Human body size data automatic calculation method, device, equipment, medium and program product

Abstract

This application relates to the field of image processing technology and provides a method, apparatus, device, medium, and program product for automatic calculation of human body size data. The method includes: constructing a three-dimensional skeletal node set of the human body based on a frontal and right-side photograph; inputting the three-dimensional skeletal node set into a parameterized human morphological model for key point constraint optimization fitting to obtain a three-dimensional body surface mesh model; and calculating circumference, length-type dimensions, and body shape parameters based on the three-dimensional body surface mesh model. This invention solves the problem of systematic underestimation of circumference-type dimensions caused by the lack of depth direction information in single-view photographs, fills the gap in existing photographic body measurement technology in the extraction of discrete body shape parameters, and enables the calculation results to be directly used for personalized customization production of clothing patterns.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to a method, apparatus, device, medium, and program product for automatically calculating human body size data. Background Technology

[0002] In the field of custom clothing, accurate acquisition of body measurements is a prerequisite for achieving precise customization. Traditional manual measurement methods rely on tailors holding a soft measuring tape to measure each part of the user's body one by one. This is not only time-consuming, but the measurement results are also affected by subjective factors such as the tailor's operating technique and the amount of force applied. Different tailors may have significant deviations in the measurement results for the same user, which cannot meet the dimensional accuracy requirements of haute couture clothing customization, nor can it support users to complete the measurement process online.

[0003] Existing technologies propose methods for estimating anthropometric dimensions based on single photographs. These methods extract the human silhouette or skeletal nodes from a single frontal photograph and combine this with height parameters to estimate various body dimensions. However, a single photograph only provides two-dimensional silhouette information of the human body on a single projection plane, failing to capture thickness distribution data in the depth direction. This results in systematic errors in circumference measurements such as chest, abdomen, and hip circumference due to the lack of front-to-back thickness components, making it difficult to meet the accuracy requirements of custom clothing. Furthermore, existing photograph-based anthropometric methods can only output partial linear dimensions, unable to structurally extract discrete body shape parameters such as shoulder, belly, leg, and hip shapes. These body shape parameters are essential inputs for clothing pattern design; their absence will prevent the pattern from fitting the user's specific body shape, significantly compromising the fit of customized clothing. Summary of the Invention

[0004] This invention provides a method, apparatus, device, medium, and program product for automatically calculating human body size data, in order to solve the problem of systematic underestimation of circumference-type dimensions caused by the lack of depth direction information in single-view photographs. This fills the gap in the extraction of discrete body shape parameters in existing photographic body measurement technology, enabling the calculation results to be directly used for personalized customization production of clothing patterns.

[0005] In a first aspect, embodiments of this application provide a method for automatically calculating human body dimensions, including: Construct a set of three-dimensional skeletal nodes for the human body based on front and right side photos; The three-dimensional skeleton node set is input into the parameterized human morphology model for key point constraint optimization and fitting to obtain a three-dimensional body surface mesh model. The girth, length, and body shape parameters are calculated based on the three-dimensional body surface mesh model.

[0006] Optionally, in a first implementation of the first aspect of the present invention, constructing a set of three-dimensional skeletal nodes of the human body based on a frontal photograph and a right-side photograph includes: The vertical pixel distance between the top of the head and the ankle in the frontal photo is calculated based on the user's height parameters to obtain the frontal scale factor. Based on the user's height parameters, the vertical pixel distance between the pixel coordinates of the top of the head and the pixel coordinates of the ankle joint in the right-side photo is converted to obtain the side scale factor. Based on the aforementioned frontal scale factor, the horizontal and vertical pixel coordinates of the first skeletal node in the frontal photograph are converted into the left-right and vertical components of the human body; based on the aforementioned side scale factor, the horizontal pixel coordinates of the second skeletal node in the right-side photograph are converted into the front-back components of the human body. The left-right direction component, the vertical direction component, and the front-back direction component of the human body are combined to obtain a three-dimensional skeleton node set.

