video;
  • the first force-receiving condition of the force-receiving part is obtained from the digital twin intelligent health prediction model, which stores the different force-receiving conditions of different body parts of the user Simulated vibration information under;
  • the first risk score of the current action of the user is determined.

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Signal Processing (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Public Health (AREA)
  • Pathology (AREA)
  • Social Psychology (AREA)
  • Databases & Information Systems (AREA)
  • Primary Health Care (AREA)
  • Data Mining & Analysis (AREA)
  • Biomedical Technology (AREA)
  • Computing Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Psychiatry (AREA)
  • Epidemiology (AREA)
  • Human Computer Interaction (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Theoretical Computer Science (AREA)
  • Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
  • Medical Treatment And Welfare Office Work (AREA)

Abstract

一种基于振动检测的数字孪生智能健康预测方法及装置，所述方法包括：当接收到健康预测指令时，采集用户的运动视频（101）；基于用户的运动视频识别出所述用户当前动作对应的受力部位（102）；判断所述受力部位是否位于所述移动终端的拍摄范围内（103）；若所述受力部位位于所述移动终端的拍摄范围内，则根据所述运动视频获取所述受力部位的振动视频（104）；根据所述受力部位的振动视频确定所述受力部位的第一振动信息（105）；根据所述第一振动信息从数字孪生智能健康预测模型中获取所述受力部位的第一受力状况，所述数字孪生智能健康预测模型存储了所述用户的不同身体部位在不同受力状况下的模拟振动信息（106）；根据所述第一受力状况，确定所述用户当前动作的第一危险度评分（107）。通过上述方法实现了实时监测用户在健身时当前动作的危险度。

