A health assessment method and system based on 3D human posture image model

Abstract

The present invention relates to the technical field of human health assessment, and in particular to a health assessment method and system based on a 3D human posture image model. The method comprises the following steps: acquiring human annular scan data; extracting dynamic posture sequences of the human annular scan data in upright, flexed, and lateral positions, and performing phase-shifted fringe projection on the human annular scan data according to the dynamic posture sequence to obtain three-dimensional point cloud data of the human body; analyzing the sagittal, coronal, and transverse planes of the three-dimensional point cloud data of the human body for orthogonal projection to generate a multi-angle projection image group of the human body; identifying the skeletal contour features of the multi-angle projection image group of the human body to establish a dynamic marking system for anatomical landmarks. The present invention achieves precision, standardization, and personalization of 3D human posture health assessment through high-precision dynamic acquisition and multi-dimensional spatial analysis, thereby improving the accuracy and comprehensiveness of human posture health assessment.

Description

Technical Field

[0001] The present invention relates to the technical field of human health assessment, and in particular to a health assessment method and system based on a 3D human posture image model. Background Art

[0002] Early health assessments relied primarily on physician experience and simple two-dimensional image analysis, which failed to fully and accurately reflect human posture and health status. With advances in deep learning and sensor technology, three-dimensional human posture capture technology has continued to mature, evolving from initial marker-based motion capture systems to markerless posture recognition technologies that utilize multi-camera and multi-sensor fusion, significantly improving the accuracy and real-time nature of data acquisition. In recent years, health assessment methods based on 3D posture images, combined with machine learning models, have enabled intelligent identification of motor dysfunction, skeletal abnormalities, and poor posture by analyzing the dynamic changes of key skeletal points. However, existing technologies for studying healthy sleep posture primarily rely on pressure sensing, analyzing the pressure distribution of various body parts in a reclining position. This approach has numerous limitations. For one thing, relying solely on pressure data cannot fully capture human posture information, nor can it accurately determine key data such as the angles and spatial relationships of various body parts, resulting in inaccurate and incomplete posture assessments. Furthermore, pressure sensing technology is significantly affected by environmental factors. For example, the material and thickness of the mattress can interfere with the accuracy of pressure data, thereby affecting the final health assessment results. Summary of the Invention

[0003] Based on this, it is necessary to provide a health assessment method and system based on a 3D human posture image model to solve at least one of the above technical problems.

[0004] To achieve the above object, a health assessment method based on a 3D human posture image model is provided, the method comprising the following steps:

[0005] Step S1: acquiring human body annular scan data; extracting dynamic posture sequences of the human body annular scan data in upright, forward bending, and side bending positions, and performing phase-shifted fringe projection on the human body annular scan data according to the dynamic posture sequences to obtain three-dimensional point cloud data of the human body;

[0006] Step S2: Analyze the sagittal, coronal, and transverse planes of the three-dimensional point cloud data of the human body and perform orthogonal projection to generate a multi-angle projection image group of the human body; identify the bone contour features of the multi-angle projection image group of the human body to establish a dynamic marking system for anatomical landmarks; use the dynamic marking system to perform spatial posture calibration on the bone contour features to generate spatial posture calibration parameters;

[0007] Step S3: Inputting the spatial posture calibration parameters into a preset health posture assessment database for multi-dimensional comparative analysis, and calculating the deviation index between the parts of the human body 3D point cloud data and the standard value based on the analysis results to construct a biomechanical load distribution map and a joint range of motion thermal map;

[0008] Step S4: construct a 3D human posture image model based on the multi-angle projection image group of the human body; perform posture health assessment on the 3D human posture image model through the biomechanical load distribution diagram and the joint range of motion thermal map, and perform personalized posture correction according to the assessment results, thereby generating a personalized posture correction report.

[0009] This invention captures dynamic posture sequences in upright, flexed, and lateral positions and combines them with phase-shifted fringe projection technology to achieve high-precision three-dimensional reconstruction of human posture in multiple motion states, enhancing data integrity and dynamic expressiveness. Utilizing multi-angle orthogonal projections of sagittal, coronal, and transverse planes, it automatically identifies skeletal contour features and establishes a dynamic labeling system for spatial posture calibration, ensuring the accuracy and repeatability of subsequent analysis. By comparing calibration parameters with a healthy posture database across multiple dimensions, a deviation index is calculated, accurately characterizing the difference between the individual and the standard. Biomechanical load distribution maps and joint range of motion heatmaps are further generated to achieve multidimensional health assessment. An intuitive 3D posture image model is constructed based on the multi-angle projections. Overlay analysis of heatmaps and load maps enhances visualization and interactivity, making it easier for users and physicians to intuitively understand posture issues. Based on the individual's load distribution and range of motion analysis, a personalized correction strategy is developed, ultimately generating a customized posture correction report, helping to improve the accuracy and effectiveness of posture interventions. Therefore, the present invention realizes the precision, standardization and personalization of 3D human posture health assessment through high-precision dynamic acquisition and multi-dimensional spatial analysis, thereby improving the accuracy and comprehensiveness of human posture health assessment.

[0010] Preferably, step S1 includes the following steps:

[0011] Step S11: performing a 360° circular scan of the human body using a scanning device to obtain human body circular scan data;

[0012] Step S12: extracting a scanning sequence of the human body annular scanning data, and performing inter-frame synchronous registration on the human body annular scanning data according to the scanning sequence to generate synchronously corrected human body scanning frame data;

[0013] Step S13: extracting dynamic skeleton points from the human body scan frame data; annotating the dynamic skeleton points with postures to generate dynamic posture sequences of upright position, forward bending position, and side bending position;

[0014] Step S15: performing spatial posture parameter decomposition on the dynamic posture sequence to obtain human posture distribution control data; performing fringe phase encoding and multi-phase offset fringe pattern generation on the human body scanning frame data to obtain a multi-phase offset encoding atlas;

[0015] Step S16: performing phase unpacking on the multi-phase offset coding atlas using the human body posture distribution control data to generate a dense phase map; performing triangulation and depth inversion calculation on the dense phase map to generate high-precision three-dimensional point cloud data of the human body.

[0016] This invention utilizes a scanning device to perform a 360° circular scan of the human body, acquiring complete raw scan data surrounding the body surface. This improves data integrity and provides comprehensive information support for subsequent modeling and analysis. Inter-frame synchronous registration methods uniformly correct the scan sequence, effectively addressing spatiotemporal errors and dynamic misalignment caused by human motion in multi-frame data, and enhancing the stability and accuracy of point cloud reconstruction. Dynamic skeleton point extraction and posture annotation ensure that the scan data not only contains surface morphology but also dynamic posture information (such as upright, flexed, and lateral bending), providing a structured foundation for subsequent individual posture assessment and motor function analysis. Dynamic posture sequences are converted into posture distribution control data, enabling spatially decoupled control of different human postures. Fringe phase encoding and multi-phase offset processing are combined to enhance the accuracy and anti-interference capability of depth acquisition. Leveraging phase unpacking, dense mapping, and depth inversion techniques, high-density, low-error 3D point cloud data is generated from the phase atlas, achieving submillimeter accuracy and effectively supporting subsequent spatial analysis and visualization modeling. The overall process supports dynamic scanning and motion recognition scenarios. It is not only suitable for static modeling, but also for scenarios with high requirements for dynamic features such as rehabilitation assessment and motion analysis, expanding the application boundaries.

[0017] Preferably, the posture annotation of the dynamic skeleton points in step S13 includes:

[0018] The three-dimensional coordinates of the dynamic skeleton points are collected by an optical motion capture system to obtain the three-dimensional coordinates of the skeleton points, wherein the sampling frequency is set to ≥30 frames / second, the spatial accuracy is ±2mm, the coordinate unit is millimeter, and the right-hand coordinate system is referenced;

[0019] Definition of upright position identification criteria: overall spinal inclination ≤5°, left and right shoulder height difference ≤10mm, knee flexion angle ≤10°;

[0020] Definition of flexion position identification criteria: overall spinal anteversion angle ≥30° and ≤90°, cervical to lumbar curvature radius ≤400 mm, hip flexion angle ≥45°;

[0021] Scoliosis identification criteria were defined as follows: lateral spinal curvature angle ≥20° and ≤60°, left and right hip height difference ≥15mm, and the horizontal distance of the shoulder-hip line deviating from the central axis ≥30mm;

[0022] Based on the defined upright position recognition standards, forward bending position recognition standards and lateral bending position recognition standards, the three-dimensional coordinates of the skeleton points are classified and labeled, thereby obtaining the dynamic posture sequences of upright position, forward bending position and lateral bending position.

[0023] This invention utilizes an optical motion capture system to ensure high temporal resolution and spatial accuracy of human skeletal coordinates, within the constraints of a sampling frequency of ≥30 frames / second and a spatial accuracy of ±2mm. This ensures that the coordinates of the captured human skeleton points have high temporal resolution and spatial accuracy, adapting to the analysis requirements of complex dynamic posture changes. By quantitatively defining upright, flexed, and lateral positions (e.g., overall spinal inclination, shoulder height difference, and joint flexion angles), an objective, repeatable, and programmable posture recognition standard system is established, eliminating the subjectivity and inconsistency of traditional manual assessment. Based on a matching algorithm based on the three-dimensional coordinates of the skeleton points and the set recognition criteria, each frame of the scanned data is automatically classified, generating high-quality posture sequence data that provides a precise annotation foundation for subsequent three-dimensional modeling and health assessment. By utilizing recognition criteria that incorporate multiple geometric indicators (e.g., inclination angle, curvature radius, joint angle, horizontal offset), the system can accurately identify complex non-standard postures such as flexion and lateral bending, improving sensitivity to abnormal postures and potential health risks. The three-dimensional coordinates of the skeleton points are based on the right-hand coordinate system and are expressed in millimeters. This facilitates compatibility with existing three-dimensional reconstruction algorithms, anatomical databases, and posture simulation models, and supports cross-platform and cross-system data interaction and processing.