[0007] Optionally, in a second implementation of the first aspect of the present invention, the step of converting the horizontal and vertical pixel coordinates of the first skeletal node in the frontal photograph into human body left-right and human body vertical components based on the frontal scale factor; and converting the horizontal pixel coordinates of the second skeletal node in the right side photograph into human body front-back components based on the side scale factor, includes: The frontal photo is input into the human keypoint detection model to detect skeletal nodes, and the horizontal and vertical pixel coordinates of the first skeletal node are obtained. The right side photo is input into the human keypoint detection model to detect skeletal nodes, and the horizontal and vertical pixel coordinates of the second skeletal node are obtained. Based on the aforementioned frontal scale factor, the horizontal pixel coordinates of the first skeletal node in the frontal photograph are converted to obtain the left-right direction component of the human body, and the vertical pixel coordinates of the first skeletal node in the frontal photograph are converted to obtain the vertical direction component. Based on the side scale factor, the horizontal pixel coordinates of the second skeletal node in the right side view photo are converted to obtain the front-back direction components of the human body.

[0008] Optionally, in a third implementation of the first aspect of the present invention, the step of inputting the three-dimensional skeleton node set into a parameterized human morphological model for key point constraint optimization fitting to obtain a three-dimensional body surface mesh model includes: A joint objective function is constructed based on the keypoint alignment loss term, shape regularization term, and pose regularization term. The three-dimensional skeleton node set is then input into the parameterized human morphology model for iterative optimization and solution to obtain the optimal shape parameters and optimal pose parameters. Based on the optimal shape parameters and the optimal posture parameters, the parametric human morphology model is subjected to surface deformation and soft tissue thickness compensation to obtain a three-dimensional body surface mesh model.

[0009] Optionally, in a fourth implementation of the first aspect of the present invention, the step of deforming the parametric human morphology model based on the optimal shape parameters and the optimal pose parameters and applying soft tissue thickness compensation to obtain a three-dimensional body surface mesh model includes: The optimal shape parameters and the optimal posture parameters are input into the shape hybrid deformation function and skin deformation function of the parameterized human morphology model to perform surface deformation and obtain the initial mesh model. The body mass index (BMI) is calculated based on the user's weight and height parameters. Soft tissue thickness compensation proportional to the overscalar amount of the BMI is then applied to the vertices of the abdominal and hip regions in the initial mesh model along the direction of the body surface normal vector, resulting in a three-dimensional body surface mesh model.

[0010] Optionally, in a fifth implementation of the first aspect of the present invention, the step of calculating the girth dimension, length class dimension, and body shape parameters based on the three-dimensional body surface mesh model includes: The cross-sectional height corresponding to each girth is determined based on the vertical component of each anatomical node in the three-dimensional skeleton node set. A horizontal cutting plane is constructed at each cross-sectional height of the three-dimensional body surface mesh model and the cross-sectional contour curve is extracted. The arc length integral is performed on each cross-sectional contour curve to obtain the girth dimension. Using the vertices of the body surface in the three-dimensional body surface mesh model as the start and end points, the geodesic distance of each body surface measurement path is calculated on the topology map of the three-dimensional body surface mesh model to obtain the length-class dimension; The surface geometric features of each key anatomical region in the three-dimensional body surface mesh model are extracted respectively, and discrete classification is performed based on the surface geometric features to obtain body shape parameters.

[0011] Secondly, embodiments of this application provide an automatic human body size data estimation device, comprising: The building module is used to construct a set of three-dimensional skeletal nodes of the human body based on front and right side photos; The fitting module is used to input the three-dimensional skeleton node set into the parameterized human morphology model for key point constraint optimization fitting to obtain a three-dimensional body surface mesh model. The calculation module is used to calculate the girth dimension, length class dimension and body shape parameters based on the three-dimensional body surface mesh model.

[0012] Thirdly, embodiments of this application provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the above-described automatic human body size data estimation method.

[0013] Fourthly, embodiments of this application provide a readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described automatic human body size data estimation method.

[0014] Fifthly, embodiments of this application provide a computer program product, which includes a computer program that, when executed by a processor, enables the implementation of the steps of the above-described automatic human body size data calculation method.