Description

基于振动检测的数字孪生智能健康预测方法及装置
本申请要求于2019年8月31日提交中国专利局、申请号为2019108198371、申请名称为“基于振动检测的数字孪生智能健康预测方法及装置”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及互联网技术领域，尤其涉及一种基于振动检测的数字孪生智能健康预测方法及装置。
背景技术
随着互联网的不断发展，“互联网+”是创新2.0下的互联网发展的新业态，通俗的说，“互联网+”就是“互联网+各个传统行业”，但这并不是简单的两者相加，而是利用信息通信技术以及互联网平台，让互联网与传统行业进行深度融合，创造新的发展生态。它代表一种新的社会形态，即充分发挥互联网在社会资源配置中的优化和集成作用，将互联网的创新成果深度融合于经济、社会各域之中，提升全社会的创新力和生产力，形成更广泛的以互联网为基础设施和实现工具的经济发展新形态。
传统的健康监测机制一般是采用本地化检测设备，如用户在运动过程中佩带手环，手环可以采集运动过程中用户的身体数据，通过手环进行本地化的振动检测，以及健康预测等，但是这样需要购买具有振动检测功能的手环，而且由于手环只能佩戴在手上，无法对身体其他部位进行检测，检测区域小，难以满足日益增多的用户在各种运动场景中的智能化的健康预测需求。
发明内容
本申请实施例提供基于振动检测的数字孪生智能健康预测方法及装置，对用户的健身动作进行振动检测，通过数字孪生智能健康预测模型得到用户当前动作的危险度评分，从而实现实时监测用户在健身时当前动作的危险度。
具体的，本申请实施例所公开的振动检测方法中的数据传输流程可以基于互联网+技术，形成本地+云端或服务器的分布式智能化振动检测系统，一方面本地可以通过采集装置进行精确的原始影像采集和预处理，另一方面云端或服务器可以基于获取到的分布式数据，结合通过大数据技术统计分析得到的各类专用健康预测模型，预测被检测目标的健康状况，实现互联网与传统健康监测行业的深度融合，提高健康监测的智能性和准确度，满足日益增多的各类运动场景中的智能化的健康预测需求。
本申请第一方面提供一种基于振动检测的数字孪生智能健康预测方法，所述方法应用于移动终端，所述移动终端包括摄像头，所述方法包括：
当接收到健康预测指令时，采集用户的运动视频；
基于所述用户的运动视频识别出所述用户当前动作对应的受力部位；
判断所述受力部位是否位于所述移动终端的拍摄范围内；
若所述受力部位位于所述移动终端的拍摄范围内，则根据所述运动视频获取所述受力部位的振动视频；
根据所述受力部位的振动视频确定所述受力部位的第一振动信息；
根据所述第一振动信息从数字孪生智能健康预测模型中获取所述受力部位的第一受力状况，所述数字孪生智能健康预测模型存储了所述用户的不同身体部位在不同受力状况下的模拟振动信息；
根据所述第一受力状况，确定所述用户当前动作的第一危险度评分。
本申请第二方面提供了一种基于振动检测的数字孪生智能健康预测装置，所述装置包括：
采集单元，用于当接收到健康预测指令时，采集用户的运动视频；
识别单元，用于基于所述用户的运动视频识别出所述用户当前动作对应的受力部位；
判断单元，用于判断所述受力部位是否位于所述移动终端的拍摄范围内；
第一获取单元，用于若所述受力部位位于所述移动终端的拍摄范围内，则根据所述运动视频获取所述受力部位的振动视频；
第一确定单元，用于根据所述受力部位的振动视频确定所述受力部位的第一振动信息；
第二获取单元，根据所述第一振动信息从数字孪生智能健康预测模型中获取所述受力部位的第一受力状况，所述数字孪生智能健康预测模型存储了所述用户的不同身体部位在不同受力状况下的模拟振动信息；
第二确定单元，用于根据所述第一受力状况，确定所述用户当前动作的第一危险度评分。
本申请第三方面提供了一种移动终端，所述移动终端包括处理器、存储器、通信接口，以及一个或多个程序，所述一个或多个程序被存储在所述存储器中，并且被配置由所述处理器执行，所述程序包括用于执行上述任一方法中的步骤的指令。
本申请第四方面提供了一种计算机可读存储介质，所述计算机可读存储介质存储用于电子数据交换的计算机程序，其中，所述计算机程序使得计算机执行上述任一方法中所描述的部分或全部步骤。
可以看到，通过本申请提供的基于振动检测的数字孪生智能健康预测方法及装置，当接收到健康预测指令时，采集用户的运动视频，基于用户的运动视频识别出用户当前动作对应的受力部位，根据运动视频获取受力部位的振动视频，根据受力部位的振动视频确定受力部位的第一振动信息，根据第一振动信息从数字孪生智能健康预测模型中获取受力部位的第一受力状况，根据第一受力状况，确定用户当前动作的第一危险度评分。这样，用户在健身时，可以通过移动终端实时获取用户的运动视频，移动终端对用户的运动视频进行处理得 到用户当前动作的危险度评分，实现了实时监测用户在健身时当前动作的危险度，并且，通过移动终端对用户进行检测时，检测区域大，可以满足用户在各种运动场景中的健康预测需求。
附图说明
为了更清楚地说明本申请实施例中的技术方案，下面将对实施例中所需使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1a为本申请实施例提供的一种基于振动检测的数字孪生智能健康预测系统的示意图；
图1b为本申请实施例提供的一种基于振动检测的数字孪生智能健康预测方法的流程图；
图2为本申请实施例提供的一种基于振动检测的数字孪生智能健康预测装置的示意图；
图3为本申请实施例提供的一种部位力学模型的示意图；
图4为本申请实施例提供的一种移动终端的结构示意图；
图5为本申请实施例提供的一种用户肘部运动的示意图。
具体实施方式
本申请实施例提供基于振动检测的数字孪生智能健康预测方法及装置，对用户的健身动作进行振动检测，通过数字孪生智能健康预测模型得到用户当前动作的危险度评分，从而实现实时监测用户在健身时当前动作的危险度。
为了使本技术领域的人员更好地理解本申请方案，下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本申请一部分的实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都应当属于本申请保护的范围。
本申请的说明书和权利要求书及上述附图中的术语“第一”、“第二”、“第三”、“第四”等是用于区别不同对象，而不是用于描述特定顺序。此外，术语“包括”和“具有”以及它们任何变形，意图在于覆盖不排他的包含。例如包含了一系列步骤或单元的过程、方法、系统、产品或设备没有限定于已列出的步骤或单元，而是可选地还包括没有列出的步骤或单元，或可选地还包括对于这些过程、方法、产品或设备固有的其它步骤或单元。
下面对本申请实施例进行详细说明。
本申请实施例提供的基于振动检测的数字孪生智能健康预测方法及装置，用户在健身时，通过移动终端采集用户健身过程中的受力部位的振动视频，对 受力部位的振动视频进行处理，得到振动信息，通过数字孪生智能健康预测模型得到用户当前动作的危险度评分。这样，用户在健身时，可以通过移动终端实时获取用户的运动视频，移动终端对用户的运动视频进行处理得到用户当前动作的危险度评分，实现了实时监测用户在健身时当前动作的危险度，并且，通过移动终端对用户进行检测时，检测区域大，可以满足用户在各种运动场景中的健康预测需求。
首先参见图1a，图1a为本申请一个实施例提供的一种基于振动检测的数字孪生智能健康预测系统的示意图，该健康预测系统包括移动终端和用户，移动终端可以是手机、平板电脑、笔记本电脑、移动互联网设备、或其他类型带有摄像头的终端设备，用户可以进行各种健身动作，移动终端包括摄像头，移动终端可以通过摄像头拍摄用户的实时动作。其中，如图1a所示的健康预测系统100中，移动终端101为手机，手机包括后置摄像头，后置摄像头有一定的拍摄范围，用户102的健身动作为俯卧撑，其中：
移动终端101用于，当接收到健康预测指令时，采集用户102的运动视频；基于所述用户102的运动视频识别出所述用户102当前动作对应的受力部位；判断所述受力部位是否位于所述移动终端101的拍摄范围内；若所述受力部位位于所述移动终端101的拍摄范围内，则根据所述运动视频获取所述受力部位的振动视频；根据所述受力部位的振动视频确定所述受力部位的第一振动信息；根据所述第一振动信息从数字孪生智能健康预测模型中获取所述受力部位的第一受力状况，所述数字孪生智能健康预测模型存储了所述用户102的不同身体部位在不同受力状况下的模拟振动信息；根据所述第一受力状况，确定所述用户102当前动作的第一危险度评分。
