An intelligent evaluation method for teenagers' body postures based on 3D modeling

Abstract

The present invention relates to the field of 3D modeling technology, and discloses an intelligent evaluation method for the body postures of teenagers based on 3D modeling, including: First step, spatially model the body postures of teenagers by using topology methods, represent the body postures of teenagers as a high-dimensional manifold space, and capture the topological relationships between various postures in the posture space through homology group analysis of the high-dimensional manifold space; Second step, after obtaining the topological structure of the body postures of teenagers, apply manifold learning technology to reduce the dimension of the high-dimensional posture data, and use the locally linear embedding method to optimize the expression of the data. By combining topology methods and homology group analysis, the modeling of the body postures of teenagers in the high-dimensional manifold space is realized, the topological relationships between various postures in the posture space can be effectively captured, a 3D posture model is obtained, and the problem that complex posture changes cannot be accurately described in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of 3D modeling, and specifically to an intelligent evaluation method for adolescent body postures based on 3D modeling. Background Art

[0002] With the changes in modern social lifestyles, the physical health problems of adolescents have attracted extensive attention. Long-term bad sitting postures, standing postures, and unhealthy postures lacking appropriate exercise have become one of the main causes of spinal diseases, muscle fatigue, joint problems, and poor body postures in adolescents. The above problems will affect the physical health of adolescents, have a negative impact on their mental health, and even have a long-term impact on their future quality of life. Therefore, early detection and correction of adolescents' bad postures have become an important task in the fields of modern health management and education.

[0003] To effectively evaluate and correct adolescents' body postures, most studies have started to adopt 3D modeling technology. The posture evaluation method based on 3D modeling can provide three-dimensional spatial posture data compared with the traditional two-dimensional image analysis method, and can capture the body posture changes of adolescents in different environments and activity states in real time and dynamically. Therefore, the 3D modeling method has broad application prospects in posture evaluation, health management, and posture correction.

[0004] However, the existing posture evaluation methods based on 3D modeling still face challenges, which are mainly reflected in the following aspects:

[0005] Existing 3D modeling technologies usually use simplified geometric models to describe human postures, lack an in-depth understanding of the posture space, and cannot accurately reflect the complex changes between different postures of adolescents. As a result, in a dynamic environment, especially when adolescents are moving quickly and making small posture changes, accurate measurement results cannot be obtained.

[0006] Most traditional posture evaluation methods rely on static images or simple sensor data and cannot reflect the change process of postures in real time and accurately. Especially when adolescents are performing various activities, the rapid changes in postures cannot be effectively captured by traditional methods, resulting in lag and inaccuracy in posture evaluation.

[0007] Most current posture correction technologies adopt a one-size-fits-all standard solution, ignoring the differences between individuals. This method fails to fully consider the specific situations of each adolescent. Therefore, personalized posture correction guidance cannot be provided for each adolescent, resulting in limited and poor sustainability of the correction effect.

[0008] Regarding the problems raised, how to combine the latest science and technology to provide accurate, real-time, and personalized posture assessment and correction solutions has become an important topic in posture research. Therefore, those skilled in the art provide an intelligent assessment method for adolescent body postures based on 3D modeling to solve the problems raised above. Summary of the Invention

[0009] Aiming at the deficiencies of the prior art, the present invention provides an intelligent assessment method for adolescent body postures based on 3D modeling to solve the problems raised in the above background technology.