[0015] In one solution provided by the aforementioned automatic human body size calculation method, device, equipment, medium, and program products, a dual-view combination of a frontal and a right-side photograph is used. The left-right and vertical components of the human body are extracted from the frontal photograph, and the front-back components are extracted from the right-side photograph. The skeletal node coordinates from both views are then fused and combined in a three-dimensional coordinate system to form a three-dimensional skeletal node set. This solves the problem of systematic underestimation of circumference dimensions caused by the lack of depth information in single-view photographs. This invention uses the three-dimensional skeletal node set as a constraint to drive a parametric human morphological model for keypoint constraint optimization and fitting, generating a three-dimensional body surface mesh model covering the user's entire body surface. Compared to the two-dimensional contour fitting method using a single photograph, this three-dimensional body surface mesh model completely restores the three-dimensional morphological details of the human body, such as the forward protrusion of the chest, the protrusion of the abdomen, and the backward protrusion of the buttocks, providing a unified and reliable geometric data source for subsequent calculations of various dimensions. Based on the aforementioned three-dimensional body surface mesh model, circumference-type dimensions are directly calculated by integrating the arc length of the horizontal cross-section contour, and length-type dimensions are calculated by geodesic distances between anatomical landmark nodes. Both calculations directly affect the actual geometric shape of the three-dimensional body surface mesh, avoiding the systematic errors introduced by relying on empirical formulas for indirect estimation in existing technologies. Simultaneously, this invention extracts and discretizes geometric features such as acromion height difference, anteroposterior protrusion, and lateral knee offset from key anatomical regions in the three-dimensional body surface mesh model, achieving structured and automatic output of body shape parameters such as shoulder shape, belly shape, hip shape, leg shape, chest shape, and arm shape. This fills the gap in discrete body shape parameter extraction in existing photogrammetry technology, enabling the calculated results to be directly used for personalized custom production of clothing patterns. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of the structure of an automatic human body size data estimation system according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating an automatic human body size data estimation method according to an embodiment of the present invention; Figure 3 yes Figure 2 A schematic diagram of the implementation process of step S10; Figure 4 yes Figure 3 A schematic diagram of the implementation process of step S13; Figure 5 yes Figure 2 A schematic diagram of the implementation process of step S20; Figure 6 yes Figure 2 A schematic diagram of the implementation process of step S30; Figure 7 This is a schematic diagram of the structure of an automatic human body size data calculation device according to an embodiment of the present invention; Figure 8 This is a schematic diagram of the structure of a computer device according to an embodiment of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] It should be understood that, when used in this specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or collections thereof. It should also be understood that, as used in this specification and the appended claims, the term "and / or" refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0020] Furthermore, in the description of this invention and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0021] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of the invention include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0022] It should be understood that the sequence number of each step in the following embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0023] To illustrate the technical solution of the present invention, specific embodiments are described below.

[0024] To address the problems mentioned above in the background art, this application provides a method, apparatus, device, medium, and program product for automatically calculating human body dimensions. The automatic calculation method for human body dimensions provided by this invention can be applied to, for example... Figure 1 The automatic human body size calculation system shown includes a client and a server.

[0025] In one embodiment, such as Figure 2 As shown, an automatic calculation method for human body dimensions is provided, which can be applied to... Figure 1 Taking the automatic calculation system for human body dimensions as an example, the following steps are included: S10: Construct a set of three-dimensional skeletal nodes for the human body based on the front and right side photos; S20: Input the set of three-dimensional skeleton nodes into the parameterized human morphology model and perform key point constraint optimization fitting to obtain a three-dimensional body surface mesh model; S30: Calculate the girth, length, and body shape parameters based on the three-dimensional body surface mesh model.

[0026] In this embodiment, by combining a frontal and a right-side photograph, the left-right and vertical components of the human body are extracted from the frontal photograph, and the front-back components are extracted from the right-side photograph. The coordinates of the skeletal nodes from the two perspectives are then fused and combined in a three-dimensional coordinate system to form a three-dimensional skeletal node set. This solves the problem of systematic underestimation of girth-type dimensions caused by the lack of depth direction information in single-view photographs. This invention uses the three-dimensional skeletal node set as a constraint to drive the parameterized human morphological model for key point constraint optimization and fitting, generating a three-dimensional body surface mesh model covering the user's entire body surface. Compared to the two-dimensional contour fitting method of a single photograph, this three-dimensional body surface mesh model completely restores the three-dimensional morphological details of the human body, such as the forward protrusion of the chest, the protrusion of the abdomen, and the backward protrusion of the buttocks, providing a unified and reliable geometric data source for subsequent calculations of various dimensions. Based on the three-dimensional body surface mesh model, girth-type dimensions are directly calculated by integrating the arc length of the horizontal sectional contour, and length-type dimensions are calculated by the geodesic distance between anatomical landmark nodes. Both calculations directly affect the actual geometric shape of the three-dimensional body surface mesh, avoiding the systematic errors introduced by relying on empirical formulas for indirect estimation in existing technologies. Meanwhile, this invention extracts geometric features such as acromion height difference, anterior-posterior protrusion, and lateral offset of the knee joint from key anatomical regions in a three-dimensional body surface mesh model and performs discrete classification. This enables the structured and automatic output of body shape parameters such as shoulder shape, belly shape, hip shape, leg shape, chest shape, and arm shape, filling the gap in the extraction of discrete body shape parameters in existing photo-measurement technology. This allows the calculation results to be directly used for personalized customization production of clothing patterns.