具体的，如图1a所示，移动终端101接收到健康预测指令时，通过摄像头采集用户102的运动视频，即采集用户102进行俯卧撑的运动视频，移动终端101基于用户102进行俯卧撑的视频识别出用户102当前动作为俯卧撑，进而判断出用户102当前动作对应的受力部位为肘部关节，移动终端101判断肘部关节位于拍摄范围内，进而获取肘部关节的振动视频，由于用户102进行俯卧撑时涉及到双臂的肘部关节，因此移动终端101分别获取双臂的肘部关节的振动视频，然后判断用户102当前俯卧撑动作的危险度评分，如果用户102做俯卧撑的频率过高，那么肘部关节的振动频率也会过高，因此得到的危险度评分较高，移动终端101可以及时向用户102发出告警或者提示，提示用户102降低频率，或者暂时停止运动。这样，用户在健身时，可以通过移动终端实时获取用户的运动视频，移动终端对用户的运动视频进行处理得到用户当前动作的危险度评分，实现了实时监测用户在健身时当前动作的危险度，并且，通过移动终端对用户进行检测时，检测区域大，可以满足用户在各种运动场景中的健康预测需求。
参见图1b，图1b为本申请一个实施例提供的一种基于振动检测的数字孪 生智能健康预测方法的流程图，本申请实施例提供的基于振动检测的数字孪生智能健康预测方法应用于移动终端，移动终端包括摄像头。其中，如图1b所示，本申请的一个实施例提供的一种基于振动检测的数字孪生智能健康预测方法可以包括：
101、当接收到健康预测指令时，采集用户的运动视频。
具体的，用户在健身时，可以开启移动终端的健康预测功能，移动终端接收到健康预测指令时，开启摄像头，采集用户的运动视频。
其中，移动终端可以是手机、平板电脑、笔记本电脑、移动互联网设备、或其他类型带有摄像头的终端设备。
102、基于所述用户的运动视频识别出所述用户当前动作对应的受力部位。
用户在运动过程中，通过移动终端对用户进行实时影像采集，基于采集得到的用户的运动视频，识别出用户的当前动作，从而确定用户当前动作对应的受力部位。
具体地，当移动终端为手机时，手机界面显示健身拉伤预防功能入口，用户点击进入健身拉伤预防功能后，手机开启摄像头对用户进行实时健身影像采集，用户在健身过程中，手机会基于用户的当前动作，识别出该动作对应的受力部位。例如，用户的当前动作为俯卧撑时，受力部位为手臂，用户的当前动作为仰卧起坐时，受力部位为腹部。
103、判断所述受力部位是否位于所述移动终端的拍摄范围内。
具体的，由于移动终端摆放的位置不同，或者移动终端拍摄用户的角度不同，存在移动终端没有拍摄到用户的受力部位的情况，所以需要判断受力部位是否位于移动终端的拍摄范围内。
在一种可能的示例中，判断受力部位是否位于移动终端的拍摄范围内的方法可以是：
将受力部位的特征点与运动视频的图像帧中用户各个身体部位的特征点进行对比，如果对比成功，则说明受力部位位于移动终端的拍摄范围内，如果对比失败，则说明受力部位不位于移动终端的拍摄范围内。
104、若所述受力部位位于所述移动终端的拍摄范围内，则根据所述运动视频获取所述受力部位的振动视频。
具体的，若受力部位位于移动终端的拍摄范围内，则在运动视频中确定该受力部位的位置，然后从运动视频中选取该受力部位的振动视频。
105、根据所述受力部位的振动视频确定所述受力部位的第一振动信息。
具体的，采用拉格朗日运动放大方法对所述受力部位的振动视频进行处理，以得到振动放大视频，对该振动放大视频进行处理，得到受力部位的第一振动信息。
106、根据所述第一振动信息从数字孪生智能健康预测模型中获取所述受力部位的第一受力状况，所述数字孪生智能健康预测模型存储了所述用户的不 同身体部位在不同受力状况下的模拟振动信息。
其中，数字孪生技术作为实现制造的物理世界和信息世界之间的交互融合的核心技术，是除了人工智能、机器学习、AR/VR、区块链之外的另外一个技术风向。数字孪生技术通过数字化的形式对某一物理实体过去和目前的行为或流程进行动态呈现。作为一种充分利用数据、智能并集成多学科的技术，数字孪生技术在实践智能制造理念和目标中提供更加实时、高效、智能的服务。
具体的，数字孪生智能健康预测模型与移动终端同步运行,通过数字孪生智能健康预测模型模拟人体的运动情况，能及时的预测人体健身过程中运动部位是否容易拉伤等，通过本申请提供的基于振动检测的数字孪生智能健康预测方法，将空间图形处理与数字孪生技术结合，实现了计算机视觉技术与人工智能的结合，具体通过振动检测来判断人体运动情况。其中，数字孪生智能健康预测模型需要预先构建，由于对于不同用户而言，用户的身高、体重、性别、年龄、身体状况、所处环境等可能不同，所以在预先构建数字孪生智能健康预测模型时，首先需要通过移动终端获取用户的人体参数，输入用户的人体参数生成数字孪生智能健康预测模型，构建完成后，数字孪生智能健康预测模型模拟用户的不同身体部位在不同受力状况下的振动信息，并且将模拟出来的振动信息存储在数字孪生智能健康预测模型中。
107、根据所述第一受力状况，确定所述用户当前动作的第一危险度评分。
具体的，用户的不同受力部位有不同的受力范围和受力频率，如果受力过大或者受力频率过高，很有可能造成用户的受力部位拉伤或者用户的身体损伤，所以根据第一受力状况，可以确定用户当前动作的危险度评分。
例如，用户在做俯卧撑时，如果手臂运动的频率过高，容易造成手臂肌肉拉伤；再例如，用户在做动态臀桥时，如果臀部提起高度过高，容易造成腰部肌肉损伤。
可见，通过本申请实施例提供的基于振动检测的数字孪生智能健康预测方法，当接收到健康预测指令时，采集用户的运动视频，基于用户的运动视频识别出用户当前动作对应的受力部位，根据运动视频获取受力部位的振动视频，根据受力部位的振动视频确定受力部位的第一振动信息，根据第一振动信息从数字孪生智能健康预测模型中获取受力部位的第一受力状况，根据第一受力状况，确定用户当前动作的第一危险度评分。这样，用户在健身时，可以通过移动终端实时获取用户的运动视频，移动终端对用户的运动视频进行处理得到用户当前动作的危险度评分，实现了实时监测用户在健身时当前动作的危险度，并且，通过移动终端对用户进行检测时，检测区域大，可以满足用户在各种运动场景中的健康预测需求。
本申请另一个实施例提供了另一种基于振动检测的数字孪生智能健康预测方法，可以包括：
201、当接收到健康预测指令时，采集用户的运动视频。
具体的，用户在健身时，可以开启移动终端的健康预测功能，移动终端接收到健康预测指令时，开启摄像头，采集用户的运动视频。
其中，移动终端可以是手机、平板电脑、笔记本电脑、移动互联网设备、或其他类型带有摄像头的终端设备。
202、基于用户的运动视频识别出用户当前动作对应的受力部位。
用户在运动过程中，通过移动终端对用户进行实时影像采集，基于采集得到的用户的运动视频，识别出用户的当前动作，从而确定用户当前动作对应的受力部位。
具体地，当移动终端为手机时，手机界面显示健身拉伤预防功能入口，用户点击进入健身拉伤预防功能后，手机开启摄像头对用户进行实时健身影像采集，用户在健身过程中，手机会基于用户的当前动作，识别出该动作对应的受力部位。
203、判断受力部位是否位于移动终端的拍摄范围内。
具体的，由于移动终端摆放的位置不同，或者移动终端拍摄用户的角度不同，存在移动终端没有拍摄到用户的受力部位的情况，所以需要判断受力部位是否位于移动终端的拍摄范围内。
在一种可能的示例中，判断受力部位是否位于移动终端的拍摄范围内的方法可以是：将受力部位的特征点与运动视频的图像帧中用户各个身体部位的特征点进行对比，如果对比成功，则说明受力部位位于移动终端的拍摄范围内，如果对比失败，则说明受力部位不位于移动终端的拍摄范围内。
204、若受力部位位于移动终端的拍摄范围内，则根据运动视频获取受力部位的振动视频。
具体的，若受力部位位于移动终端的拍摄范围内，则在运动视频中确定该受力部位的位置，然后从运动视频中选取该受力部位的振动视频。
受力部位的振动视频中包含受力部位的运动过程，这个运动过程非常微小，需要进行放大以便后续振动信息的提取。采用拉格朗日运动放大方法，可以通过对视频中目标特征点进行跟踪运动轨迹和聚类实现对微小运动的放大。
205、对受力部位的振动视频的图像帧进行校准，以得到稳定的多个运动特征点。