[0010] To achieve the above objectives, the present invention is realized through the following technical solutions: An intelligent assessment method for adolescent body postures based on 3D modeling, including:

[0011] In the first step, by using topology methods to perform spatial modeling on adolescent body postures, representing the adolescent body postures as a high-dimensional manifold space, and capturing the topological relationships between various postures in the posture space through homology group analysis of the high-dimensional manifold space;

[0012] In the second step, after obtaining the topological structure of the adolescent body postures, applying manifold learning techniques to reduce the dimension of the high-dimensional posture data, and using the locally linear embedding method to optimize the expression of the data, so that different changes in the adolescent body postures can be represented in a lower-dimensional space and local structure information is retained;

[0013] In the third step, based on the low-dimensional representation of the posture data, using the wave function model of quantum mechanics to describe the time evolution of the adolescent body postures, and constructing a wave function to capture the probability fluctuations of the adolescent body postures, so that during dynamic posture assessment, the change trajectory of the postures and the transition states between different postures can be reflected in real time;

[0014] In the fourth step, based on the description of the quantum wave function, using a generative adversarial network to generate 3D posture data. The generator generates a new posture model by learning the adolescent posture data, and the discriminator evaluates the difference between the generated posture and the real data to optimize the output of the generator;

[0015] In the fifth step, after optimizing the generated 3D posture model through the generative adversarial network, combining the elasticity mechanics model in physics to optimize the posture by modeling the elastic deformation of each part of the adolescent body;

[0016] In the sixth step, based on the optimization of the physical model, using the muscle dynamics model and the neural feedback mechanism to simulate the muscle activities and nerve reflexes of each part of the adolescent body;

[0017] Step 7: By integrating the results of the physics model and neurofeedback, and combining the optimized 3D pose model with real-time feedback using holographic technology, project the real-time pose evaluation results onto a holographic display device, enabling teenagers to observe their current body postures through intuitive 3D images;

[0018] Step 8: By updating the holographic feedback results in real time and combining personalized correction suggestions, provide adjustment guidance tailored to the specific situation of teenagers during the dynamic pose change process.

[0019] Preferably, the topological method includes performing topological transformation on each pose state in the high-dimensional manifold space through homology group analysis, and using the following formula to calculate the topological invariant in the pose space:

[0020]

[0021] where H k( M ) is the k-th order homology group, representing the topological invariance in the pose space;

[0022] Ker ( d k) is the kernel of the k-th order derivative, representing the part that is not affected by pose changes;

[0023] Im ( d k+1) is the image of the (k + 1)-th order derivative, representing the part of pose changes;

[0024] k represents the order of the homology group.

[0025] Preferably, the manifold learning method reduces the dimensionality of high-dimensional pose data through local linear embedding technology, using the optimization objective formula:

[0026]

[0027] where f is a function that maps high-dimensional pose data to a low-dimensional space, x i is the i-th original data point, x j is the j-th original data point, N ( i ) is the neighborhood of point x i w ij represents the relationship strength between point x i and x j , and n is the total number of data points.

[0028] Preferably, the wave function model describes the evolution process of the pose based on the Schrödinger equation in quantum mechanics. The Schrödinger equation is as follows:

[0029]

[0030] where \(i\) is the imaginary unit; \(h\) is Planck's constant, representing the unit of the quantum scale;

[0031] \(\frac{\partial}{\partial t}\psi(x,t)\) ( x,t ) is the partial derivative of the wave function \(\psi(x,t)\) with respect to time \(t\); ( x,t ) and \(\hat{H}\)

[0032] is the Hamiltonian operator;

[0033] \(\psi(x,t)\) ( x,t ) is the wave function that describes the probability distribution of the posture of teenagers at time \(t\) and position \(x\).

[0034] Preferably, the quantum wave function model determines the expected value of each posture state through the measurement operator of quantum mechanics, and the calculation formula is:

[0035]

[0036] where \(\langle\hat{O}\rangle\) is the expected value of the operator \(\hat{O}\), representing the average result of the measurement of the posture of teenagers; \(\hat{O}\)

[0037] \(\psi(x,t)\) is the wave function that describes the probability distribution of the posture of teenagers at time \(t\) and position \(x\); \(\psi^*(x,t)\) * (x,t) is the complex conjugate of the wave function; \(dx\) represents the integral over the posture space.