[0024] Preferably, in step S2, analyzing the sagittal plane, coronal plane, and transverse section of the three-dimensional point cloud data of the human body for orthogonal projection includes:

[0025] Normalize the spatial coordinate center of gravity of the human body's three-dimensional point cloud data to generate standardized three-dimensional posture point cloud data;

[0026] Extract the local principal axis of the standardized 3D pose point cloud data, and use the local principal axis to calibrate the body axis vector field of the 3D pose point cloud data to generate structure alignment vector calibration data;

[0027] The structural alignment vector calibration data is rearranged in depth mapping according to the sagittal plane direction to generate a sagittal orthogonal projection image;

[0028] The structural alignment vector calibration data is aggregated and compressed in the coronal direction to generate a coronal orthogonal projection image.

[0029] The structural alignment vector calibration data is sliced ​​and resampled in the cross-sectional direction to generate a cross-sectional orthogonal projection image;

[0030] The orthogonal projection images of the sagittal plane, coronal plane and transverse section are registered and bound with angle labels to generate a multi-angle projection image group of the human body.

[0031] The present invention effectively eliminates the influence of individual height and body shape differences on posture analysis results through spatial coordinate center of gravity normalization processing, so that the three-dimensional point cloud data has a consistent reference frame, and improves the alignment accuracy and analysis versatility between subsequent diverse body posture data. By adopting the local principal axis extraction and body axis vector field calibration strategy, vector field calibration data that conforms to the biological structure of the human body can be constructed, so that each point in the point cloud has a clear anatomical direction meaning, and strengthens the structural constraints of feature recognition and posture comparison. By rearranging the depth mapping of the sagittal plane, coronal plane and transverse section, aggregation compression and hierarchical slice resampling, the spatial depth and structural features in the three-dimensional posture are effectively retained, and accurate visual expression of the human body posture in the two-dimensional plane is achieved. Through image group registration and angle label binding, the output multi-angle projection image has consistent spatial semantics and data labels, which is convenient for input into convolutional neural networks, posture recognition models or biomechanical simulation platforms, and realizes efficient automatic posture recognition and evaluation. The orthogonal projection process divides the original three-dimensional point cloud into multidimensional structural views organized by anatomical planes, so that abnormal posture features such as scoliosis, anteversion or vertebral rotation can be significantly enhanced in contrast on a specific projection plane, thereby improving the sensitivity and accuracy of structural abnormality identification.

[0032] Preferably, identifying the skeleton contour features of the multi-angle projection image group of the human body in step S2 to establish a dynamic marking system for anatomical landmarks includes:

[0033] Identify the main skeleton contours of a multi-angle projection image group of the human body;

[0034] Analyze the connected domain of the multi-angle projection image group of the human body according to the main skeleton contour to obtain the primary skeleton distribution map;

[0035] Detect key morphological turning points on the primary skeleton distribution map to generate candidate anatomical structure corner point data, and perform spatial consistency test on the candidate anatomical structure corner point data based on depth prior to generate a high-confidence anatomical landmark point set;

[0036] Perform temporal pose dynamic matching and trajectory interpolation on high-confidence anatomical landmark points to generate dynamic annotation sequence data;

[0037] By dynamically annotating sequence data, the posture angle change modeling and structural stability evaluation of high-confidence anatomical landmark points are performed to generate a dynamic labeling system for anatomical landmark points.

[0038] This method automatically identifies the main skeletal contours in multi-angle projection image sets and generates a primary skeleton distribution map based on connected domain analysis, eliminating the need for traditional manual annotation. This significantly improves the automation and efficiency of posture structure recognition, making it suitable for processing and analyzing large-scale posture data. A consistency check mechanism combining key morphological turning point detection with a deep prior ensures that only anatomical landmarks with structural consistency and high spatial confidence are retained, effectively mitigating feature location deviations caused by factors such as image occlusion and projection blur, thereby enhancing the accuracy and robustness of anatomical recognition. By performing temporal dynamic matching and trajectory interpolation on a set of high-confidence anatomical landmarks, a continuous posture change annotation sequence is established. This helps track the structural changes in an individual's posture during exercise or weight-bearing, providing data support for dynamic posture assessment and disease early warning. Based on the dynamic annotation sequence data, a posture angle change modeling mechanism is established, which accurately measures the displacement trajectory and rotation angle of key anatomical parts. Combined with structural stability assessment, this mechanism further characterizes the biomechanical risk characteristics during posture changes, enhancing the depth and reliability of healthy posture assessment.

[0039] Preferably, in step S2, performing spatial posture calibration on the skeleton contour features using the dynamic marking system includes:

[0040] Perform spatial center-of-gravity registration on the anatomical landmark sequence in the dynamic marking system to generate initial posture alignment reference data;

[0041] Perform three-dimensional rigid registration on the bone contour features, and use the initial posture alignment reference data to perform coarse registration on the rigidly registered bone contour features to generate a skeleton model;

[0042] Perform non-rigid fine-tuning optimization based on landmark points on the coarse registration skeleton model to generate a high-precision pose alignment model;

[0043] Extract joint angles and solve the posture transformation matrix for the high-precision posture alignment model to generate a local posture transformation parameter set;

[0044] The local attitude transformation parameter set is globally reconciled and error convergence controlled to generate spatial attitude calibration parameters.

[0045] The present invention automatically generates initial posture alignment reference data by performing spatial center of gravity registration on the anatomical landmark point sequence in the dynamic marking system, effectively avoiding manual point selection errors, and is conducive to improving the convergence speed and initialization robustness of subsequent registration steps. Three-dimensional rigid registration of skeletal contour features can quickly unify the individual's main viewing direction in space and align it with the standard coordinate system, and maintain basic structural consistency in operations such as multi-frame data fusion and dynamic posture comparison. On the basis of rigid registration, the introduction of landmark-driven non-rigid fine-tuning optimization can effectively repair local alignment deviations caused by motion differences, soft tissue deformation or scanning errors, and improve the registration accuracy and detail fidelity of the posture model. The joint angle information and posture transformation matrix extracted based on the high-precision posture alignment model can provide high-dimensional parameter support for complex joint motion modeling, motion recognition and motion analysis, significantly improving the expressive power of human motion data. By performing global reconciliation and error convergence control on the local posture transformation parameter set, the continuity and physiological rationality of the posture transformation relationship between each joint or torso segment can be ensured, thereby building a stable and accurate spatial posture calibration model. The generated spatial posture calibration parameters can be widely used in the normalization processing and time series comparison of multi-frame dynamic posture data, improving the comparability and quantifiable feature expression ability of the posture evolution process, and providing key technical support for posture training and behavior analysis.

[0046] Preferably, step S3 includes the following steps:

[0047] Step S31: inputting the spatial posture calibration parameters into a preset health posture assessment database for posture similarity matching to generate a standard posture matching analysis result;

[0048] Step S32: Calculate the three-dimensional deviation vector of each skeleton joint node based on the standard posture matching result data to generate a joint deviation index matrix;

[0049] Step S33: analyzing the point cloud voxel density of the human body three-dimensional point cloud data, and performing mechanical structure mapping in combination with the joint deviation index matrix to generate a biomechanical load distribution map;

[0050] Step S34: Perform angle interval analysis and motion amplitude reconstruction based on the joint deviation index matrix to generate a joint range of motion heat map.

[0051] The present invention automatically identifies the structural differences between the standard healthy posture and the standard healthy posture by inputting the spatial posture calibration parameters into a preset healthy posture assessment database for posture similarity matching, and provides a unified control model for posture comparison in different populations or specific disease states. Based on the standard matching results, the three-dimensional deviation vector of each skeleton joint node is calculated and the joint deviation index matrix is ​​constructed, which can comprehensively quantify the size, direction and range of the posture error of each joint in space, and effectively improve the resolution and accuracy of posture abnormality assessment. By using the voxel density analysis of the human body three-dimensional point cloud data and the joint deviation index matrix for fusion modeling, it can effectively reflect the biomechanical risk areas such as local stress concentration and load offset caused by posture abnormality, and provide a basis for clinical rehabilitation and posture correction. By performing angle interval analysis and motion amplitude reconstruction on the joint deviation index, a heat map that intuitively reflects the activity capacity of each joint is generated, which helps to quickly identify activity-restricted areas, evaluate functional degradation trends, and support rehabilitation course design and effect tracking.

[0052] Preferably, step S31 includes the following steps:

[0053] Step S311: When the posture calibration parameters meet any of the following conditions, it is determined that the preliminary matching has failed and preliminary posture matching failure data is generated: the displacement error of the posture key node exceeds the set threshold of 10cm; the joint angle deviation exceeds 15°; the overall posture stability score is lower than 0.6;

[0054] Step S312: When the following conditions are met simultaneously, it is determined that there is a characteristic region posture mismatch and characteristic posture mismatch data is obtained: the angular coordination between the core joint groups decreases by more than 25%; the motion trajectory deviation exceeds the standard range of the historical health model; the matching residual is greater than the set upper limit of 0.08 in three consecutive evaluations, where the core joint group includes the shoulder, spine, and pelvis;

[0055] Step S313: When all of the following conditions occur, the overall posture is determined to be abnormal and posture abnormality data is obtained: the overall 3D posture alignment error exceeds 12%; the body symmetry score is lower than 0.7; the motion sequence continuity score is lower than 0.65; and the error cannot be corrected by the adaptive compensation algorithm during the matching process;

[0056] Step S314: Integrate the preliminary posture matching failure data, the characteristic posture mismatch data, and the posture abnormality data to generate the analysis results of the standard posture matching.