在一种可能的示例中，对受力部位的振动视频的图像帧进行校准，以得到温度的多个运动特征点的方法可以是：选择受力部位的振动视频中的至少一帧图像；根据该至少一帧图像确定该受力部位的振动视频中在视频采集过程中处于相对静止状态的参考特征点；将该至少一帧图像包括的每一帧图像按照N个不同圆心进行圆形图像截取，以得到N个基础圆形分区，N为大于3的整数；从该N个基础圆形分区中选出目标圆形分区，该目标圆形分区包含相对运动特征点，该相对运动特征点为针对该参考特征点进行相对运动的运动特征点；将该目标圆形分区按照预设窗口进行截取，以得到多个截取分区，该预设 窗口的大小和形状根据该受力部位的肌肉形态确定；依次从该多个截取分区中获取运动距离处于预设范围的相对运动特征点，并累计已获取相对运动特征点的数值，该预设范围为该受力部位在正常受力状态下的振动幅度范围；当累计的数值不小于预设数值时，确定已获取的相对运动特征点为该稳定的多个运动特征点。
具体地，采用拉格朗日放大方法对振动视频进行放大，首先需要获得振动视频中稳定的多个运动特征点，即微小运动的点，以便与振动视频中的静止点(背景点)和剧烈运动的点进行区分，对于振动视频来说，除了拍摄到振动图像，还包括一些相对静止的背景图像，例如，用户在做引体向上时，支撑的横杆就是相对静止的背景图像，再例如，用户在做俯卧撑时，地面就是相对静止的背景图像。获取相对静止状态的物体上的点作为参考特征点，然后根据预设的运动特征点提取策略，提取振动视频中稳定的多个运动特征点。
可以看出，对于振动视频中的至少一帧图像来说，并不是每个区域都包含运动的点，如果对图像帧中的每个区域进行一一排查，获得稳定的运动特征点，需要耗费大量的时间。那么，可以采用适当的分区策略提取运动特征点以提升效率。
206、对稳定的多个运动特征点进行处理，以得到第一振动放大视频。
在一种可能的示例中，对稳定的多个运动特征点进行处理，以得到第一振动放大视频的方法可以是：对该多个运动特征点进行跟踪，以得到该多个运动特征点的轨迹向量；采用聚类算法对该多个运动特征点的轨迹向量进行聚类，以得到K类运动层；从该K类运动层中获得需要进行放大的目标运动层；对该目标运动层中运动特征点的偏移距离乘以一个放大倍数放大，以得到放大的运动层；对该放大的运动层进行渲染，以得到该第一振动放大视频。
具体的，对多个运动特征点进行跟踪，获得其对应的轨迹向量，轨迹向量用数值来描述运动特征点的运动方向、运动距离和亮度变化等；再采用聚类算法对多个运动特征点的轨迹向量进行聚类，获得K类运动层，根据轨迹向量的相关性和相似度进行K类运动层的划分，可以使得不同运动层中包含不同类别的运动，以便选择K类运动层中微小运动对应的运动层进行放大处理，获得放大的运动层。最后，因为对运动层的放大造成目标视频对应的图像帧中包括一些空白区域，需要进行渲染对图像帧进行填充。
可以看出，用户运动过程中，有些振动是微小的，如果直接提取振动信息会导致准确率下降，通过将振动进行放大，可以提高振动信息提取的准确率，以便于后续处理。
207、获取第一振动放大视频对应的图像帧，采用相位相关算法对图像帧进行计算，以得到图像帧间的第一交叉互功率谱。
其中，采用相位相关算法对所述图像帧进行计算的公式为：
Figure PCTCN2020104822-appb-000001
其中，F a为a帧图像的傅立叶变换，为b帧图像的傅里叶变换的共轭信号，除式的下边为两个傅里叶变换的信号的相关积的模。R为本步骤的计算结果交叉互功率谱。
获得交叉互功率谱后，其中包含频域噪音，因此可对其进行滤波处理，提升信噪比，以便提升后续提取的振动信息的准确度。
可选的，在采用相位相关算法对图像帧进行计算，以得到图像帧间的交叉互功率谱之后，还包括：
获取交叉互功率谱中的一个或多个相关峰，相关峰为频域信号；根据各相关峰在振动视频中对应的位置及相关峰的频段确定各相关峰对应的滤波策略；按照各相关峰对应的滤波策略对各相关峰进行滤波处理。
208、对第一交叉互功率谱进行反傅立叶变换，以得到第一振动信息。
具体的，交叉互功率谱反应的是频域上的振动信息，而需要在时域对振动信息进行分析，则需要进行反傅里叶变换(或逆傅里叶变换)。进行反傅里叶变换采用的公式为：
Figure PCTCN2020104822-appb-000002
其中，
Figure PCTCN2020104822-appb-000003
表示对交叉互功率谱进行反傅里叶变换，R′是滤波处理后得到的交叉互功率谱，r是振动视频中像素的振动信息。
209、根据第一振动信息从数字孪生智能健康预测模型中获取受力部位的第一受力状况。
其中，数字孪生智能健康预测模型存储了用户的不同身体部位在不同受力状况下的模拟振动信息。
在一种可能的示例中，根据第一振动信息从数字孪生智能健康预测模型中获取受力部位的第一受力状况之前，还包括：获取用户的人体参数，人体参数包括年龄、性别、身高、体重、体脂率、心率、血压的任意一种组合；根据人体参数构建与用户的M个身体部位对应的M个部位力学模型，其中，M为正整数，M个身体部位与M个部位力学模型一一对应；分别对M个部位力学模型中的M个指定位置施加径向力，其中，M个部位力学模型与M个指定位置一一对应，径向力为模拟M个身体部位的受力；确定M个部位力学模型的M个移动距离，其中，M个部位力学模型与M个移动距离一一对应；将M个移动距离输入预设动力学算法进行计算，以得到M个部位力学模型的M条模拟 振动信息。
具体的，部位力学模型是虚拟的身体部位，也就是说，通过采用结构件力学模型来实现对用户的身体部位的虚拟化。进一步的，通过三维扫描用户的身体部位构建部位力学模型。
其中，在分别对M个部位力学模型中的M个指定位置施加径向力的步骤中，径向力即为径向运动的摩擦力，径向力y沿径向均匀施加在部位力学模型的指定位置，从而抵消了来自这些径向力的振动，所以假设摩擦力只沿径向运动。
当对部位力学模型施加径向力后,需要对该径向力进行计算。
举例来说，参见图3，图3为本申请一个实施例提供的一种部位力学模型示意图。设单位长度为h，入口半径为R1，出口半径是R2。相应地，其进口和出口地区
Figure PCTCN2020104822-appb-000004
设t是一个作用于这个轴向元素的单位向量，n1和n2分别为入口和出口的法向单位向量。假设n1和n2之间的夹角足够小，将该部力学模型看作圆锥或圆柱体。设v1和v2分别为进出口速度的平均值。计算该部位力学模型所受的径向力的公式如下：
Figure PCTCN2020104822-appb-000005
其中，
Figure PCTCN2020104822-appb-000006
Figure PCTCN2020104822-appb-000007
Figure PCTCN2020104822-appb-000008
在部位的每个细胞单元中计算出的这个力，由于细胞运动，形成了部位上的所有力。通过扫描来的计算机辅助设计(CAD)模型，确定指定地方在部位相对于轴心点位置的相对位置，将摩擦力与振动模型联系起来。
其中，在将M个移动距离输入预设动力学算法进行计算的步骤中，预设动力学算法为：
Figure PCTCN2020104822-appb-000009
其中，Is为惯性矩阵，Ks为刚度矩阵，Cs为联结矩阵，Mb为力矩矩阵。
其中，径向力由公式
Figure PCTCN2020104822-appb-000010
计算得到，可被分解成平行于三个轴的三部分力，确定指定位置处在三个轴上的力矩M1、M2和M3，生成矩阵M b＝[M 1 M 2 M 3] T，代入上述预设动力学算法进行计算，得到角位移，利用被检测区域和指定位置的固定距离，将计算得到的角位移进一步转化为线性位移，再使用三角关系将线性位移投影到二维(Y-Z)平面上。
其中，数字孪生智能健康预测模型中存储的模拟振动信息包括振动波形图、模态图和热力图。
可以看出，人体包含的身体部位很多，分析所有身体部位的健康状况过于繁杂，所以可以采取单个力学模型，对各身体部位在不同健康状况下的运动情况进行分析，不考虑这些身体部位运动造成的相互影响，而是通过测量各身体部位在不同情况下的运动情况来模拟它们共同作用产生的能量。由于人体在运动时涉及到的身体部位基本上是固定的，所以只分析这些主要参与运动的身体部位，从而降低建模的复杂性。
210、根据第一受力状况，确定用户当前动作的第一危险度评分。
具体的，用户的不同受力部位有不同的受力范围和受力频率，如果受力过大或者受力频率过高，很有可能造成用户的受力部位拉伤或者用户的身体损伤，所以根据第一受力状况，可以确定用户当前动作的危险度评分。