[0038] Preferably, during the training process of the generative adversarial network, the generator outputs a new 3D posture model, and the generator and discriminator are optimized through the following optimization objective formula:

[0039]

[0040] where \(G\) is the generator; \(D\) is the discriminator, \(D(x)\) is the output probability of the discriminator for the real data \(x\); \(G(z)\) is the fake data generated by the generator;

[0041] \(\log D(x)\) is the logarithmic probability that the discriminator determines the input \(x\) as real data.

[0042] \(\log(1 - D(G(z)))\) is the logarithmic probability that the discriminator determines the generated fake data \(G(z)\) as fake data;

[0043] \(E_x\) x～p(x) is the expected value with respect to the real data \(x\); \(E_z\) z～p(z) is the expected value with respect to the noise \(z\).

[0044] Preferably, the generative adversarial network is combined with a variational autoencoder. The variational autoencoder is used to generate pose representations in a low-dimensional latent space, and then the generative adversarial network is used to optimize the generated low-dimensional latent space representations to generate a 3D pose model.

[0045] Preferably, the optimized 3D pose model is optimized by the theory of elasticity in physics, and the optimization formula is:

[0046]

[0047] where E is the total energy of the elastic body, representing the energy generated during the pose adjustment process;

[0048] σ is the stress tensor, representing the internal force acting on each part of the body;

[0049] ε is the strain tensor, representing the degree of deformation of each part of the body;

[0050] V is the volume element; dV is the differential of the volume element.

[0051] Preferably, the physical model optimization is combined with a muscle dynamics model and a neural feedback mechanism to simulate the muscle activities and nerve reflexes of each part of the adolescent body, so that during the posture adjustment process, the muscle and nerve responses can obtain appropriate feedback and control.

[0052] Preferably, the optimized 3D pose model provides real-time feedback on the current pose of the adolescent through holographic technology, updates the pose evaluation results in real time, and combines personalized correction suggestions to provide dynamic pose adjustment guidance to ensure that the adolescent can correct bad postures in real-time feedback.

[0053] The present invention provides an intelligent evaluation method for the body posture of adolescents based on 3D modeling. It has the following

[0054] Beneficial effects:

[0055] 1. By combining the topological method and homology group analysis, the present invention realizes the modeling of the body posture of adolescents in a high-dimensional manifold space, can effectively capture the topological relationship between various postures in the posture space, obtains a 3D pose model, and solves the problem that complex pose changes cannot be accurately described in the prior art.

[0056] 2. By introducing a quantum mechanical wave function model to describe the temporal evolution of the pose, the present invention can reflect the change trajectory of the body posture of adolescents and the transition state between different postures in real time, obtain a real-time evaluation of dynamic pose changes, and effectively solve the problem of insufficient accuracy of traditional methods in dealing with rapid and minute pose changes of adolescents.

[0057] 3. By combining a muscle dynamics model and a neural feedback mechanism, the present invention provides personalized posture feedback and correction suggestions, which can be adjusted in real time according to the specific posture needs of teenagers, obtaining a scientific and personalized posture correction plan, and solving the deficiency of lacking personalized posture correction in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 It is a flowchart of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0059] To enable those skilled in the art to understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0060] The present invention will be described in detail below with reference to the accompanying drawings:

[0061] Embodiment:

[0062] Please refer to the attached Figure 1 , the embodiment of the present invention provides an intelligent evaluation method for the body posture of teenagers based on 3D modeling, including:

[0063] In the first step, by using the topology method to perform spatial modeling on the body posture of teenagers, representing the body posture of teenagers as a high-dimensional manifold space, and capturing the topological relationship between various postures in the posture space through homology group analysis of the high-dimensional manifold space;

[0064] In the second step, after obtaining the topological structure of the body posture of teenagers, applying manifold learning technology to reduce the dimension of the high-dimensional posture data, and using the locally linear embedding method to optimize the expression of the data, so that different changes in the body posture of teenagers can be represented in a lower-dimensional space and the local structure information is retained;