[0057] By incorporating multiple criteria, including key node displacement errors, joint angle deviations, and overall posture stability scores, this system can quickly and accurately identify initial posture matching failures, avoiding redundant computations in subsequent analysis and improving overall system efficiency. By dynamically monitoring the angular coordination, motion trajectory, and matching residuals of the core joint group (shoulder, spine, and pelvis), it accurately identifies local posture mismatches, promptly identifying subtle but critical posture dysfunctions and improving the sensitivity and specificity of anomaly detection. By integrating metrics such as three-dimensional posture alignment error, body symmetry scores, and motion sequence continuity, combined with an adaptive compensation algorithm, it ensures that only uncorrectable anomalies are identified as overall posture anomalies, reducing false positives and improving the reliability of posture anomaly detection. By integrating data on initial matching failures, feature mismatches, and overall anomalies, it enables a multi-dimensional, multi-level analysis of posture matching status, providing reliable and comprehensive data support for subsequent posture correction, health assessment, and clinical decision-making.

[0058] Preferably, step S4 includes the following steps:

[0059] Step S41: reconstructing the depth profile of the multi-angle projection image group of the human body using the coarse registration skeleton model to construct a 3D human posture image model; mapping the biomechanical load distribution map to the surface of the 3D human posture image model to generate load stress distribution modeling data;

[0060] Step S42: Projecting the joint range of motion heat map onto the skeleton structure of the 3D human posture image model to generate a dynamic range of motion parameter set; performing a multi-dimensional assessment of posture health based on the load stress distribution modeling data and the dynamic range of motion parameter set to generate a posture abnormality marking layer;

[0061] Step S43: Identifying posture optimization areas based on the posture abnormality marker layer, matching personalized correction strategies with a preset standard posture model, and generating posture correction control parameters; applying the posture correction control parameters to key posture nodes of the 3D human posture image model to generate posture correction simulation results;

[0062] Step S44: Perform personalized interpretation and generate an evaluation summary of the posture correction simulation results, and finally output a personalized posture correction report.

[0063] This invention utilizes a coarsely registered skeleton model to reconstruct the depth profile of multi-angle projection image sets, achieving a precise construction of a three-dimensional human posture image model. This provides an accurate spatial foundation for subsequent biomechanical load mapping and posture function analysis. Biomechanical load distribution maps and joint range of motion heat maps are mapped to the three-dimensional model surface and skeleton structure, respectively, enabling a dynamic combined assessment of load stress and mobility, effectively revealing potential areas of posture abnormalities and enhancing the scientific and comprehensive nature of posture health assessment. Based on a posture abnormality marker layer, abnormal areas are intelligently identified and personalized correction strategies are matched with standard posture models, enabling precise localization and targeted regulation of abnormal areas, improving the relevance and effectiveness of correction. Correction and control parameters are applied to key nodes of the three-dimensional posture model to perform dynamic simulations of posture correction, facilitating verification of the actual effectiveness of correction strategies in a virtual environment, reducing clinical trial-and-error costs and enhancing the scientific nature of personalized interventions. By providing detailed personalized interpretation of the posture correction simulation results and generating an evaluation summary, an easy-to-understand and professional posture correction report is ultimately generated, helping patients and clinicians fully understand their posture status and correction plan.

[0064] In this specification, a health assessment system based on a 3D human posture image model is provided, which is used to perform the above-mentioned health assessment method based on a 3D human posture image model. The health assessment system based on a 3D human posture image model includes:

[0065] The point cloud projection module is used to obtain human annular scan data; extract the dynamic posture sequence of the human annular scan data in upright position, forward bending position and side bending position, and perform phase shift fringe projection on the human annular scan data according to the dynamic posture sequence to obtain three-dimensional point cloud data of the human body;

[0066] The posture analysis module is used to analyze the sagittal, coronal, and transverse planes of the human body's three-dimensional point cloud data for orthogonal projection, generating a set of multi-angle projection images of the human body; identifying the skeletal contour features of the multi-angle projection image set of the human body to establish a dynamic marking system for anatomical landmarks; and using the dynamic marking system to perform spatial posture calibration on the skeletal contour features to generate spatial posture calibration parameters.

[0067] The posture deviation calculation module is used to input the spatial posture calibration parameters into the preset health posture assessment database for multi-dimensional comparative analysis. Based on the analysis results, it calculates the deviation index between the parts in the human body 3D point cloud data and the standard value to construct a biomechanical load distribution map and joint range of motion thermal map;

[0068] The health assessment module is used to construct a 3D human posture image model based on a group of multi-angle projection images of the human body; the 3D human posture image model is evaluated for posture health through biomechanical load distribution diagrams and joint range of motion thermal maps, and personalized posture correction is performed based on the evaluation results, thereby generating a personalized posture correction report.

[0069] The beneficial effect of the present invention is that it accurately captures the dynamic posture sequence of the human body in upright, flexed and lateral positions through circular scanning combined with phase-shifted fringe projection technology, effectively improving the integrity and dynamic capture capability of the three-dimensional point cloud data, and providing a solid data foundation for subsequent posture analysis. By performing orthogonal projection of the three-dimensional point cloud data in the sagittal, coronal and transverse planes, a multi-angle projection image group is generated. Combined with the bone contour feature recognition and the dynamic marking system of anatomical landmarks, accurate spatial calibration of the human skeleton posture is achieved, effectively improving the spatial accuracy and dynamic consistency of posture analysis. The spatial posture calibration parameters are matched and compared with the healthy posture database in multiple dimensions to accurately calculate the deviation index of each part of the skeleton, and a biomechanical load distribution map and joint range of motion heat map are constructed, thereby enhancing the quantitative assessment capability of posture abnormalities and the understanding of structural mechanics. A three-dimensional human posture model is constructed based on the multi-angle projection image group, integrating load distribution and range of motion information to achieve comprehensive health assessment and precise correction of individual posture, and generate a personalized posture correction report. These modules work together to form a complete closed loop, from human body dynamic data collection and spatial posture analysis to health assessment and personalized intervention. This significantly enhances the automation, intelligence, and practical value of human posture health management. Therefore, through high-precision dynamic data collection and multidimensional spatial analysis, this invention achieves precise, standardized, and personalized 3D human posture health assessment, improving its accuracy and comprehensiveness. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Figure 1 A schematic diagram of the steps of a health assessment method based on a 3D human posture image model;

[0071] Figure 2 for Figure 1 Detailed implementation steps of step S3 in FIG.

[0072] Figure 3 for Figure 1 Detailed implementation steps of step S4 in FIG.

[0073] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0074] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. It is obvious that the embodiments described are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts are within the scope of protection of the present invention.

[0075] In addition, the accompanying drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. Identical reference numerals in the figures denote identical or similar parts, and thus repetitive descriptions thereof will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities that do not necessarily correspond to physically or logically separate entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor and / or microcontroller approaches.

[0076] It should be understood that although the terms "first," "second," and the like may be used herein to describe various elements, these elements should not be limited by these terms. These terms are used solely to distinguish one element from another. For example, a first element may be referred to as a second element, and similarly, a second element may be referred to as a first element, without departing from the scope of the exemplary embodiments. The term "and / or" as used herein includes any and all combinations of one or more of the listed associated items.

[0077] To achieve this, please refer to Figures 1 to 3 A health assessment method based on a 3D human body posture image model comprises the following steps:

[0078] Step S1: acquiring human body annular scan data; extracting dynamic posture sequences of the human body annular scan data in upright, forward bending, and side bending positions, and performing phase-shifted fringe projection on the human body annular scan data according to the dynamic posture sequences to obtain three-dimensional point cloud data of the human body;

[0079] Step S2: Analyze the sagittal, coronal, and transverse planes of the three-dimensional point cloud data of the human body and perform orthogonal projection to generate a multi-angle projection image group of the human body; identify the bone contour features of the multi-angle projection image group of the human body to establish a dynamic marking system for anatomical landmarks; use the dynamic marking system to perform spatial posture calibration on the bone contour features to generate spatial posture calibration parameters;

[0080] Step S3: Inputting the spatial posture calibration parameters into a preset health posture assessment database for multi-dimensional comparative analysis, and calculating the deviation index between the parts of the human body 3D point cloud data and the standard value based on the analysis results to construct a biomechanical load distribution map and a joint range of motion thermal map;

[0081] Step S4: construct a 3D human posture image model based on the multi-angle projection image group of the human body; perform posture health assessment on the 3D human posture image model through the biomechanical load distribution diagram and the joint range of motion thermal map, and perform personalized posture correction according to the assessment results, thereby generating a personalized posture correction report.

[0082] This invention captures dynamic posture sequences in upright, flexed, and lateral positions and combines them with phase-shifted fringe projection technology to achieve high-precision three-dimensional reconstruction of human posture in multiple motion states, enhancing data integrity and dynamic expressiveness. Utilizing multi-angle orthogonal projections of sagittal, coronal, and transverse planes, it automatically identifies skeletal contour features and establishes a dynamic labeling system for spatial posture calibration, ensuring the accuracy and repeatability of subsequent analysis. By comparing calibration parameters with a healthy posture database across multiple dimensions, a deviation index is calculated, accurately characterizing the difference between the individual and the standard. Biomechanical load distribution maps and joint range of motion heatmaps are further generated to achieve multidimensional health assessment. An intuitive 3D posture image model is constructed based on the multi-angle projections. Overlay analysis of heatmaps and load maps enhances visualization and interactivity, making it easier for users and physicians to intuitively understand posture issues. Based on the individual's load distribution and range of motion analysis, a personalized correction strategy is developed, ultimately generating a customized posture correction report, helping to improve the accuracy and effectiveness of posture interventions. Therefore, the present invention realizes the precision, standardization and personalization of 3D human posture health assessment through high-precision dynamic acquisition and multi-dimensional spatial analysis, thereby improving the accuracy and comprehensiveness of human posture health assessment.