在一种可能的示例中，根据第一受力状况，确定用户当前动作的第一危险度评分之后，还包括：当第一危险度评分为肌肉拉伤可能性评分时，判断肌肉拉伤可能性评分是否超出预设可能性评分；若肌肉拉伤可能性评分超出预设可能性评分，则发出第一预警消息，第一预警消息用于提示用户调整运动姿势或者停止运动以避免受力部位拉伤；当第一危险度评分为身体机能评分时，判断身体机能评分是否低于预设评分；若身体机能评分低于预设评分，则发出第二预警消息，第二预警消息用于提示用户减缓运动频率或者停止运动以避免用户的身体损伤。
可以看出，用户在健身时，可以通过移动终端实时获取用户的运动视频，移动终端对用户的运动视频进行处理得到用户当前动作的危险度评分，实现了实时监测用户在健身时当前动作的危险度，并对用户进行预警提示。
211、若受力部位不位于移动终端的拍摄范围内，则确定与受力部位关联且位于移动终端的拍摄范围内的关联部位。
在一种可能的示例中，确定与受力部位关联且位于移动终端的拍摄范围内的关联部位的方法可以是：确定与受力部位关联的至少一个关联部位；判断至少一个关联部位中是否存在位于移动终端的拍摄范围内的第一关联部位；若存在，则确定第一关联部位为与受力部位关联且位于移动终端的拍摄范围内的关 联部位；若不存在，则开启广角摄像模式以使得至少一个关联部位中存在位于移动终端的拍摄范围内的第二关联部位；确定第二关联部位为与受力部位关联且位于移动终端的拍摄范围内的关联部位。
可以看出，当受力部位没有位于移动终端的拍摄范围内时，可以确定位于拍摄范围内的与受力部位的关联部位，或者开启广角摄像模式确定位于拍摄范围内的与受力部位的关联部位，以便于后续根据关联部位的振动信息确定受力部位的振动信息。
在另一种可能的示例中，若受力部位在移动终端的拍摄范围内，还需要识别用户的衣着特征，从而判断能否准确获取受力部位的振动视频，若识别到受力部位处用户的衣着特征覆盖了受力部位的振动特征，那么无法准确获取受力部位的振动视频，则需要在移动终端的拍摄范围内重新确定与受力部位相关且可以准确获取振动视频的关联部位，基于该关联部位的位置，获取该关联部位的振动视频。
212、根据运动视频获取关联部位的振动视频。
213、根据关联部位的振动视频确定关联部位的振动信息。
214、根据关联部位的振动信息得到受力部位的第二振动信息。
215、根据第二振动信息从数字孪生智能健康预测模型中获取受力部位的第二受力状况。
216、根据第二受力状况，确定用户当前动作的第二危险度评分。
可见，在本申请实施例中，通过移动终端获取用户健身过程中的运动视频，确认受力部位，根据受力部位是否在移动终端的拍摄范围内进行不同处理，得到受力部位的振动信息，将受力部位的振动信息与数字孪生智能健康预测模型存储的该受力部位在不受力状况下的模拟振动信息进行对比，以确定该受力部位的受力状况，从而确定用户当前动作的危险度评分，并且，根据危险度评分对用户进行预警。这样，当用户的健身动作幅度过大、频率过高或者姿势错误时，可以对用户进行报警提示，从而实现实时检测用户在健身时当前动作的危险度，规范化用户的健身动作，避免给身体健康带来负面影响。
参见图5，图5为本申请实施例提供的一种用户肘部运动的示意图。如图5所示，用户在做肘部运动，例如用户在锻炼肱二头肌时，肘部会处于弯曲状态和伸直状态的重复过程，用户肘部伸直时，肘部关节部位从如图5所示的肘部1，运动过程中，肘部关节部位如图5所示的肘部2和肘部3，用户肘部弯曲时，肘部关节部位如图5所示的肘部4，用户在做肘部运动时，肘部关节部位进行有规律的振动，从而可以通过用户手机等终端获取肘部关节部位的振动视频，从而对振动视频进行分析，判断用户的锻炼动作是否标准，以及是否会造成肘部关节损伤。
参见图2，图2为本申请的另一个实施例提供的一种基于振动检测的数字孪生智能健康预测装置的示意图。其中，如图2所示，本申请的另一个实施例 提供的一种基于振动检测的数字孪生智能健康预测装置可以包括：
采集单元201，用于当接收到健康预测指令时，采集用户的运动视频；
识别单元202，用于基于所述用户的运动视频识别出所述用户当前动作对应的受力部位；
判断单元203，用于判断所述受力部位是否位于所述移动终端的拍摄范围内；
第一获取单元204，用于若所述受力部位位于所述移动终端的拍摄范围内，则根据所述运动视频获取所述受力部位的振动视频；
第一确定单元205，用于根据所述受力部位的振动视频确定所述受力部位的第一振动信息；
第二获取单元206，根据所述第一振动信息从数字孪生智能健康预测模型中获取所述受力部位的第一受力状况，所述数字孪生智能健康预测模型存储了所述用户的不同身体部位在不同受力状况下的模拟振动信息；
第二确定单元207，用于根据所述第一受力状况，确定所述用户当前动作的第一危险度评分。
本申请实施例基于振动检测的数字孪生智能健康预测装置的具体实施可参见上述基于振动检测的数字孪生智能健康预测方法的各实施例，在此不做赘述。
参见图4，图4为本申请的实施例提供的一种移动终端的结构示意图。其中，如图4所示，本申请的实施例提供的一种移动终端可以包括：
处理器401，例如CPU。
存储器402，可选的，存储器可以为高速RAM存储器，也可以是稳定的存储器，例如磁盘存储器。
通信接口403，用于实现处理器401和存储器402之间的连接通信。
本领域技术人员可以理解，图4中示出的移动终端的结构并不构成对移动终端的限定，可以包括比图示更多或更少的部件，或者组合某些部件，或者不同的部件布置。
如图4所示，存储器402中可以包括操作系统、网络通信模块以及健康预测的程序。操作系统是管理和控制移动终端硬件和软件资源的程序，支持健康预测的程序以及其他软件或程序的运行。网络通信模块用于实现存储器402内部各组件之间的通信，以及与移动终端中其他硬件和软件之间通信。
在图4所示的移动终端中，处理器401用于执行存储器402中存储的健康预测的程序，实现以下步骤：
当接收到健康预测指令时，采集用户的运动视频；
基于所述用户的运动视频识别出所述用户当前动作对应的受力部位；
判断所述受力部位是否位于所述移动终端的拍摄范围内；
若所述受力部位位于所述移动终端的拍摄范围内，则根据所述运动视频获 取所述受力部位的振动视频；
根据所述受力部位的振动视频确定所述受力部位的第一振动信息；
根据所述第一振动信息从数字孪生智能健康预测模型中获取所述受力部位的第一受力状况，所述数字孪生智能健康预测模型存储了所述用户的不同身体部位在不同受力状况下的模拟振动信息；
根据所述第一受力状况，确定所述用户当前动作的第一危险度评分。
本申请实施例提供的移动终端的具体实施可参见上述基于振动检测的数字孪生智能健康预测方法的各实施例，在此不做赘述。
本申请的另一个实施例提供了一种计算机可读存储介质，计算机可读存储介质存储有计算机程序，计算机程序被处理器执行以实现以下步骤：
当接收到健康预测指令时，采集用户的运动视频；
基于所述用户的运动视频识别出所述用户当前动作对应的受力部位；
判断所述受力部位是否位于所述移动终端的拍摄范围内；
若所述受力部位位于所述移动终端的拍摄范围内，则根据所述运动视频获取所述受力部位的振动视频；
根据所述受力部位的振动视频确定所述受力部位的第一振动信息；
根据所述第一振动信息从数字孪生智能健康预测模型中获取所述受力部位的第一受力状况，所述数字孪生智能健康预测模型存储了所述用户的不同身体部位在不同受力状况下的模拟振动信息；
根据所述第一受力状况，确定所述用户当前动作的第一危险度评分。
本申请计算机可读存储介质的具体实施可参见上述基于振动检测的数字孪生智能健康预测方法的各实施例，在此不做赘述。
需要说明的是，对于前述的各方法实施例，为了简单描述，故将其都表述为一系列的动作组合，但是本领域技术人员应该知悉，本申请并不受所描述的动作顺序的限制，因为依据本申请，某些步骤可以采用其他顺序或者同时进行。其次，本领域技术人员也应该知悉，说明书中所描述的实施例均属于优选实施例，所涉及的动作和模块并不一定是本申请所必须的。在上述实施例中，对各个实施例的描述都各有侧重，某个实施例中没有详述的部分，可以参见其他实施例的相关描述。
以上所述，以上实施例仅用以说明本申请的技术方案，而非对其限制；尽管参照前述实施例对本申请进行了详细的说明，本领域的普通技术人员应当理解：其依然可以对前述各实施例记载的技术方案进行修改，或者对其中部分技术特征进行等同替换；而这些修改或者替换，并不使相应技术方案的本质脱离本申请各实施例技术方案的范围。