[0065] In the third step, based on the low-dimensional representation of the posture data, using the wave function model of quantum mechanics to describe the time evolution of the body posture of teenagers, and constructing a wave function to capture the probability fluctuations of the body posture of teenagers, so that during dynamic posture evaluation, the change trajectory of the posture and the transition state between different postures can be reflected in real time;

[0066] In the fourth step, based on the description of the quantum wave function, using a generative adversarial network to generate 3D posture data. The generator generates a new posture model by learning the posture data of teenagers, and the discriminator evaluates the difference between the generated posture and the real data to optimize the output of the generator;

[0067] In the fifth step, after optimizing the generated 3D pose model through the generative adversarial network, the pose is optimized by combining the elasticity mechanics model in physics, and the elastic deformation of each part of the adolescent body is modeled;

[0068] In the sixth step, on the basis of the optimized physical model, the muscle dynamics model and the neural feedback mechanism are used to simulate the muscle activities and neural reflexes of each part of the adolescent body;

[0069] In the seventh step, by synthesizing the results of the physics model and neural feedback, and combining holographic technology to provide real-time feedback on the optimized 3D pose model, the real-time pose evaluation results are projected onto the holographic display device, and adolescents can observe their current body postures through intuitive 3D images;

[0070] In the eighth step, by updating the holographic feedback results in real time and combining personalized correction suggestions, during the dynamic pose change process, adjustment guidance for the specific situation of adolescents is provided.

[0071] The advantage of the first step is that it can model the complex changes in body postures and capture the topological features in pose changes. It helps to solve the problem that the complex relationships in the pose space cannot be effectively represented in traditional modeling methods, and provides a theoretical basis for subsequent pose modeling and evaluation.

[0072] The advantage of the second step is to reduce the complexity of high-dimensional data, enabling pose data to be effectively represented in a low-dimensional space. It retains important local structural information and avoids information loss.

[0073] The advantage of the third step is that the wave function can capture the tiny changes in poses in a probabilistic way, enhancing the accurate description of dynamic pose changes. It realizes the precise evaluation of fast and tiny pose changes and solves the deficiencies in traditional dynamic evaluation methods.

[0074] The advantage of the fourth step is that the generative adversarial network can continuously optimize the generation model, improving the accuracy and authenticity of 3D pose data. Through adversarial training, the gap between the generated pose data and the actual data is avoided, making the generated pose model conform to the real adolescent body postures.

[0075] The advantage of the fifth step is to optimize the pose through physical constraints, ensuring compliance with biomechanical principles during pose adjustment. By simulating the elastic responses of body parts, it can provide scientific and operable pose correction schemes.

[0076] The advantage of the sixth step is that it can simulate the real-time feedback of muscles and nerves during actual pose adjustment, ensuring the personalization and scientific nature of the pose correction scheme. It provides real-time feedback and adjustment, enhancing the effectiveness and operability of pose correction.

[0077] The benefit of the seventh step is to provide intuitive holographic feedback, enabling teenagers to observe and correct bad postures in real time, enhancing interactivity and sense of participation. The real-time updated feedback information helps teenagers continuously adjust their postures during activities.

[0078] The benefit of the eighth step is to provide personalized posture correction programs, solving the problem of "one-size-fits-all" in traditional technologies. Through continuous feedback and personalized suggestions, it helps teenagers gradually develop correct body postures, with significant long-term effects.

[0079] The topological method includes performing topological transformations on each posture state in the high-dimensional manifold space through homology group analysis, and using the following formula to calculate the topological invariants in the posture space:

[0080]

[0081] where H k( M ) is the k-th homology group, representing the topological invariance in the posture space;

[0082] Ker ( d k) is the kernel of the k-th derivative, representing the part that is not affected by posture changes;

[0083] Im ( d k+1) is the image of the (k + 1)-th derivative, representing the part of posture changes;

[0084] k represents the order of the homology group.