[0027] In one embodiment, such as Figure 3 As shown, step S10 specifically includes the following steps: S11: Based on the user's height parameters, the vertical pixel distance between the pixel coordinates of the top of the head and the pixel coordinates of the ankle joint in the frontal photo is converted to obtain the frontal scale factor; S12: Based on the user's height parameters, the vertical pixel distance between the pixel coordinates of the top of the head and the pixel coordinates of the ankle joint in the right-side photo is converted to obtain the side scale factor. S13: Based on the frontal scale factor, convert the horizontal and vertical pixel coordinates of the first skeletal node in the frontal photo into the left-right and vertical components of the human body; based on the side scale factor, convert the horizontal pixel coordinates of the second skeletal node in the right side photo into the front-back components of the human body. S14: Combine the left-right direction components, the vertical direction components, and the front-back direction components of the human body to obtain a set of three-dimensional skeleton nodes.

[0028] In this embodiment, after the frontal and right-side photos are captured on the mobile device, the server performs sharpness enhancement, noise reduction, distortion correction, and local contrast enhancement on both photos to ensure stable pixel boundaries for key areas such as the top of the head, ankles, shoulders, waist, and hips during detection. In the frontal photo, the server detects the pixel coordinates of the highest point of the head and the midpoint of the line connecting the two ankles, calculates the pixel distance between them in the vertical direction of the image, and then performs a physical scale conversion based on the user-entered height parameters to obtain the frontal scale factor. In the right-side photo, the server uses the same method to detect the vertical pixel distance between the highest point of the head and the ankle reference point, and calculates the side scale factor based on the same user height parameters. The scale conversion relationship can be expressed as: , ,in, This indicates the frontal scale factor, expressed in centimeters per pixel. This indicates the side dimension factor, expressed in centimeters per pixel. This indicates the user's height parameter, in centimeters; This represents the vertical pixel distance between the highest point of the head and the ankle reference point in a frontal photograph, in pixels. This represents the vertical pixel distance between the highest point of the head and the ankle reference point in the right-side photo, expressed in pixels. When verifying the consistency of the frontal and side scale factors, the relative deviation between them can be controlled within 5%, and this relative deviation is used as the retake threshold. This retake threshold can be adjusted based on fluctuations in mobile shooting distance, human keypoint positioning errors, and allowable errors in circumference dimensions. When the relative deviation between the frontal and side scale factors exceeds this retake threshold, the server prompts the user to re-capture the corresponding photo, reducing the distortion in the 3D skeleton node set ratio caused by shooting distance differences being transmitted to the left-right and front-back directions of the human body.

[0029] The server inputs a frontal photo into the human keypoint detection model to obtain the horizontal and vertical pixel coordinates of the first bone node. It then inputs a right-side photo into the same model to obtain the horizontal and vertical pixel coordinates of the second bone node. Based on the obtained frontal scale factor, the horizontal pixel offset of the first bone node is converted into the left-right component of the human body, and the vertical pixel offset is converted into the vertical component of the human body. Finally, based on the side scale factor, the horizontal pixel offset of the second bone node is converted into the front-back component of the human body. This preserves the ability of the frontal photo to describe the width and height of the human body and introduces the ability of the right-side photo to supplement the thickness of the human body. After the conversion, the server uses the human standing reference point as the origin of the 3D coordinate system and establishes a unified human coordinate system between the frontal and right-side photos based on the highest point of the head, the ankle reference point, the human midline, the acromion, and the hip joint. Before combining the left-right, vertical, and front-back components of the human body, the server verifies the consistency of human body dimensions, posture deviation, and node height in the two photos, ensuring that the width and height information corresponding to the frontal photo and the thickness information corresponding to the right-side photo are within the same spatial reference relationship. For anatomical nodes that can be stably identified in both the frontal and right-side photos, the server determines the fusion coordinates based on node detection confidence, node height deviation, and the connection relationship between adjacent bones. For left-side nodes in the right-side photo that cannot be stably identified due to body occlusion, the server performs constrained completion based on the left-right symmetry node relationship, the torso midline constraint, and the length constraint of adjacent nodes, thereby forming a set of 3D skeletal nodes.