Claims (10)

  1. 一种基于振动检测的数字孪生智能健康预测方法，其特征在于，所述方法应用于移动终端，所述移动终端包括摄像头，所述方法包括：
    当接收到健康预测指令时，采集用户的运动视频；
    基于所述用户的运动视频识别出所述用户当前动作对应的受力部位；
    判断所述受力部位是否位于所述移动终端的拍摄范围内；
    若所述受力部位位于所述移动终端的拍摄范围内，则根据所述运动视频获取所述受力部位的振动视频；
    根据所述受力部位的振动视频确定所述受力部位的第一振动信息；
    根据所述第一振动信息从数字孪生智能健康预测模型中获取所述受力部位的第一受力状况，所述数字孪生智能健康预测模型存储了所述用户的不同身体部位在不同受力状况下的模拟振动信息；
    根据所述第一受力状况，确定所述用户当前动作的第一危险度评分。
  2. 根据权利要求1所述的方法，其特征在于，所述根据所述受力部位的振动视频确定所述受力部位的第一振动信息包括：
    对所述受力部位的振动视频的图像帧进行校准，以得到稳定的多个运动特征点；
    对所述多个运动特征点进行跟踪，以得到所述多个运动特征点的轨迹向量；
    采用聚类算法对所述多个运动特征点的轨迹向量进行聚类，以得到K类运动层；
    从所述K类运动层中获得需要进行放大的目标运动层；
    对所述目标运动层中运动特征点的偏移距离乘以一个放大倍数放大，以得到放大的运动层；
    对所述放大的运动层进行渲染，以得到所述第一振动放大视频；
    获取所述第一振动放大视频对应的图像帧，采用相位相关算法对所述图像帧进行计算，以得到图像帧间的第一交叉互功率谱；
    对所述第一交叉互功率谱进行反傅立叶变换，以得到所述第一振动信息。
  3. 根据权利要求2所述的方法，其特征在于，所述对所述受力部位的振动视频的图像帧进行校准，以得到稳定的多个运动特征点，包括：
    选择所述受力部位的振动视频中的至少一帧图像；
    根据所述至少一帧图像确定所述受力部位的振动视频中在视频采集过程中处于相对静止状态的参考特征点；
    将所述至少一帧图像包括的每一帧图像按照N个不同圆心进行圆形图像截取，以得到N个基础圆形分区，N为大于3的整数；
    从所述N个基础圆形分区中选出目标圆形分区，所述目标圆形分区包含相对运动特征点，所述相对运动特征点为针对所述参考特征点进行相对运动的 运动特征点；
    将所述目标圆形分区按照预设窗口进行截取，以得到多个截取分区，所述预设窗口的大小和形状根据所述受力部位的肌肉形态确定；
    依次从所述多个截取分区中获取运动距离处于预设范围的相对运动特征点，并累计已获取相对运动特征点的数值，所述预设范围为所述受力部位在正常受力状态下的振动幅度范围；
    当累计的数值不小于预设数值时，确定已获取的相对运动特征点为所述稳定的多个运动特征点。
  4. 根据权利要求1至3任一项所述的方法，其特征在于，所述方法还包括：
    若所述受力部位不位于所述移动终端的拍摄范围内，则确定与所述受力部位关联且位于所述移动终端的拍摄范围内的关联部位；
    根据所述运动视频获取所述关联部位的振动视频；
    根据所述关联部位的振动视频确定所述关联部位的振动信息；
    根据所述关联部位的振动信息得到所述受力部位的第二振动信息；
    根据所述第二振动信息从数字孪生智能健康预测模型中获取所述受力部位的第二受力状况；
    根据所述第二受力状况，确定所述用户当前动作的第二危险度评分。
  5. 根据权利要求4所述的方法，其特征在于，所述确定与所述受力部位关联且位于所述移动终端的拍摄范围内的关联部位包括：
    确定与所述受力部位关联的至少一个关联部位；
    判断所述至少一个关联部位中是否存在位于所述移动终端的拍摄范围内的第一关联部位；
    若存在，则确定所述第一关联部位为与所述受力部位关联且位于所述移动终端的拍摄范围内的关联部位；
    若不存在，则开启广角摄像模式以使得所述至少一个关联部位中存在位于所述移动终端的拍摄范围内的第二关联部位；
    确定所述第二关联部位为与所述受力部位关联且位于所述移动终端的拍摄范围内的关联部位。
  6. 根据权利要求1所述的方法，其特征在于，所述根据所述第一振动信息从数字孪生智能健康预测模型中获取所述受力部位的第一受力状况之前，所述方法还包括：
    获取所述用户的人体参数，所述人体参数包括年龄、性别、身高、体重、体脂率、心率、血压的任意一种组合；
    根据所述人体参数构建与所述用户的M个身体部位对应的M个部位力学模型，其中，M为正整数，所述M个身体部位与所述M个部位力学模型一一对应；
    分别对所述M个部位力学模型中的M个指定位置施加径向力，其中，所述M个部位力学模型与所述M个指定位置一一对应，所述径向力为模拟所述M个身体部位的受力；
    确定所述M个部位力学模型的M个移动距离，其中，所述M个部位力学模型与所述M个移动距离一一对应；
    将所述M个移动距离输入预设动力学算法进行计算，以得到所述M个部位力学模型的M条模拟振动信息。
  7. 根据权利要求1所述的方法，其特征在于，所述根据所述第一受力状况，确定所述用户当前动作的第一危险度评分之后，所述方法还包括：
    当所述第一危险度评分为肌肉拉伤可能性评分时，判断所述肌肉拉伤可能性评分是否超出预设可能性评分；
    若所述肌肉拉伤可能性评分超出所述预设可能性评分，则发出第一预警消息，所述第一预警消息用于提示所述用户调整运动姿势或者停止运动以避免所述受力部位拉伤；
    当所述第一危险度评分为身体机能评分时，判断所述身体机能评分是否低于预设评分；
    若所述身体机能评分低于所述预设评分，则发出第二预警消息，所述第二预警消息用于提示所述用户减缓运动频率或者停止运动以避免所述用户的身体损伤。
  8. 一种基于振动检测的数字孪生智能健康预测装置，其特征在于，所述装置包括：
    采集单元，用于当接收到健康预测指令时，采集用户的运动视频；
    识别单元，用于基于所述用户的运动视频识别出所述用户当前动作对应的受力部位；
    判断单元，用于判断所述受力部位是否位于所述移动终端的拍摄范围内；
    第一获取单元，用于若所述受力部位位于所述移动终端的拍摄范围内，则根据所述运动视频获取所述受力部位的振动视频；
    第一确定单元，用于根据所述受力部位的振动视频确定所述受力部位的第一振动信息；
    第二获取单元，根据所述第一振动信息从数字孪生智能健康预测模型中获取所述受力部位的第一受力状况，所述数字孪生智能健康预测模型存储了所述用户的不同身体部位在不同受力状况下的模拟振动信息；
    第二确定单元，用于根据所述第一受力状况，确定所述用户当前动作的第一危险度评分。
  9. 一种移动终端，其特征在于，所述移动终端包括处理器、存储器、通信接口，以及一个或多个程序，所述一个或多个程序被存储在所述存储器中，并且被配置由所述处理器执行，所述程序包括用于执行如权利要求1至7任一 项所述的方法中的步骤的指令。
  10. 一种计算机可读存储介质，其特征在于，所述计算机可读存储介质存储用于电子数据交换的计算机程序，其中，所述计算机程序使得计算机执行如权利要求1至7任一项所述的方法。
PCT/CN2020/104822 2019-08-31 2020-07-27 基于振动检测的数字孪生智能健康预测方法及装置 WO2021036635A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910819837.1A CN110600132B (zh) 2019-08-31 2019-08-31 基于振动检测的数字孪生智能健康预测方法及装置
CN201910819837.1 2019-08-31