[0085] Through homology group analysis, the topological relationships between various postures in the posture space can be effectively captured. Different from traditional methods, the topological method can handle complex posture transformations and can identify topological invariants that are not easily detected in postures, thereby accurately modeling complex body posture changes.

[0086] Using homology group analysis, through the calculation of the k-th homology group, the topological invariance in the posture space can be retained. Topological invariance represents the inherent characteristics between postures, which is particularly important for dealing with complex posture changes and can ensure that key biological characteristics and motion patterns are not lost during the transformation of postures.

[0087] High-dimensional posture data contains a large amount of redundant information, and homology group analysis reduces the complexity of posture data to the lowest by extracting topological invariance.

[0088] The manifold learning method reduces the dimension of high-dimensional posture data through local linear embedding technology, using the optimization objective formula:

[0089]

[0090] Among them, f is a function that maps high-dimensional pose data to a low-dimensional space, and x i is the i-th original data point, and x j is the j-th original data point, N ( i ) is the neighborhood of point x i , and w ij represents the relationship strength between point x i and x j . n is the total number of data points.

[0091] High-dimensional pose data usually contains a large amount of redundant information, making data processing and calculation very complex. By reducing the data dimension to a low-dimensional space, the computational complexity can be effectively reduced, making subsequent pose analysis and modeling more efficient.

[0092] The locally linear embedding technique can preserve the local structure of the original data during the dimensionality reduction process. Specifically, the locally linear embedding method can preserve the neighborhood relationship between data points, so that the low-dimensional space after dimensionality reduction can accurately represent the structure of similar data points in the high-dimensional space.

[0093] The dimensionality reduction operation through optimizing the objective formula can ensure that the performance of high-dimensional data in the low-dimensional space can still reflect the real body pose characteristics. The locally linear embedding method preserves the relationship between data points, making the data after dimensionality reduction conform to the actual body pose changes and improving the accuracy of the model.

[0094] The redundant information in high-dimensional data will affect the performance of the model. The locally linear embedding technique reduces the data dimension, preserves important feature information, removes unnecessary noise and redundancy, and ensures the simplicity and interpretability of the model.

[0095] The wave function model is based on the Schrödinger equation in quantum mechanics to describe the evolution process of the pose. The Schrödinger equation is as follows:

[0096]

[0097] Among them, i is the imaginary unit; h is the Planck constant, representing the unit of the quantum scale;

[0098] &ψ ( x,t ) is the partial derivative of the wave function ψ ( x,t ) with respect to time t;

[0099] is the Hamiltonian operator;

[0100] ψ ( x,t )is the wave function that describes the probability distribution of the posture of adolescents at time t and position x.

[0101] The Schrödinger equation can accurately describe the change of posture over time through the evolution of the quantum wave function, especially the subtle dynamic posture changes. Compared with traditional physical modeling methods, the wave function model can capture extremely subtle posture fluctuations, which is particularly important for real-time assessment of the posture changes of adolescents, especially in sports or rapidly changing postures.

[0102] The Schrödinger equation describes the wave function that changes with time. Therefore, with the help of this model, the body posture changes of adolescents can be tracked in real time in a dynamic environment. This method can accurately simulate the transition state of the posture, reflect the smooth change between different postures, and improve the accuracy and real-time performance of dynamic posture assessment.

[0103] Through the wave function model, the Schrödinger equation can capture the current posture state and predict the future change trajectory of the posture. Its prediction ability is particularly applicable to the posture changes of adolescents in daily activities and sports, and can provide predictive guidance for posture correction.

[0104] The posture change itself is a highly complex dynamic system. The wave function model can handle the changes of the system from the perspective of quantum mechanics, providing more detailed posture analysis and evaluation than traditional methods. It can effectively handle high-dimensional and non-linear posture changes and is not restricted by the linear assumptions in traditional methods.