[0083] In the embodiment of the present invention, reference Figure 1 FIG. 1 is a flow chart showing the steps of a health assessment method based on a 3D human posture image model according to the present invention. In this example, the health assessment method based on a 3D human posture image model includes the following steps:

[0084] Step S1: acquiring human body annular scan data; extracting dynamic posture sequences of the human body annular scan data in upright, forward bending, and side bending positions, and performing phase-shifted fringe projection on the human body annular scan data according to the dynamic posture sequences to obtain three-dimensional point cloud data of the human body;

[0085] Step S2: Analyze the sagittal, coronal, and transverse planes of the three-dimensional point cloud data of the human body and perform orthogonal projection to generate a multi-angle projection image group of the human body; identify the bone contour features of the multi-angle projection image group of the human body to establish a dynamic marking system for anatomical landmarks; use the dynamic marking system to perform spatial posture calibration on the bone contour features to generate spatial posture calibration parameters;

[0086] Step S3: Inputting the spatial posture calibration parameters into a preset health posture assessment database for multi-dimensional comparative analysis, and calculating the deviation index between the parts of the human body 3D point cloud data and the standard value based on the analysis results to construct a biomechanical load distribution map and a joint range of motion thermal map;

[0087] Step S4: construct a 3D human posture image model based on the multi-angle projection image group of the human body; perform posture health assessment on the 3D human posture image model through the biomechanical load distribution diagram and the joint range of motion thermal map, and perform personalized posture correction according to the assessment results, thereby generating a personalized posture correction report.

[0088] In this embodiment of the present invention, a high-precision, multi-view 3D scanner is used to perform a 360° circular scan of the human body, acquiring data in upright, flexed, and lateral positions. An image pose recognition algorithm (such as OpenPose or MediaPipe) is then used to classify and annotate the scanned data sequence, generating a dynamic pose sequence set. In each pose, phase-shift profilometry is used to dynamically optically encode the body surface. The depth of each pixel is calculated using a multi-phase interpolation method, combining parallax and geometric constraints to generate high-density 3D point cloud data of the human body. The point cloud data is then filtered, denoised, and edge-filled to output a complete human point cloud model. The human point cloud data is then orthogonally projected onto the sagittal, coronal, and transverse planes. High-resolution projection images are generated at each viewpoint, forming a multi-angle image set covering the entire body. Based on deep learning image semantic segmentation models (such as U-Net or HRNet), the outlines of the main skeletal structures of the human body (cervical spine, thoracic spine, pelvis, etc.) are extracted; traceable anatomical landmarks are set up at key nodes, and a dynamic landmark marking system (DLMS) is constructed to track the displacement changes of landmarks under different postures. The position changes of multi-frame anatomical landmarks in DLMS are used to establish a multi-view coordinate system registration model; through the rigid body transformation matrix (Rigid Transformation Matrix) and the posture inverse algorithm, a complete set of spatial posture calibration parameters (including rotation angle, offset vector, etc.) is generated. The posture calibration parameters are input into the health posture assessment database (containing a large number of three-dimensional skeleton posture models of standard people); the deviations of different parts (such as spinal angle, pelvic tilt, etc.) from the standard values ​​are compared, and the formula is used to calculate the position of the landmarks. ;in For the The deviation index of the part, For the The posture parameters measured at the part, is the standard value. The finite element estimation algorithm (FEM) is used to simulate the pressure conditions of each joint and bone on the human body point cloud; a color heat map of the joint range of motion is generated based on the angle change amplitude and the active radius; the output results are presented in the form of a visual atlas for downstream evaluation and personalized intervention. A complete 3D human posture model is reconstructed based on point cloud and skeleton data; multi-angle projection images are integrated to improve modeling accuracy, and topology correction and skin texture mapping are performed. The biomechanical load map and joint heat map are superimposed on the 3D model for comprehensive analysis; based on deviation indicators, abnormal load areas and joint stiffness indexes, trunk asymmetry, scoliosis degree, posture stability, etc. are evaluated. Based on the analysis results, the system generates a correction report including the following contents: the location and direction of posture deviation, visual heat map and load map, personalized posture training suggestions (such as stretching, corrective movements, use of assistive devices), and a posture improvement progress tracking template.

[0089] Preferably, step S1 includes the following steps:

[0090] Step S11: performing a 360° circular scan of the human body using a scanning device to obtain human body circular scan data;

[0091] Step S12: extracting a scanning sequence of the human body annular scanning data, and performing inter-frame synchronous registration on the human body annular scanning data according to the scanning sequence to generate synchronously corrected human body scanning frame data;

[0092] Step S13: extracting dynamic skeleton points from the human body scan frame data; annotating the dynamic skeleton points with postures to generate dynamic posture sequences of upright position, forward bending position, and side bending position;

[0093] Step S15: performing spatial posture parameter decomposition on the dynamic posture sequence to obtain human posture distribution control data; performing fringe phase encoding and multi-phase offset fringe pattern generation on the human body scanning frame data to obtain a multi-phase offset encoding atlas;

[0094] Step S16: performing phase unpacking on the multi-phase offset coding atlas using the human body posture distribution control data to generate a dense phase map; performing triangulation and depth inversion calculation on the dense phase map to generate high-precision three-dimensional point cloud data of the human body.

[0095] In one embodiment of the present invention, a structured light scanning device (or a ToF + RGB fusion system) with multi-view synchronous imaging is employed, with multiple high-resolution cameras and projection units positioned around the human body. The projection system is controlled to project structured light stripe patterns onto the human body at fixed angle intervals, while the multiple cameras simultaneously capture images reflected from the human body surface. A synchronized triggering mechanism and a high-frame-rate camera (>60fps) are used to ensure motion blur-free scanning throughout the entire process, outputting a raw image sequence of a circular scan of the human body. The viewpoint parameters and timestamps of each frame in the raw image sequence are extracted to establish a time-angle index model. Inter-frame synchronous registration algorithms (such as optical flow registration or depth alignment based on the ICP algorithm) are used to synchronize and spatially align frames from different viewpoints. The resulting synchronized and corrected human scan frame dataset is output, representing the alignment of all frames at the same time and in a unified spatial coordinate system. Based on a deep learning skeleton recognition network (such as OpenPose 3D or MediaPipe Skeleton), a sequence of 3D skeleton key points (such as shoulders, elbows, spine, and pelvis) is extracted from the synchronized rectified scan frames. Based on the temporal changes of the skeleton points, a dynamic pose classifier (such as a support vector machine or LSTM classification model) is applied to label the sequence as: upright, forward bending, and lateral bending. The resulting dynamic pose sequence dataset is then output with the three labeled categories. For each type of dynamic posture sequence, spatial posture parameters (pitch, yaw, roll) are extracted based on Euler angle decomposition or quaternion decomposition algorithms. Each frame of posture is represented as a six-degree-of-freedom (6DoF) posture vector to construct complete human posture distribution control data. Phase-Shift Encoding (Phase-Shift Encoding) is applied to each frame of data to generate a structured light fringe pattern: four sets of fringe patterns with 90° phase shifts (i.e., 0°, 90°, 180°, and 270°) are projected to obtain a multi-phase offset encoding atlas. This outputs a high-quality structured light encoding atlas for subsequent 3D reconstruction. The human posture distribution control data is used to perform spatially guided unpacking of the encoding atlas to avoid phase errors caused by posture changes. Phase calculations are performed for each pixel (using: ;in ∼ is the grayscale value of the four phase-shifted images) to generate a dense phase map; triangulation and depth inversion calculation are performed based on the phase map: the depth value of each pixel is calculated by combining the phase difference value through the calibrated spatial baseline of the projection-camera pair; the depth map is converted into three-dimensional coordinates, and the point clouds of each view are fused. The point cloud registration + fusion filtering algorithm (such as Poisson reconstruction) is used to generate the final: high-precision human body three-dimensional point cloud data (accuracy can reach sub-millimeter level).

[0096] Preferably, the posture annotation of the dynamic skeleton points in step S13 includes:

[0097] The three-dimensional coordinates of the dynamic skeleton points are collected by an optical motion capture system to obtain the three-dimensional coordinates of the skeleton points, wherein the sampling frequency is set to ≥30 frames / second, the spatial accuracy is ±2mm, the coordinate unit is millimeter, and the right-hand coordinate system is referenced;

[0098] Definition of upright position identification criteria: overall spinal inclination ≤5°, left and right shoulder height difference ≤10mm, knee flexion angle ≤10°;

[0099] Definition of flexion position identification criteria: overall spinal anteversion angle ≥30° and ≤90°, cervical to lumbar curvature radius ≤400 mm, hip flexion angle ≥45°;

[0100] Scoliosis identification criteria were defined as follows: lateral spinal curvature angle ≥20° and ≤60°, left and right hip height difference ≥15mm, and the horizontal distance of the shoulder-hip line deviating from the central axis ≥30mm;

[0101] Based on the defined upright position recognition standards, forward bending position recognition standards and lateral bending position recognition standards, the three-dimensional coordinates of the skeleton points are classified and labeled, thereby obtaining the dynamic posture sequences of upright position, forward bending position and lateral bending position.