Publications (1)

Publication Number Publication Date
WO2021036635A1 true WO2021036635A1 (zh) 2021-03-04

Family

ID=68856704

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/104822 WO2021036635A1 (zh) 2019-08-31 2020-07-27 基于振动检测的数字孪生智能健康预测方法及装置

Country Status (2)

Country Link
CN (1) CN110600132B (zh)
WO (1) WO2021036635A1 (zh)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114822849A (zh) * 2022-05-06 2022-07-29 深圳市第二人民医院(深圳市转化医学研究院) 基于数字孪生的数据监测方法、装置、设备和存储介质
CN114968162A (zh) * 2022-06-20 2022-08-30 阿维塔科技(重庆)有限公司 车载食品的信息展示方法及装置
CN115543094A (zh) * 2022-11-28 2022-12-30 杭州轻宇宙科技有限公司 数字孪生虚拟人与人体的交互方法、系统、电子设备
CN115597659A (zh) * 2022-09-21 2023-01-13 山东锐翊电力工程有限公司（Cn） 一种变电站智能安全管控方法

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110600132B (zh) * 2019-08-31 2023-12-15 深圳市广宁股份有限公司 基于振动检测的数字孪生智能健康预测方法及装置
CN112132955B (zh) * 2020-09-01 2024-02-06 大连理工大学 人体骨骼的数字孪生体构建方法
CN114918976B (zh) * 2022-06-16 2022-12-02 慧之安信息技术股份有限公司 基于数字孪生技术的关节机器人健康状态评估方法