[0105] The quantum wave function model determines the expected value of each posture state through the measurement operator of quantum mechanics. The calculation formula is:

[0106]

[0107] where is the expected value of the operator , representing the average result of the measurement of the posture of adolescents; is the operator, representing the observable quantity of the measurement of the posture of adolescents;

[0108] ψ ( x,t ) is the wave function that describes the probability distribution of the posture of adolescents at time t and position x; ψ * (x,t) is the complex conjugate of the wave function; dx represents the integration over the posture space.

[0109] The quantum measurement operator can calculate the expected value of the posture state and provide a numerical value in complex posture evaluation. Compared with traditional posture measurement methods, the quantum measurement theory can use the probability distribution of the wave function to optimize the measurement results, reduce measurement errors, and improve measurement stability.

[0110] By calculating the expected value of the wave function, the posture states of teenagers at different time points and different positions can be obtained. This method can integrate multi-measurement data to ensure that the dynamic changes of postures are completely captured, enabling the system to reflect the overall trend of teenagers' body postures.

[0111] Since the quantum measurement operator can calculate the expected value of the posture in real time, this method can dynamically adjust the posture evaluation model according to the latest measurement data, enabling the system to respond to posture changes in real time and providing posture correction guidance for teenagers.

[0112] Traditional posture evaluation methods are vulnerable to noise interference. However, the quantum measurement method can effectively reduce the influence of external interference on measurement results through the mathematical description of the wave function, improve the stability of posture evaluation, and ensure that the system can still provide evaluation data in complex environments.

[0113] During the training process of the generative adversarial network, the generator outputs a new 3D posture model, and the generator and discriminator are optimized through the following optimization objective formula:

[0114]

[0115] Among them, G is the generator; D is the discriminator, D(x) is the output probability of the discriminator for the real data x; G(z) is the fake data generated by the generator;

[0116] logD(x) is the logarithmic probability that the discriminator determines the input x as real data.

[0117] log(1 - D(G(z))) is the logarithmic probability that the discriminator determines the generated fake data G(z) as fake data;

[0118] E x～p(x) is the expected value for the real data x; E z～p(z) is the expected value for the noise z.

[0119] The generative adversarial network combines with the variational autoencoder. The variational autoencoder is used to generate the posture representation in the low-dimensional latent space, and then the generated low-dimensional latent space representation is optimized through the generative adversarial network to generate a 3D posture model.

[0120] Through the adversarial training of the generator and the discriminator in the generative adversarial network, the generator continuously optimizes its output to generate a 3D posture model close to the real one. The discriminator optimizes the output of the generator based on its determination of the real data and the generated data, ensuring that the finally generated 3D posture model has a high degree of authenticity and accuracy.

[0121] By introducing a variational autoencoder, the present invention maps the pose data to a low-dimensional latent space through the variational autoencoder, making the representation of the pose data compact and easy to process. Subsequently, the generative adversarial network optimizes the low-dimensional latent space to generate diverse and reasonable 3D pose models. This method can avoid the problem of insufficient pose diversity in traditional methods, enabling the generated pose models to cover a variety of different pose variations and enhancing the flexibility of the system.

[0122] The optimized 3D pose model is optimized through the theory of elasticity in physics, and the optimization formula is:

[0123]

[0124] where E is the total energy of the elastic body, representing the energy generated during the pose adjustment process;

[0125] σ is the stress tensor, representing the internal forces acting on various parts of the body;

[0126] ε is the strain tensor, representing the degree of deformation of various parts of the body;

[0127] V is the volume element; dV is the differential of the volume element.

[0128] Through the description of stress and strain, the elasticity model ensures that the pose optimization process conforms to the basic principles of biomechanics. During the actual pose adjustment process, various parts of the body will deform when an external force is applied. The elasticity model can describe the deformation to ensure that the optimized pose will not cause unreasonable biomechanical reactions and avoid muscle and bone injuries caused by pose adjustment.