[0102] In an embodiment of the present invention, an optical motion capture system (such as Vicon or OptiTrack) is used, and one or more infrared high-speed cameras are deployed to form a closed capture area. Highly reflective markers are attached to key bone points of the human body (head, cervical spine, acromion, elbow, wrist, thoracic spine, lumbar spine, hip, knee, and ankle) to construct a human skeleton point marking model. The system configuration is as follows: sampling frequency: ≥30 frames / second, spatial accuracy: ±2mm, coordinate unit: millimeter (mm), reference coordinate system: right-handed coordinate system (X forward, Y left, Z up), and a three-dimensional coordinate data sequence of skeleton points in continuous time frames is output. The angle between the line from the seventh cervical vertebra (C7) to the sacral vertebra (S1) and the Z axis is calculated, which is the overall inclination of the spine. The absolute height difference between the left shoulder high point and the right shoulder high point on the Z axis is calculated, which is the height difference between the left and right shoulders. The angle between the thigh vector and the calf vector is calculated, which is the knee flexion angle. The above-mentioned upright position recognition features are standardized by a preset threshold. The anteversion angle between the line connecting C7 and S1 and the Z axis is calculated, representing the spinal anteversion angle. The spinal axis is reconstructed using spline interpolation. The local radius of curvature of the fitted curve is calculated as the cervical-lumbar curvature radius. The angle between the trunk and thigh is calculated as the hip flexion angle. Similarly, a preset threshold is used to standardize the features for flexion position recognition described above. The offset angle between the line connecting the thoracic spine midpoint and the pelvic center in the YZ plane is calculated, representing the scoliosis angle. The height difference between the left and right hips on the Z axis is calculated, representing the left-right hip height difference. The maximum offset of the line connecting the left shoulder to the right hip in the X-axis relative to the midline (the sagittal plane) is calculated, representing the distance the shoulder-hip line deviates from the midline. A preset threshold is used to standardize the features for scoliosis position recognition described above. The skeleton data of each frame is traversed, and each frame is sequentially extracted and judged to meet a specific recognition criteria group (upright, flexion, scoliosis). A priority judgment mechanism is used: if multiple categories match, scoliosis > flexion > upright is prioritized. The classification results of consecutive frames are combined into a posture time series, and a dynamic posture sequence is output.

[0103] Preferably, in step S2, analyzing the sagittal plane, coronal plane, and transverse section of the three-dimensional point cloud data of the human body for orthogonal projection includes:

[0104] Normalize the spatial coordinate center of gravity of the human body's three-dimensional point cloud data to generate standardized three-dimensional posture point cloud data;

[0105] Extract the local principal axis of the standardized 3D pose point cloud data, and use the local principal axis to calibrate the body axis vector field of the 3D pose point cloud data to generate structure alignment vector calibration data;

[0106] The structural alignment vector calibration data is rearranged in depth mapping according to the sagittal plane direction to generate a sagittal orthogonal projection image;

[0107] The structural alignment vector calibration data is aggregated and compressed in the coronal direction to generate a coronal orthogonal projection image.

[0108] The structural alignment vector calibration data is sliced ​​and resampled in the cross-sectional direction to generate a cross-sectional orthogonal projection image;

[0109] The orthogonal projection images of the sagittal plane, coronal plane and transverse section are registered and bound with angle labels to generate a multi-angle projection image group of the human body.

[0110] In an embodiment of the present invention, the acquired 3D point cloud data of the human body undergoes spatial coordinate normalization. This process uses the body's overall center of gravity as a reference and spatially translates the data of all points to achieve normalization of the point cloud data's center of gravity. This step ensures that subsequent analysis is based on a unified spatial reference system, which helps maintain posture consistency and data stability. From the normalized point cloud data, the principal directions reflecting the body's overall structure are extracted, including the longitudinal, transverse, and anterior-posterior directions. These extracted directions serve as the body's structural reference axes for spatial structural alignment of the human body. Subsequently, based on these structural directions, each point in the 3D point cloud is assigned directional information corresponding to its spatial position, completing the calibration of the structural alignment vector. Based on the calibrated structural directions, the 3D point cloud of the human body is rearranged along the left-right direction to display the body's posture characteristics when viewed from the side. Specifically, the point cloud data is projected onto a plane reflecting the body's anterior-posterior and posterior-posterior dimensions, forming an orthogonal projection in the sagittal plane. This image clearly illustrates the contour changes of the spine, head, hips, and other parts when viewed from the side. Using the anterior-posterior axis as the aggregation dimension, the point cloud data is compressed and integrated in this direction to generate a projection image of the human body structure viewed from the front. This image effectively demonstrates the left-right symmetry and degree of deformation of parts such as the shoulders, thorax, and pelvis, making it suitable for structural assessment and posture analysis. By slicing the body layer by layer along the longitudinal direction, the point cloud data at different heights is hierarchically divided and projected to generate a cross-sectional image viewed from top to bottom. This image facilitates analysis of body features such as width distribution, rotation angles, and local symmetry at different height levels, and is particularly suitable for analyzing the abdominal, waist, and pelvic structures. The projection images generated in these three directions are spatially aligned to ensure visual consistency in image center position, image size, and edge information. An angle identification label is then assigned to each image to clearly indicate the viewing direction it represents, such as side view, front view, or top view, facilitating subsequent analysis and model processing. Finally, a set of human projection images from multiple viewing angles is generated, supporting applications such as posture recognition, structural comparison, and intelligent analysis.

[0111] Preferably, identifying the skeleton contour features of the multi-angle projection image group of the human body in step S2 to establish a dynamic marking system for anatomical landmarks includes:

[0112] Identify the main skeleton contours of a multi-angle projection image group of the human body;

[0113] Analyze the connected domain of the multi-angle projection image group of the human body according to the main skeleton contour to obtain the primary skeleton distribution map;

[0114] Detect key morphological turning points on the primary skeleton distribution map to generate candidate anatomical structure corner point data, and perform spatial consistency test on the candidate anatomical structure corner point data based on depth prior to generate a high-confidence anatomical landmark point set;

[0115] Perform temporal pose dynamic matching and trajectory interpolation on high-confidence anatomical landmark points to generate dynamic annotation sequence data;

[0116] By dynamically annotating sequence data, the posture angle change modeling and structural stability evaluation of high-confidence anatomical landmark points are performed to generate a dynamic labeling system for anatomical landmark points.

[0117] In an embodiment of the present invention, a multi-angle projection image set of the human body is preprocessed, including image enhancement, edge refinement, and noise filtering. Subsequently, a recognition algorithm based on shape gradient changes and edge continuity is used to extract the main skeletal contour regions in each projection image. The extracted main skeletal contours typically include structures such as the skull edge, spinal midline, shoulder contour, pelvic boundary, and limb backbone lines. These regions possess distinct geometric features and are easily separated from the background. Within the obtained main skeletal contour images, image connectivity analysis methods are used to identify regions with continuous pixel connectivity, thereby demarcating connected domains corresponding to different anatomical structures. Combining the topological relationships, directional information, and contour extension characteristics of each connected domain, a rough skeleton orientation map, i.e., a primary skeleton distribution map, is constructed to describe the spatial extension paths of the main bony structures of the human body. Within the primary skeleton distribution map, morphological turning points along its structural path are further identified. These turning points typically correspond to locations where bones connect, bend, or protrude, such as the transition from the cervical to thoracic vertebrae, the acromion, and the outer edge of the hip bone. By calculating the positions, angle changes, and neighborhood geometric features of these turning points, a set of candidate anatomical corner point data is extracted. These candidate corner points are then spatially verified for consistency across projections at different angles. This verification is based on the principle of rationality in human posture changes. Incorporating human structural symmetry, depth information hierarchy, and anatomical knowledge, outliers and artifacts are eliminated, ultimately resulting in a set of highly reliable anatomical landmarks. This set covers key body regions and exhibits good spatial stability and anatomical significance. Based on the changes in human motion between frames, dynamic trajectories of high-confidence anatomical landmarks are matched across time points. By tracking points between consecutive frames, dynamic trajectories of anatomical landmarks over time are generated. In the event of missing landmarks due to occlusion, noise, or other factors, trajectory interpolation algorithms based on motion patterns and structural coherence are used to complete them, generating complete dynamic annotation sequence data. Based on this dynamic annotation sequence data, the posture and angle changes of each anatomical landmark at different time points are modeled, and the amplitude, frequency, and coordination of these changes are calculated. By combining the relative position relationship and movement direction trend between each point, the stability and coordination of the anatomical structure in the dynamic process are evaluated, thus forming a dynamic marking system of anatomical landmarks for motion analysis and posture recognition.

[0118] Of particular importance, the key morphological turning point detection of the primary skeleton distribution map also includes:

[0119] Perform curvature change detection on the primary skeleton distribution map to generate local curvature extreme point data of the skeleton;

[0120] Perform directional gradient analysis on the local curvature extreme point data of the skeleton to generate candidate posture mutation point data;

[0121] Perform spatial continuity filtering on candidate posture mutation point data to generate turning point stability screening data;

[0122] Perform structural consistency matching on turning point stability screening data to generate corresponding point data of potential anatomical structures;

[0123] Identify the semantic geometric features of potential anatomical structure corresponding point data to obtain candidate anatomical structure corner point data.

[0124] In this embodiment of the present invention, the contours of continuous skeleton segments in the primary skeleton distribution map are traced to extract local geometric features of each skeletal path segment. The local curvature changes of the skeletal path are assessed using the spatial angle between consecutive points, the curve change trend, and the morphological fluctuations. Locations with significant curvature changes are marked, and extreme curvature points that form abrupt trends are extracted. This process can initially identify key areas where the skeletal structure undergoes deformation or directional deviation. Based on the acquired curvature extreme points, directional gradient information is further calculated for the area surrounding each extreme point, including features such as the magnitude of sudden changes in the skeleton's main direction, left-right deflection trends, and direction discontinuities. A threshold is set to filter the magnitude of directional changes, retaining only points with obvious directional abrupt changes as candidate pose mutation points. These points typically correspond to joint transitions, structural connections, or posture changes. Spatial continuity analysis is performed on the distribution of candidate pose mutation points throughout the primary skeleton map. Candidate points are excluded if they fail to form a stable structural continuity within their adjacent areas or lack consistency with the upstream and downstream skeleton directions. Only points that exhibit recurring, directional continuity, and stable position across multiple adjacent projections or frames are retained to form the stability screening data for turning points. This process effectively eliminates false turning points caused by image noise or local errors. Based on this stability screening, each candidate turning point is spatially matched to a known human anatomical structure model. Anatomical structure attribution is determined for each candidate point by comparing relative distances and angular relationships between points, as well as the topological paths of their skeletal branches. Points with high consistency in structural attributes with a known anatomical site are identified as potential anatomical points, providing a physical basis for subsequent semantic recognition. Finally, semantic recognition is performed on these potential anatomical points, combining their geometric morphological features with anatomical functional characteristics for classification. This process primarily determines whether a point belongs to a semantically distinct turning structure, such as the acromion, hip spine, knee joint, or ankle point, based on information such as its spatial position, morphological characteristics, and kinematic significance. Points selected through this recognition process are ultimately confirmed as candidate anatomical turning points, providing key support for dynamic posture modeling.