[0030] In one embodiment, such as Figure 4 As shown, step S13 specifically includes the following steps: S131: Input the frontal photo into the human key point detection model to detect skeletal nodes and obtain the horizontal and vertical pixel coordinates of the first skeletal node. Input the right side photo into the human key point detection model to detect skeletal nodes and obtain the horizontal and vertical pixel coordinates of the second skeletal node. S132: Based on the frontal scale factor, the horizontal pixel coordinates of the first bone node in the frontal photo are converted to obtain the left-right direction component of the human body, and the vertical pixel coordinates of the first bone node in the frontal photo are converted to obtain the vertical direction component. S133: Based on the side scale factor, the horizontal pixel coordinates of the second bone node in the right side photo are converted to obtain the front-back direction components of the human body.

[0031] In this embodiment, after receiving the preprocessed frontal and right-side photos, the server inputs the frontal photo into the human body key point detection model. The human body key point detection model jointly identifies the human body contour, limb connection relationships, and anatomical landmark positions, and outputs a first set of skeletal nodes. Each node in the first set of skeletal nodes contains horizontal pixel coordinates, vertical pixel coordinates, and detection confidence. The covered nodes may include clothing measurement-related locations such as the left acromion, right acromion, cervical vertebrae, sternal manubrium, left and right axillary points, midpoint of the lumbar vertebrae, left and right hip joints, left and right knee joints, left and right ankle joints, left and right elbow joints, and left and right wrist joints. At the same time, the server inputs the right-side photo into the same human body key point detection model to obtain a second set of skeletal nodes. The second set of skeletal nodes mainly provides the lateral projection information required in the front-back thickness direction of the human body. Among them, the right acromion, right axillary point, right hip joint, right knee joint, right ankle joint, right elbow joint, right wrist joint, and nodes near the midline of the torso have high visibility. To reduce the impact of false positives on 3D reconstruction, a detection confidence threshold can be set to 0.85. Nodes with a confidence level below this threshold will not be directly used as stable coordinates in the conversion. This detection confidence threshold can be adjusted based on the false positive rate, false negative rate, and node localization error of the human keypoint detection model in the validation samples. When a node's confidence level is lower than the detection confidence threshold, the server uses adjacent skeleton connection relationships, left-right symmetry constraints, or manual re-shooting prompts to reduce the impact of clothing wrinkles, occlusion, pose deviation, and background interference on centimeter-scale conversion.

[0032] The server uses the pixel coordinates of the human body's midline in the frontal photo as the zero-point reference for the left and right directions of the human body. Combined with the already obtained frontal scale factor, it converts the horizontal pixel offset of the first bone node into the left-right component of the human body. Simultaneously, it uses the vertical pixel coordinates of the ankle joint reference point as the zero-point reference for the human body's height direction, converting the vertical pixel offset of the first bone node into the vertical component of the human body. The conversion relationship can be expressed as follows: , ,in, Indicates the first The left-right component of each skeletal node in the human body, in centimeters; Indicates the first Each skeletal node is a vertical component of the human body calculated from a frontal photograph, in centimeters. Indicates the first in the frontal photo The horizontal pixel coordinates of the first bone node; Indicates the first in the frontal photo The vertical pixel coordinates of the first bone node; This represents the horizontal pixel coordinates of the human body's midline in a frontal photograph; This represents the vertical pixel coordinates of the ankle joint reference point in a frontal photograph. This represents the frontal scale factor, expressed in centimeters per pixel. Since the vertical direction of image coordinates increases downwards, while the vertical direction in human 3D coordinates uses upwards as positive, the vertical conversion is done by subtracting the node's vertical pixel coordinates from the ankle joint reference point coordinates.