[0129] This method adopts an optimization strategy of energy minimization, keeping the body pose adjustment process within the physically feasible range and ensuring the smoothness and naturalness of pose changes.

[0130] Through the description of the stress tensor and the strain tensor, elasticity can analyze the deformation of body parts during pose adjustment. This is particularly important for pose assessment because the degree of deformation and the internal force distribution of various body parts will have different effects on the overall pose.

[0131] The physical model optimization combines the muscle dynamics model and the neural feedback mechanism to simulate the muscle activities and neural reflexes of various parts of the adolescent body, so that during the pose adjustment process, the muscle and nerve responses can receive appropriate feedback and control.

[0132] The optimized 3D pose model provides real-time feedback on the current pose of the adolescent through holographic technology, updates the pose assessment results in real time, and combines personalized correction suggestions to provide dynamic pose adjustment guidance, ensuring that the adolescent can correct bad postures in real-time feedback.

[0133] The present invention combines a muscle dynamics model and a neurofeedback mechanism, and through real-time feedback holographic technology, can provide personalized, real-time and dynamic posture correction solutions for teenagers. By simulating muscle activities and nerve reflexes, the system ensures that the body's internal responses during posture adjustment are scientifically controlled. At the same time, through holographic feedback, teenagers can observe and adjust their postures in real time, effectively improving the accuracy, comfort and persistence of posture correction. This method can enhance the personalization and accuracy of posture correction, and can provide support for the long-term health management of teenagers, helping to prevent and correct teenagers' posture problems, and thus promoting the healthy growth of teenagers.

[0134] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent evaluation method for teenagers' body postures based on 3D modeling, characterized in that, Including: The first step is to perform spatial modeling on the body postures of teenagers by using topology methods, represent the body postures of teenagers as a high-dimensional manifold space, and capture the topological relationships between various postures in the posture space through homology group analysis of the high-dimensional manifold space; The second step is to apply manifold learning techniques to reduce the dimension of the high-dimensional posture data after obtaining the topological structure of the body postures of teenagers, and use the locally linear embedding method to optimize the expression of the data, so that different changes in the body postures of teenagers can be represented in a lower-dimensional space while retaining local structure information; The third step is to describe the time evolution of the body postures of teenagers through the wave function model of quantum mechanics on the basis of the low-dimensional representation of the posture data, capture the probability fluctuations of the body postures of teenagers by constructing a wave function, so that during dynamic posture evaluation, the change trajectory of the posture and the transition state between different postures can be reflected in real time; The fourth step is to generate 3D posture data by using a generative adversarial network on the basis of the description of the quantum wave function. The generator generates a new posture model by learning the posture data of teenagers, and the discriminator evaluates the difference between the generated posture and the real data to optimize the output of the generator; The fifth step is to optimize the posture by combining the elastic mechanics model in physics after optimizing the 3D posture model generated by the generative adversarial network, and model the elastic deformation of each part of the teenager's body; The sixth step is to use the muscle dynamics model and the neural feedback mechanism to simulate the muscle activities and nerve reflexes of each part of the teenager's body on the basis of the optimized physical model; The seventh step is to project the optimized 3D posture model in real time by combining the results of the physical model and neural feedback through holographic technology, project the real-time posture evaluation results onto a holographic display device, and teenagers can observe their current body postures through intuitive 3D images; The eighth step is to provide adjustment guidance for the specific situation of teenagers during the dynamic posture change process by updating the holographic feedback results in real time and combining personalized correction suggestions.

2. The intelligent evaluation method for the body posture of teenagers based on 3D modeling according to claim 1, wherein, The topology method includes performing topological transformation on each posture state in the high-dimensional manifold space through homology group analysis, and using the following formula to calculate the topological invariant in the posture space: Among them, H k (M) is the k-th homology group, representing the topological invariance in the attitude space; Ker(d k ) is the kernel of the k-th order derivative, representing the part that is not affected by attitude changes; Im(d k+1 ) is the image of the (k + 1)-th order derivative, representing the part of the attitude change; k represents the order of the homology group.