[0125] Preferably, in step S2, performing spatial posture calibration on the skeleton contour features using the dynamic marking system includes:

[0126] Perform spatial center-of-gravity registration on the anatomical landmark sequence in the dynamic marking system to generate initial posture alignment reference data;

[0127] Perform three-dimensional rigid registration on the bone contour features, and use the initial posture alignment reference data to perform coarse registration on the rigidly registered bone contour features to generate a skeleton model;

[0128] Perform non-rigid fine-tuning optimization based on landmark points on the coarse registration skeleton model to generate a high-precision pose alignment model;

[0129] Extract joint angles and solve the posture transformation matrix for the high-precision posture alignment model to generate a local posture transformation parameter set;

[0130] The local attitude transformation parameter set is globally reconciled and error convergence controlled to generate spatial attitude calibration parameters.

[0131] In an embodiment of the present invention, spatial coordinates are calculated for a sequence of anatomical landmarks collected in a dynamic labeling system to determine their overall center of gravity. By aligning the centers of gravity of the landmark sequences in different poses, spatial deviations caused by translation are eliminated, and unified initial pose alignment reference data is generated. This step provides a stable spatial reference for subsequent registration, ensuring overall positional consistency across different skeletal poses. Next, the skeletal contour features are converted into a three-dimensional point cloud model, and rotational and translational registration is performed in three-dimensional space based on the principle of rigid transformation. Using the initial pose alignment reference data generated in the first step, the rigidly registered skeletal contour model is adjusted to achieve a rough spatial alignment, resulting in a coarsely registered skeleton model. This model exhibits good structural alignment but does not yet address local detail discrepancies. Based on the coarse registration, a non-rigid registration algorithm is used to fine-tune the skeleton model. By applying local positional constraints to key anatomical landmarks in the dynamic labeling system and combining surface deformation with the model, the local pose of the skeletal contour is adjusted to more accurately match the actual human skeletal morphology. This process eliminates local deformation errors, improves registration accuracy, and generates a high-precision pose alignment model. Using a high-precision posture alignment model, angles are measured at the human skeletal joints to extract the rotation angles and spatial orientation changes of each joint. Based on this joint angle information, the corresponding local posture transformation matrix is ​​calculated to form a set of local posture transformation parameters. These parameters accurately describe the spatial rotation and transformation relationships of each skeletal component and serve as the foundation for subsequent posture analysis and adjustment. Finally, the local posture transformation parameter set is globally harmonized to ensure consistency in spatial position and orientation between each local transformation. Through iterative calculations to achieve error convergence, the transformation parameters are adjusted to achieve optimal coordination of the overall posture. This process effectively reduces registration errors, ensures the accuracy and stability of the spatial posture calibration parameters, and ultimately generates reliable spatial posture calibration parameters.

[0132] It is particularly important to perform three-dimensional rigid registration of the bone contour features and use the initial posture alignment reference data to perform coarse registration of the rigidly registered bone contour features. Skeleton model generation also includes:

[0133] Extract key bone feature points of bone contour features;

[0134] Based on the key bone feature point set, perform 3D rigid transformation estimation to generate a preliminary rigid registration matrix;

[0135] Performing rigid transformation on bone contour features using the preliminary rigid registration matrix to generate rigid registration bone contour data;

[0136] By using the preset initial alignment reference data, the coarse registration error metric of the rigid registration bone contour data is calculated to generate the registration error index data;

[0137] Optimize the preliminary rigid registration matrix according to the registration error index data to generate an optimized rigid registration matrix;

[0138] The optimized rigid registration matrix is ​​used to perform a secondary transformation on the bone contour features to generate precisely aligned bone contour data;

[0139] A skeleton topology model is constructed based on the precisely aligned bone contour data to generate a coarse registration skeleton model.

[0140] In an embodiment of the present invention, a set of representative key skeletal feature points are identified and extracted from skeletal contour feature data. These points typically correspond to important anatomical nodes of the human skeleton, ensuring that subsequent registration processes can achieve accurate alignment based on these key points. Based on the extracted key skeletal feature points, a preliminary rigid registration matrix is ​​estimated by calculating the optimal rigid transformation relationship between the point set in space. This matrix, which contains rotation and translation parameters, preliminarily describes the spatial transformation of the skeletal contour from its original pose to its target pose. Using the preliminarily estimated rigid registration matrix, the skeletal contour features are spatially rigidly transformed so that their overall shape and position align with the target reference pose, thereby forming preliminary rigidly registered skeletal contour data. Using pre-set initial alignment reference data as a standard, an error assessment is performed on the rigidly registered skeletal contour data. The distance differences between the skeletal key points and the reference data are calculated to form a registration error index, which reflects the quality of the registration accuracy. Based on the registration error index, the parameters of the preliminary rigid registration matrix are adjusted through an iterative optimization method to reduce the registration error. The optimization process aims to find a more accurate rotation and translation combination to achieve a higher precision in the spatial matching between the skeletal contour features and the reference data. Using the optimized rigid registration matrix, a second rigid transformation is performed on the skeletal contour features, resulting in more precisely aligned skeletal contour data, ensuring that the skeletal morphology closely matches the initial pose reference. Based on this precisely aligned skeletal contour data, the skeleton's topological connections are established, integrating the skeletal nodes and their spatial connections to generate a complete coarsely registered skeletal model. This model preserves the skeletal spatial structure while also providing a foundation for subsequent non-rigid fine-tuning.

[0141] As an example of the present invention, refer to Figure 2 As shown, in this example, step S3 includes:

[0142] Step S31: inputting the spatial posture calibration parameters into a preset health posture assessment database for posture similarity matching to generate a standard posture matching analysis result;

[0143] Step S32: Calculate the three-dimensional deviation vector of each skeleton joint node based on the standard posture matching result data to generate a joint deviation index matrix;

[0144] Step S33: analyzing the point cloud voxel density of the human body three-dimensional point cloud data, and performing mechanical structure mapping in combination with the joint deviation index matrix to generate a biomechanical load distribution map;

[0145] Step S34: Perform angle interval analysis and motion amplitude reconstruction based on the joint deviation index matrix to generate a joint range of motion heat map.

[0146] In this embodiment of the present invention, the parameters obtained through spatial posture calibration are integrated and input into a pre-established healthy posture assessment database. This database contains multiple sets of rigorously collected and screened standard posture template data. Each template corresponds to the three-dimensional coordinates and posture feature information of key skeletal nodes of the human body in a specific posture. The system compares the input spatial posture parameters with each standard template in the database frame by frame, using similarity calculation methods (such as point-to-point distance calculation and angle difference analysis) to evaluate the degree of match between the input posture and each standard template. The system then sorts the standard posture templates based on the degree of match and selects the one that most closely matches the input posture. The system then outputs the corresponding matching result data, including the matching standard template number, matching score, and error distribution during the matching process. Based on the best matching standard posture selected in step S31, the three-dimensional spatial deviation between the current posture node and the standard node is calculated for each key joint node of the human skeleton. Specifically, the system extracts the skeletal node coordinates of the current and standard postures, calculates the difference along the X, Y, and Z coordinate axes, and obtains the three-dimensional joint deviation vector. The magnitude of each joint deviation vector is then calculated to form a joint deviation index. The deviation indices of all joints are arranged in skeletal order to generate a joint deviation index matrix. Each element in the matrix represents the spatial deviation of the corresponding joint. This matrix serves as key input for subsequent biomechanical analysis. For the 3D human point cloud data, the 3D space is first divided into voxel units of equal volume. The number of point cloud points contained in each voxel unit is counted to generate a point cloud voxel density distribution map. Then, combined with the joint deviation index matrix from step S32, the joint deviation data is mapped to corresponding voxel regions based on the human skeletal structure and joint position. During the mapping process, a weighted distribution method is used to consider the influence range and decreasing trend of joint deviation on neighboring voxels, and the force contribution of each voxel is calculated. Next, based on the principles of force transfer and the human skeletal mechanical model, the load value within each voxel is calculated. The load value reflects the strength of the force in that area. Finally, the load value of each voxel is presented as a color gradient corresponding to the spatial position to generate a biomechanical load distribution map. The joint deviation index matrix and dynamic posture data are used to perform a detailed delineation of the range of motion of each key joint. By analyzing the angle change data in dynamic posture sequences, the joint motion amplitude and frequency within each angle interval are counted to obtain the motion amplitude distribution. Interpolation methods are used to continuously reconstruct the motion amplitude within each angle interval, forming a complete motion amplitude change curve. Subsequently, combined with the joint deviation index, the degree of joint motion restriction within each angle interval is assessed. Finally, this data is displayed as a heat map. The heat map uses color depth to distinguish the motion amplitude and degree of restriction of different joints, forming a joint motion heat map that intuitively reflects the functional status of the joints.