[0033] The server uses the midpoint of the lumbar spine or the midline reference point of the torso in the right-side image as the zero-point reference for the anterior-posterior direction of the human body. Combined with the side scale factor, it converts the horizontal pixel offset of the second bone node into the anterior-posterior component of the human body. The conversion relationship can be expressed as follows: ,in, Indicates the first The anterior-posterior component of each skeletal node in the human body, in centimeters; Indicates the first one in the right-side photo The horizontal pixel coordinates of the second bone node; This represents the horizontal pixel coordinates of the zero point reference for the front-back direction of the human body in the right-side photo. This represents the side scale factor, expressed in centimeters per pixel. A frontal photograph provides key spatial information about the human body in both left-right and height directions, while a right-side photograph supplements this with front-back information that cannot be obtained from a single frontal photograph. The converted node data can be mapped according to the same anatomical number.

[0034] In one embodiment, such as Figure 5 As shown, step S20 specifically includes the following steps: S21: Construct a joint objective function based on the keypoint alignment loss term, shape regularization term, and pose regularization term, and input the 3D skeleton node set into the parameterized human morphology model for iterative optimization to obtain the optimal shape parameters and optimal pose parameters; S22: Based on the optimal shape parameters and optimal posture parameters, deform the surface of the parametric human morphology model and apply soft tissue thickness compensation to obtain a three-dimensional body surface mesh model.

[0035] In this embodiment, the set of three-dimensional skeletal nodes is used as the external geometric constraint of the parametric human morphology model, enabling the parametric human morphology model to gradually converge from a general prior human morphology to an individualized human morphology that matches the current user's skeletal structure. The server calls the joint prediction mapping relationship driven by the shape parameter vector and posture parameter vector in the parametric human morphology model to generate the three-dimensional predicted coordinates of each anatomical key node in the current iteration state, and compares the three-dimensional predicted coordinates with the already constructed set of three-dimensional skeletal nodes point by point. For trusted nodes with a detection confidence of 0.85, the server incorporates the node spatial deviation into the key point alignment loss term, so that the positions of the acromion, axillary point, midpoint of the lumbar spine, hip joint, knee joint, ankle joint, etc., are as close as possible to the real spatial position calculated from the dual-view photos. At the same time, in order to reduce the abnormal stretching of the body surface morphology caused by the constraint of only limited skeletal nodes, a shape regularization term is introduced to constrain the shape parameters from deviating from the prior distribution of human body shape, and a posture regularization term is introduced to constrain the posture parameters to be close to the upright body state. Before entering the joint objective function, the keypoint alignment loss term is normalized according to the user's height parameter or skeletal scale. The shape regularization term and pose regularization term are normalized according to the prior covariance of the parameters or the allowable range of parameter variation. After converting all three types of loss terms into dimensionless penalty terms, they are then weighted and summed to form the joint objective function. in, This represents the joint objective function value, used to evaluate the overall matching degree between the human body model corresponding to the current shape parameters and pose parameters and the set of 3D skeleton nodes; This represents a shape parameter vector, used to characterize the direction of body shape changes such as height ratio, limb thickness, and shoulder-to-hip ratio; This represents the attitude parameter vector, used to characterize the spatial rotational state of the main joints; This represents the keypoint alignment loss term, used to constrain the spatial deviation between the model's predicted key nodes and the reliable nodes in the 3D skeleton node set. This represents the shape regularization term, used to suppress anomalous shape parameters; This represents the posture regularization term, used to suppress abnormal posture deviations; , , These represent the weight coefficients of the three types of loss terms. For example, after normalization of each loss term, the keypoint alignment weight can be used as the main weight, while the shape regularization weight and pose regularization weight can be used as stable weights. These weights can be tuned based on the keypoint spatial error, shape parameter fluctuation range, and pose deflection amplitude in the validation samples. The keypoint alignment weight should be higher than the shape regularization weight and pose regularization weight to ensure the model preferentially fits the reliable nodes obtained from the dual-view photos, while suppressing abnormal body shape stretching and abnormal posture deflection through regularization. The server can use the Adam optimizer to perform iterative solutions, with an initial learning rate of 0.001 and an upper limit of 200 iterations. Early termination occurs when the relative change rate of the objective function between two adjacent iterations is less than 0.1%. The learning rate, upper limit of iterations, and early termination condition can be tuned based on the convergence curve in the validation samples, the decrease in keypoint spatial error, and the time required for a single calculation, so that the optimization process maintains the stability of keypoint fitting while reducing ineffective iterations. After the iteration is completed, the joint objective function reaches a stable convergence state, and the optimal shape parameter and optimal posture parameter are obtained. The optimal shape parameter reflects the user's body proportion and body shape distribution, and the optimal posture parameter reflects the user's standardized standing state when taking the picture.