3. The intelligent evaluation method for the body posture of teenagers based on 3D modeling according to claim 1, characterized in that, The manifold learning method reduces the dimension of the high-dimensional posture data through locally linear embedding technology, and uses the optimization objective formula: where f is a function that maps high-dimensional pose data to a low-dimensional space, x i is the i-th original data point, x j is the j-th original data point, N(i) is the neighborhood of the point x i , w ij represents the strength of the relationship between the point x i and x j , and n is the total number of data points.

4. The intelligent evaluation method for the body posture of teenagers based on 3D modeling according to claim 1, characterized in that, The wave function model describes the evolution process of the posture based on the Schrödinger equation in quantum mechanics, and the Schrödinger equation is as follows: where i is the imaginary unit; h is the Planck constant, representing the unit of the quantum scale; &ψ(x,t) is the partial derivative of the wave function ψ(x,t) with respect to time t; is the Hamiltonian operator; ψ(x,t) is the wave function, describing the probability distribution of the teenager's posture at time t and position x.

5. A method for intelligent evaluation of teenagers' body postures based on 3D modeling according to claim 1, characterized in that, The quantum wave function model determines the expected value of each posture state through the measurement operator of quantum mechanics, and the calculation formula is: Among them, is the operator 's expected value, representing the average result of the measurement of the posture of teenagers; is the operator, representing the observed value of the measurement of the posture of teenagers; ψ(x,t) is the wave function that describes the probability distribution of the posture of adolescents at time t and position x; ψ * (x,t) is the complex conjugate of the wave function; dx represents the integration over the posture space.

6. The intelligent evaluation method for the body posture of teenagers based on 3D modeling according to claim 1, characterized in that The generative adversarial network generates a new 3D posture model during the training process, and optimizes the generator and discriminator through the following optimization objective formula: Among them, G is the generator; D is the discriminator, and D(x) is the output probability of the discriminator for the real data x; G(z) is the fake data generated by the generator; logD(x) is the logarithmic probability that the discriminator determines the input x as real data; log(1 - D(G(z))) is the logarithmic probability that the discriminator determines the generated fake data G(z) as fake data; E x～p(x) is the expected value of the true data x; E z～p(z) is the expected value of the noise z.

7. An intelligent evaluation method for the body postures of teenagers based on 3D modeling according to claim 6, characterized in that, The generative adversarial network combines with a variational autoencoder. The variational autoencoder is used to generate pose representations in the low-dimensional latent space, and then the generated low-dimensional latent space representations are optimized through the generative adversarial network to generate a 3D pose model.

8. The intelligent evaluation method for the body posture of teenagers based on 3D modeling according to claim 1, characterized in that, The optimized 3D pose model is optimized by the theory of elasticity in physics, and the optimization formula is: Among them, E is the total energy of the elastic body, representing the energy generated during the pose adjustment process; σ is the stress tensor, representing the internal force acting on each part of the body; ε is the strain tensor, representing the degree of deformation of each part of the body; V is the volume element; dV is the differential of the volume element.

9. The intelligent evaluation method for the body posture of teenagers based on 3D modeling according to claim 1, characterized in that, The optimization of the physical model combines a muscle dynamics model and a neural feedback mechanism to simulate the muscle activities and neural reflexes of each part of the adolescent body, so that during the pose adjustment process, the muscle and nerve responses can receive appropriate feedback and control.

10. The intelligent evaluation method for the body posture of teenagers based on 3D modeling according to claim 1, wherein The optimized 3D pose model provides real-time feedback on the current pose of the adolescent through holographic technology, updates the pose assessment results in real time and combines personalized correction suggestions to provide dynamic pose adjustment guidance to ensure that the adolescent can correct bad postures in real-time feedback.

Images