[0147] Preferably, step S31 includes the following steps:

[0148] Step S311: When the posture calibration parameters meet any of the following conditions, it is determined that the preliminary matching has failed and preliminary posture matching failure data is generated: the displacement error of the posture key node exceeds the set threshold of 10cm; the joint angle deviation exceeds 15°; the overall posture stability score is lower than 0.6;

[0149] Step S312: When the following conditions are met simultaneously, it is determined that there is a characteristic region posture mismatch and characteristic posture mismatch data is obtained: the angular coordination between the core joint groups decreases by more than 25%; the motion trajectory deviation exceeds the standard range of the historical health model; the matching residual is greater than the set upper limit of 0.08 in three consecutive evaluations, where the core joint group includes the shoulder, spine, and pelvis;

[0150] Step S313: When all of the following conditions occur, the overall posture is determined to be abnormal and posture abnormality data is obtained: the overall 3D posture alignment error exceeds 12%; the body symmetry score is lower than 0.7; the motion sequence continuity score is lower than 0.65; and the error cannot be corrected by the adaptive compensation algorithm during the matching process;

[0151] Step S314: Integrate the preliminary posture matching failure data, the characteristic posture mismatch data, and the posture abnormality data to generate the analysis results of the standard posture matching.

[0152] In this embodiment of the present invention, the input spatial posture calibration parameters are tested to determine whether the displacement error of each key skeletal node exceeds a preset threshold of 10 cm. The angular deviation of each joint is also calculated to determine whether it exceeds the 15-degree limit. Furthermore, the system calculates an overall posture stability score based on the multi-dimensional posture data. If the score is below 0.6, the initial matching is considered a failure. At this point, the system generates and stores preliminary posture matching failure data containing the cause of the failure and the relevant posture data. If the system detects that the angular coordination between the core joint group (including the shoulder, spine, and pelvis) has decreased by more than 25% compared to the historical healthy model, and the core joint motion trajectory has significantly deviated, exceeding the allowable range of the standard healthy model, and if the matching residual value remains greater than the set upper limit of 0.08 in three consecutive matching evaluations, the system determines that a characteristic region posture mismatch has occurred. The system records this status and generates characteristic posture mismatch data containing the aforementioned abnormal indicators and corresponding posture data for subsequent posture correction of specific regions. The system comprehensively judges the overall posture quality. When the alignment error of the overall three-dimensional posture exceeds 12%, the body symmetry score is lower than 0.7, and the continuity score of the motion sequence is lower than 0.65, and the system fails to correct the error through the adaptive compensation algorithm, it is judged that the overall posture is abnormal. At this time, the system will generate posture abnormality data that records the overall posture abnormality in detail, including the specific error value, score and specific information of compensation failure, for further in-depth analysis. The preliminary posture matching failure data, feature posture mismatch data and overall posture abnormality data generated in steps S311, S312 and S313 are integrated and processed. Through data fusion and comprehensive analysis, a standard posture matching analysis result report is formed. The report lists in detail the various judgment bases, abnormality types and related parameters, providing an accurate basis for subsequent posture evaluation and correction.

[0153] As an example of the present invention, refer to Figure 3 As shown, in this example, step S4 includes:

[0154] Step S41: reconstructing the depth profile of the multi-angle projection image group of the human body using the coarse registration skeleton model to construct a 3D human posture image model; mapping the biomechanical load distribution map to the surface of the 3D human posture image model to generate load stress distribution modeling data;

[0155] Step S42: Projecting the joint range of motion heat map onto the skeleton structure of the 3D human posture image model to generate a dynamic range of motion parameter set; performing a multi-dimensional assessment of posture health based on the load stress distribution modeling data and the dynamic range of motion parameter set to generate a posture abnormality marking layer;

[0156] Step S43: Identifying posture optimization areas based on the posture abnormality marker layer, matching personalized correction strategies with a preset standard posture model, and generating posture correction control parameters; applying the posture correction control parameters to key posture nodes of the 3D human posture image model to generate posture correction simulation results;

[0157] Step S44: Perform personalized interpretation and generate an evaluation summary of the posture correction simulation results, and finally output a personalized posture correction report.

[0158] In this embodiment of the present invention, a previously constructed coarsely registered skeleton model is used as a foundational framework to perform depth contour reconstruction on a set of multi-angle projection images of the human body. This process combines the contour information from the multi-angle projection images with the spatial constraints of the skeleton model to accurately reconstruct the three-dimensional surface morphology of the human body, thereby constructing a complete 3D human posture image model. Subsequently, biomechanical load distribution data is projected onto the surface of the 3D model using spatial mapping technology. Combined with the model's geometric features, the load stress data is converted into surface stress distribution modeling data, forming a 3D posture model incorporating mechanical information. Thermal distribution information from the joint range of motion thermogram is projected onto the skeleton structure of the 3D human posture image model to generate a dynamic range of motion parameter set. This parameter set specifically represents the range of motion and restricted range of each key joint in three-dimensional space. Next, the system combines the load stress distribution modeling data with the dynamic range of motion parameter set, employing a multi-dimensional assessment algorithm to comprehensively analyze the health status of human posture, identify abnormal areas and potential risks in posture, and generate a posture anomaly marker layer to provide visual assistance for subsequent posture optimization. Based on the generated posture anomaly marker layer, the system automatically identifies posture areas requiring optimization. By comparing and matching with the preset standard posture model and taking into account the individual differences of users, a personalized posture correction strategy is formulated. The system calculates and generates posture correction control parameters. Specific parameters include the adjustment direction and amplitude of key posture nodes. Subsequently, these control parameters are applied to the key posture nodes of the 3D human posture image model to simulate the posture effect after correction and generate posture correction simulation results to demonstrate and verify the feasibility of the correction plan. The posture correction simulation results are analyzed in detail, and personalized explanatory text and evaluation summary are generated based on the individual's initial posture data and health assessment results. The report covers the correction effect description, key adjustment areas, potential health improvement suggestions, etc., forming a final personalized posture correction report for reference and application by users and relevant health management personnel.

[0159] In this specification, a health assessment system based on a 3D human posture image model is provided, which is used to perform the above-mentioned health assessment method based on a 3D human posture image model. The health assessment system based on a 3D human posture image model includes:

[0160] The point cloud projection module is used to obtain human annular scan data; extract the dynamic posture sequence of the human annular scan data in upright position, forward bending position and side bending position, and perform phase shift fringe projection on the human annular scan data according to the dynamic posture sequence to obtain three-dimensional point cloud data of the human body;

[0161] The posture analysis module is used to analyze the sagittal, coronal, and transverse planes of the human body's three-dimensional point cloud data for orthogonal projection, generating a set of multi-angle projection images of the human body; identifying the skeletal contour features of the multi-angle projection image set of the human body to establish a dynamic marking system for anatomical landmarks; and using the dynamic marking system to perform spatial posture calibration on the skeletal contour features to generate spatial posture calibration parameters.

[0162] The posture deviation calculation module is used to input the spatial posture calibration parameters into the preset health posture assessment database for multi-dimensional comparative analysis. Based on the analysis results, it calculates the deviation index between the parts in the human body 3D point cloud data and the standard value to construct a biomechanical load distribution map and joint range of motion thermal map;

[0163] The health assessment module is used to construct a 3D human posture image model based on a group of multi-angle projection images of the human body; the 3D human posture image model is evaluated for posture health through biomechanical load distribution diagrams and joint range of motion thermal maps, and personalized posture correction is performed based on the evaluation results, thereby generating a personalized posture correction report.

[0164] The beneficial effect of the present invention is that it accurately captures the dynamic posture sequence of the human body in upright, flexed and lateral positions through circular scanning combined with phase-shifted fringe projection technology, effectively improving the integrity and dynamic capture capability of the three-dimensional point cloud data, and providing a solid data foundation for subsequent posture analysis. By performing orthogonal projection of the three-dimensional point cloud data in the sagittal, coronal and transverse planes, a multi-angle projection image group is generated. Combined with the bone contour feature recognition and the dynamic marking system of anatomical landmarks, accurate spatial calibration of the human skeleton posture is achieved, effectively improving the spatial accuracy and dynamic consistency of posture analysis. The spatial posture calibration parameters are matched and compared with the healthy posture database in multiple dimensions to accurately calculate the deviation index of each part of the skeleton, and a biomechanical load distribution map and joint range of motion heat map are constructed, thereby enhancing the quantitative assessment capability of posture abnormalities and the understanding of structural mechanics. A three-dimensional human posture model is constructed based on the multi-angle projection image group, integrating load distribution and range of motion information to achieve comprehensive health assessment and precise correction of individual posture, and generate a personalized posture correction report. These modules work together to form a complete closed loop, from human body dynamic data collection and spatial posture analysis to health assessment and personalized intervention. This significantly enhances the automation, intelligence, and practical value of human posture health management. Therefore, through high-precision dynamic data collection and multidimensional spatial analysis, this invention achieves precise, standardized, and personalized 3D human posture health assessment, improving its accuracy and comprehensiveness.

[0165] The present invention is therefore intended to be illustrative and non-restrictive in all respects, with the scope of the invention being defined by the appended claims rather than the foregoing description, and all changes that come within the meaning and range of equivalents of the application documents are intended to be embraced therein.

[0166] The foregoing description is intended only to provide specific embodiments of the present invention, which will enable those skilled in the art to understand and implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is to be construed in the widest possible manner consistent with the principles and novel features disclosed herein.