[0036] The server inputs the optimal shape and posture parameters into the shape blending deformation function and skin deformation function of the parameterized human body morphology model, causing the model mesh vertices to deform according to the individualized body shape and joint posture, generating an initial 3D mesh model. After the initial 3D mesh model is formed, the server combines the user's height and weight parameters to determine whether soft tissue-rich areas such as the abdomen and buttocks need thickness compensation, and applies restricted displacement correction to the mesh vertices of the corresponding areas along the direction of the body surface normal vector, so that the 3D body surface mesh model can express three-dimensional morphological features such as chest protrusion, abdominal protrusion, buttock protrusion, and shoulder height difference.

[0037] In one embodiment, step S22 specifically includes the following steps: The optimal shape parameters and optimal posture parameters are input into the shape hybrid deformation function and skin deformation function of the parameterized human morphology model to perform surface deformation and obtain the initial mesh model. The body mass index (BMI) is calculated based on the user's weight and height parameters. Soft tissue thickness compensation proportional to the BMI overscalar is then applied to the vertices of the abdominal and hip regions in the initial mesh model along the direction of the body surface normal vector, resulting in a three-dimensional body surface mesh model.

[0038] In this embodiment, the server uses the optimal shape parameters and optimal posture parameters as the driving inputs for the parametric human body morphology model. The parametric human body morphology model first adjusts the overall body proportions, torso thickness, shoulder-to-hip width relationship, and limb thickness distribution based on the optimal shape parameters. Then, it adjusts the spatial postures of joints such as the shoulder, elbow, hip, knee, and ankle based on the optimal posture parameters. An individualized shape offset is applied to the surface vertices of the basic human body template using a shape blending deformation function. A skinning deformation function maps the traction relationship of each joint posture to the surrounding surface vertices onto the mesh surface, thereby generating an initial mesh model that matches the 3D skeletal node set. The initial mesh model can express the user's skeletal length, shoulder width, hip width, and limb spatial orientation. However, soft tissue-rich areas such as the abdomen and buttocks may still have insufficient surface thickness due to insufficient skeletal node constraints. Therefore, soft tissue thickness compensation is needed by combining the user's weight parameters, height parameters, and local lateral contour features from the right-side profile photo. The server determines the overall compensation intensity range based on body mass index (BMI), and determines the compensation weights for the abdominal and gluteal regions based on the protrusion of the anterior abdominal edge relative to the midline of the torso and the protrusion of the posterior hip edge relative to the hip reference position. This ensures that soft tissue thickness compensation is simultaneously constrained by the relationship between body weight and height, as well as the local morphology of the lateral image. BMI can be calculated according to the following relationship: in, Body mass index, expressed in kilograms per square meter; This indicates the user's weight parameter, in kilograms. The existing meaning of the user's height parameter is retained. After calculating the Body Mass Index (BMI), the server can use 24 as the threshold for determining whether the abdominal and hip regions require expansion compensation. This threshold distinguishes between the normal body mass range and the body type range that may require thickness compensation, and can be adjusted based on the body sample distribution of the target user group. When the BMI does not exceed the threshold, the abdominal and hip regions of the initial mesh model do not undergo expansion compensation, and only the body surface morphology obtained from shape parameter fitting is retained. When the BMI exceeds the threshold, the server selects the body surface mesh vertices of the abdominal and hip regions in the initial mesh model, reads the body surface normal vector of each selected vertex, and applies displacement compensation along the outer direction of the body surface normal vector, taking into account the local protrusion weight in the right-side photo. The soft tissue thickness compensation amount can be determined according to the following relationship: in, This indicates the amount of soft tissue thickness compensation, expressed in centimeters. 0.12 represents the body mass index (BMI), expressed in kilograms per square meter; 0.12 represents the compensation slope, expressed in centimeters per BMI unit; 24 represents the BMI threshold; and 3 represents the upper limit of soft tissue thickness compensation, expressed in centimeters. Through non-negative truncation, no outward expansion compensation is performed on the abdominal and hip regions when the BMI does not exceed the threshold. Through upper limit limiting, the compensation amount is still restricted within a preset range when the BMI is high, reducing abnormal outward expansion of the mesh surface. The coefficient 0.12 can be used as an example value for the compensation slope in the abdominal and hip regions, indicating that for every unit the BMI exceeds the threshold, the corresponding soft tissue thickness is corrected by approximately 0.12 centimeters. This compensation slope can be tuned by combining sample body measurement data, errors in abdominal and hip protrusion in lateral photographs, and deviations between the 3D body surface mesh model and manual body measurement results. When direct girth measurement data is lacking, the compensation slope is used as a limiting correction coefficient and, together with the compensation upper limit, constrains the mesh outward expansion amplitude. After updating the vertices of the abdomen and buttocks regions, the server performs a smooth transition on the mesh vertices near the boundary of the compensation region and constrains the difference in normal displacement between adjacent vertices to maintain a continuous curved surface shape at the junction of the compensation region and the junction of the chest and waist and the hip and leg. When the local curvature change after smoothing exceeds the preset curvature change threshold, the server reduces the compensation weight of the corresponding vertex and re-executes boundary smoothing to finally obtain the three-dimensional body surface mesh model.