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20180011389A1 (en) * 2016-07-11 2018-01-11 Petrica-Sandel Baciu Modular Instant and Digital Back for Film TLR Cameras
CN109453497A (zh) * 2018-09-30 2019-03-12 深圳市科迈爱康科技有限公司 交互式训练方法、系统及计算机可读存储介质
CN110045608A (zh) * 2019-04-02 2019-07-23 太原理工大学 基于数字孪生的机械设备零部件结构参数动态优化方法
CN110109532A (zh) * 2018-06-11 2019-08-09 成都思悟革科技有限公司 一种基于人体姿态获得系统的人体动作比对系统
CN110600132A (zh) * 2019-08-31 2019-12-20 深圳市广宁股份有限公司 基于振动检测的数字孪生智能健康预测方法及装置

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR101665386B1 (ko) * 2010-11-15 2016-10-12 한화테크윈 주식회사 로봇 위치 추정 장치 및 방법
US20190005200A1 (en) * 2017-06-28 2019-01-03 General Electric Company Methods and systems for generating a patient digital twin
CN109635644A (zh) * 2018-11-01 2019-04-16 北京健康有益科技有限公司 一种用户动作的评价方法、装置和可读介质
CN109448815A (zh) * 2018-11-28 2019-03-08 平安科技（深圳）有限公司 自助健身方法、装置、计算机设备及存储介质
CN109621331A (zh) * 2018-12-13 2019-04-16 深圳壹账通智能科技有限公司 辅助健身方法、装置及存储介质、服务器
CN109819233B (zh) * 2019-01-21 2020-12-08 合肥哈工热气球数字科技有限公司 一种基于虚拟成像技术的数字孪生系统
CN110090012A (zh) * 2019-03-15 2019-08-06 上海图灵医疗科技有限公司 一种基于机器学习的人体疾病检测方法及检测产品
CN110189363B (zh) * 2019-05-30 2023-05-05 南京林业大学 一种机场场面移动目标低视角视频测速方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20180011389A1 (en) * 2016-07-11 2018-01-11 Petrica-Sandel Baciu Modular Instant and Digital Back for Film TLR Cameras
CN110109532A (zh) * 2018-06-11 2019-08-09 成都思悟革科技有限公司 一种基于人体姿态获得系统的人体动作比对系统
CN109453497A (zh) * 2018-09-30 2019-03-12 深圳市科迈爱康科技有限公司 交互式训练方法、系统及计算机可读存储介质
CN110045608A (zh) * 2019-04-02 2019-07-23 太原理工大学 基于数字孪生的机械设备零部件结构参数动态优化方法
CN110600132A (zh) * 2019-08-31 2019-12-20 深圳市广宁股份有限公司 基于振动检测的数字孪生智能健康预测方法及装置

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114822849A (zh) * 2022-05-06 2022-07-29 深圳市第二人民医院(深圳市转化医学研究院) 基于数字孪生的数据监测方法、装置、设备和存储介质
CN114822849B (zh) * 2022-05-06 2023-09-05 深圳市第二人民医院(深圳市转化医学研究院) 基于数字孪生的数据监测方法、装置、设备和存储介质
WO2023212994A1 (zh) * 2022-05-06 2023-11-09 深圳市第二人民医院(深圳市转化医学研究院) 基于数字孪生的数据监测方法、装置、计算机设备和存储介质
CN114968162A (zh) * 2022-06-20 2022-08-30 阿维塔科技(重庆)有限公司 车载食品的信息展示方法及装置
CN114968162B (zh) * 2022-06-20 2023-08-01 阿维塔科技(重庆)有限公司 车载食品的信息展示方法及装置
CN115597659A (zh) * 2022-09-21 2023-01-13 山东锐翊电力工程有限公司（Cn） 一种变电站智能安全管控方法
CN115597659B (zh) * 2022-09-21 2023-04-14 山东锐翊电力工程有限公司 一种变电站智能安全管控方法
CN115543094A (zh) * 2022-11-28 2022-12-30 杭州轻宇宙科技有限公司 数字孪生虚拟人与人体的交互方法、系统、电子设备
CN115543094B (zh) * 2022-11-28 2023-05-30 杭州轻宇宙科技有限公司 数字孪生虚拟人与人体的交互方法、系统、电子设备

Also Published As

Publication number Publication date
CN110600132B (zh) 2023-12-15
CN110600132A (zh) 2019-12-20

Similar Documents

Publication Publication Date Title
WO2021036635A1 (zh) 基于振动检测的数字孪生智能健康预测方法及装置
WO2021057810A1 (zh) 数据处理、训练、识别方法、装置和存储介质
Javeed et al. Wearable sensors based exertion recognition using statistical features and random forest for physical healthcare monitoring
Jain et al. Three-dimensional CNN-inspired deep learning architecture for Yoga pose recognition in the real-world environment
JP7160932B2 (ja) 動作の識別および動作情報を用いた規範的な分析の生成
US11747898B2 (en) Method and apparatus with gaze estimation
JP5527423B2 (ja) 画像処理システム、画像処理方法、及び画像処理プログラムを記憶した記憶媒体
WO2021068781A1 (zh) 一种疲劳状态识别方法、装置和设备
WO2017161734A1 (zh) 通过电视和体感配件矫正人体动作及系统
CN111274998A (zh) 帕金森病手指敲击动作识别方法及系统、存储介质及终端
Rungruanganukul et al. Deep learning based gesture classification for hand physical therapy interactive program
CN111507301B (zh) 视频处理方法、装置、计算机设备以及存储介质
Yao et al. A fall detection method based on a joint motion map using double convolutional neural networks
CN113782184A (zh) 一种基于面部关键点与特征预学习的脑卒中辅助评估系统
CN114429459A (zh) 目标检测模型的训练方法及对应的检测方法
WO2023202285A1 (zh) 图像处理方法、装置、计算机设备及存储介质
CN112507920A (zh) 一种基于时间位移和注意力机制的考试异常行为识别方法
CN114066888B (zh) 一种血流动力学指标确定方法、装置、设备及存储介质
Wang Research on the evaluation of sports training effect based on artificial intelligence technology
CN116543455A (zh) 建立帕金森症步态受损评估模型、使用方法、设备及介质
US11847810B2 (en) Face-hand correlation degree detection method and apparatus, device and storage medium
CN114022956A (zh) 一种多维度智能研判健身动作运动效果的方法
CN113869161A (zh) 一种基于人脸识别与骨骼算法实现智能跑道的方法和系统
Chen et al. EANet: Towards Lightweight Human Pose Estimation With Effective Aggregation Network
CN117423138B (zh) 基于多分支结构的人体跌倒检测方法、装置及系统

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20856497

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20856497

Country of ref document: EP

Kind code of ref document: A1