Claims

1. A health assessment method based on a 3D human posture image model, characterized in that: The following steps are involved: Step S1: acquiring human body annular scan data; extracting dynamic posture sequences of the human body annular scan data in upright, forward bending, and side bending positions, and performing phase-shifted fringe projection on the human body annular scan data according to the dynamic posture sequences to obtain three-dimensional point cloud data of the human body; Step S2: Analyze the sagittal plane, coronal plane and transverse section of the three-dimensional point cloud data of the human body and perform orthogonal projection to generate a multi-angle projection image group of the human body; Identify the skeletal contour features of a multi-angle projection image group of the human body to establish a dynamic marking system for anatomical landmarks; use the dynamic marking system to perform spatial posture calibration on the skeletal contour features and generate spatial posture calibration parameters; Step S3: Inputting the spatial posture calibration parameters into a preset health posture assessment database for multi-dimensional comparative analysis, and calculating the deviation index between the parts of the human body 3D point cloud data and the standard value based on the analysis results to construct a biomechanical load distribution map and a joint range of motion thermal map; Step S4: constructing a 3D human posture image model based on the multi-angle projection image group of the human body; performing posture health assessment on the 3D human posture image model using a biomechanical load distribution diagram and a joint range of motion thermal map, and performing personalized posture correction based on the assessment results, thereby generating a personalized posture correction report; wherein, step S4 includes the following steps: Step S41: reconstructing the depth profile of the multi-angle projection image group of the human body using the coarse registration skeleton model to construct a 3D human posture image model; mapping the biomechanical load distribution map to the surface of the 3D human posture image model to generate load stress distribution modeling data; Step S42: Projecting the joint range of motion heat map onto the skeleton structure of the 3D human posture image model to generate a dynamic range of motion parameter set; performing a multi-dimensional assessment of posture health based on the load stress distribution modeling data and the dynamic range of motion parameter set to generate a posture abnormality marking layer; Step S43: Identifying posture optimization areas based on the posture abnormality marker layer, matching personalized correction strategies with a preset standard posture model, and generating posture correction control parameters; applying the posture correction control parameters to key posture nodes of the 3D human posture image model to generate posture correction simulation results; Step S44: Perform personalized interpretation and generate an evaluation summary of the posture correction simulation results, and finally output a personalized posture correction report.

2. The health assessment method based on 3D human posture image model according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: performing a 360° circular scan of the human body using a scanning device to obtain human body circular scan data; Step S12: extracting a scanning sequence of the human body annular scanning data, and performing inter-frame synchronous registration on the human body annular scanning data according to the scanning sequence to generate synchronously corrected human body scanning frame data; Step S13: extracting dynamic skeleton points from the human body scan frame data; annotating the dynamic skeleton points with postures to generate dynamic posture sequences of upright position, forward bending position, and side bending position; Step S15: performing spatial posture parameter decomposition on the dynamic posture sequence to obtain human posture distribution control data; performing fringe phase encoding and multi-phase offset fringe pattern generation on the human body scanning frame data to obtain a multi-phase offset encoding atlas; Step S16: performing phase unpacking on the multi-phase offset coding atlas using the human body posture distribution control data to generate a dense phase map; performing triangulation and depth inversion calculation on the dense phase map to generate high-precision three-dimensional point cloud data of the human body.

3. The health assessment method based on 3D human posture image model according to claim 2, characterized in that: The step S13 of annotating the dynamic skeleton points includes: The three-dimensional coordinates of the dynamic skeleton points are collected by an optical motion capture system to obtain the three-dimensional coordinates of the skeleton points, wherein the sampling frequency is set to ≥30 frames / second, the spatial accuracy is ±2mm, the coordinate unit is millimeter, and the right-hand coordinate system is referenced; Definition of upright position identification criteria: overall spinal inclination ≤5°, left and right shoulder height difference ≤10mm, knee flexion angle ≤10°; Definition of flexion position identification criteria: overall spinal anteversion angle ≥30° and ≤90°, cervical to lumbar curvature radius ≤400 mm, hip flexion angle ≥45°; Scoliosis identification criteria were defined as follows: lateral spinal curvature angle ≥20° and ≤60°, left and right hip height difference ≥15mm, and the horizontal distance of the shoulder-hip line deviating from the central axis ≥30mm; Based on the defined upright position recognition standards, forward bending position recognition standards and lateral bending position recognition standards, the three-dimensional coordinates of the skeleton points are classified and labeled, thereby obtaining the dynamic posture sequences of upright position, forward bending position and lateral bending position.

4. The health assessment method based on 3D human posture image model according to claim 1, characterized in that: In step S2, the analysis of the sagittal plane, coronal plane, and transverse section of the three-dimensional point cloud data of the human body for orthogonal projection includes: Normalize the spatial coordinate center of gravity of the human body's three-dimensional point cloud data to generate standardized three-dimensional posture point cloud data; Extract the local principal axis of the standardized 3D pose point cloud data, and use the local principal axis to calibrate the body axis vector field of the 3D pose point cloud data to generate structure alignment vector calibration data; The structural alignment vector calibration data is rearranged in depth mapping according to the sagittal plane direction to generate a sagittal orthogonal projection image; The structural alignment vector calibration data is aggregated and compressed in the coronal direction to generate a coronal orthogonal projection image. The structural alignment vector calibration data is sliced ​​and resampled in the cross-sectional direction to generate a cross-sectional orthogonal projection image; The orthogonal projection images of the sagittal plane, coronal plane and transverse section are registered and bound with angle labels to generate a multi-angle projection image group of the human body.

5. The health assessment method based on 3D human posture image model according to claim 1, characterized in that: In step S2, identifying the skeleton contour features of the multi-angle projection image group of the human body to establish a dynamic marking system for anatomical landmarks includes: Identify the main skeleton contours of a multi-angle projection image group of the human body; Analyze the connected domain of the multi-angle projection image group of the human body according to the main skeleton contour to obtain the primary skeleton distribution map; Detect key morphological turning points on the primary skeleton distribution map to generate candidate anatomical structure corner point data, and perform spatial consistency test on the candidate anatomical structure corner point data based on depth prior to generate a high-confidence anatomical landmark point set; Perform temporal pose dynamic matching and trajectory interpolation on high-confidence anatomical landmark points to generate dynamic annotation sequence data; By dynamically annotating sequence data, the posture angle change modeling and structural stability evaluation of high-confidence anatomical landmark points are performed to generate a dynamic labeling system for anatomical landmark points.

6. The health assessment method based on 3D human posture image model according to claim 1, characterized in that: In step S2, the spatial posture calibration of the skeleton contour features using the dynamic marking system includes: Perform spatial center-of-gravity registration on the anatomical landmark sequence in the dynamic marking system to generate initial posture alignment reference data; Perform three-dimensional rigid registration on the bone contour features, and use the initial posture alignment reference data to perform coarse registration on the rigidly registered bone contour features to generate a skeleton model; Perform non-rigid fine-tuning optimization based on landmark points on the coarse registration skeleton model to generate a high-precision pose alignment model; Extract joint angles and solve the posture transformation matrix for the high-precision posture alignment model to generate a local posture transformation parameter set; The local attitude transformation parameter set is globally reconciled and error convergence controlled to generate spatial attitude calibration parameters.

7. The health assessment method based on 3D human posture image model according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: inputting the spatial posture calibration parameters into a preset health posture assessment database for posture similarity matching to generate a standard posture matching analysis result; Step S32: Calculate the three-dimensional deviation vector of each skeleton joint node based on the standard posture matching result data to generate a joint deviation index matrix; Step S33: analyzing the point cloud voxel density of the human body three-dimensional point cloud data, and performing mechanical structure mapping in combination with the joint deviation index matrix to generate a biomechanical load distribution map; Step S34: Perform angle interval analysis and motion amplitude reconstruction based on the joint deviation index matrix to generate a joint range of motion heat map.

8. The health assessment method based on 3D human posture image model according to claim 7, characterized in that: Step S31 includes the following steps: Step S311: When the posture calibration parameters meet any of the following conditions, it is determined that the preliminary matching has failed and preliminary posture matching failure data is generated: the displacement error of the posture key node exceeds the set threshold of 10cm; the joint angle deviation exceeds 15°; the overall posture stability score is lower than 0.6; Step S312: When the following conditions are met simultaneously, it is determined that there is a characteristic region posture mismatch and characteristic posture mismatch data is obtained: the angular coordination between the core joint groups decreases by more than 25%; the motion trajectory deviation exceeds the standard range of the historical health model; the matching residual is greater than the set upper limit of 0.08 in three consecutive evaluations, where the core joint group includes the shoulder, spine, and pelvis; Step S313: When all of the following conditions occur, the overall posture is determined to be abnormal and posture abnormality data is obtained: the overall 3D posture alignment error exceeds 12%; the body symmetry score is lower than 0.7; the motion sequence continuity score is lower than 0.65; and the error cannot be corrected by the adaptive compensation algorithm during the matching process; Step S314: Integrate the preliminary posture matching failure data, the characteristic posture mismatch data, and the posture abnormality data to generate the analysis results of the standard posture matching.

9. A health assessment system based on a 3D human posture image model, characterized in that: For executing the health assessment method based on the 3D human posture image model according to claim 1, the health assessment system based on the 3D human posture image model comprises: The point cloud projection module is used to obtain human annular scan data; extract the dynamic posture sequence of the human annular scan data in upright position, forward bending position and side bending position, and perform phase shift fringe projection on the human annular scan data according to the dynamic posture sequence to obtain three-dimensional point cloud data of the human body; The posture analysis module is used to analyze the sagittal, coronal, and transverse planes of the human body's three-dimensional point cloud data for orthogonal projection, generating a set of multi-angle projection images of the human body; identifying the skeletal contour features of the multi-angle projection image set of the human body to establish a dynamic marking system for anatomical landmarks; and using the dynamic marking system to perform spatial posture calibration on the skeletal contour features to generate spatial posture calibration parameters. The posture deviation calculation module is used to input the spatial posture calibration parameters into the preset health posture assessment database for multi-dimensional comparative analysis. Based on the analysis results, it calculates the deviation index between the parts in the human body 3D point cloud data and the standard value to construct a biomechanical load distribution map and joint range of motion thermal map; The health assessment module is used to construct a 3D human posture image model based on a group of multi-angle projection images of the human body; the 3D human posture image model is evaluated for posture health through biomechanical load distribution diagrams and joint range of motion thermal maps, and personalized posture correction is performed based on the evaluation results, thereby generating a personalized posture correction report.

Images