[0039] In one embodiment, such as Figure 6 As shown, step S30 specifically includes the following steps: S31: Determine the cross-sectional height corresponding to each girth based on the vertical component of each anatomical node in the three-dimensional skeleton node set. Construct horizontal cutting planes at each cross-sectional height of the three-dimensional body surface mesh model and extract the cross-sectional contour curves. Perform arc length integration on each cross-sectional contour curve to obtain the girth dimensions. S32: Using the vertices of the body surface in the three-dimensional body surface mesh model as the starting and ending points, calculate the geodesic distance of each body surface measurement path on the topology map of the three-dimensional body surface mesh model to obtain the length-class dimension; S33: Extract the surface geometric features of each key anatomical region in the three-dimensional body surface mesh model, and perform discrete classification based on the surface geometric features to obtain body shape parameters.

[0040] In this embodiment, the server uses a 3D body surface mesh model as a unified geometric data source and maps the vertical components of anatomical nodes in the 3D skeleton node set to the cross-sectional heights corresponding to different body mass items, ensuring that circumference, length, and discrete body shape parameters all originate from the same set of 3D human body surfaces. When calculating circumference, the server determines the height of the horizontal cutting plane based on the anatomical positioning relationships of the chest, neck, waist, abdomen, buttocks, upper arm, thigh, knee, calf, and ankle. For example, the chest circumference cross-sectional height can be the average of the vertical components of the left and right axillary points; the neck circumference cross-sectional height can be the vertical component corresponding to the cervical vertebrae; the mid-waist circumference cross-sectional height can be the vertical component corresponding to the midpoint of the lumbar vertebrae; the abdominal circumference cross-sectional height can be the midpoint between the midpoint of the lumbar vertebrae and the right hip joint; and the hip circumference cross-sectional height can traverse multiple candidate horizontal cross-sections from the waist to the upper thigh, and the candidate range is limited by the positions of the hip joint, the posterior edge of the buttocks, and the root of the thigh. ; Calculate the perimeter of the cross-section in each candidate horizontal cross-section, and select the position with the larger perimeter of the cross-section as the hip circumference cross-section position from the candidate horizontal cross-sections that meet the anatomical position constraints of the buttocks; For items such as upper arm circumference, thigh circumference, calf circumference and trouser leg circumference, the cross-section can be set in combination with the joint position. For example, the upper arm circumference is located about 5 cm below the armpit point, the thigh circumference is located about 10 cm below the hip joint, the calf circumference is located about 10 cm below the knee joint, and the trouser leg circumference is located about 3 cm above the ankle joint. The above distances are the implementation values ​​used in clothing measurement to be close to the maximum circumference or the garment placement, which can reduce the interference of local contraction at the joint endpoints on the cross-section perimeter. After the cross-sectional height is determined, the server constructs a horizontal cutting plane orthogonal to the vertical axis in the 3D body surface mesh model, calculates the intersection segments between the horizontal cutting plane and the triangular mesh facets, and splices them together according to the spatial connectivity of adjacent intersection segments to form a closed or approximately closed cross-sectional contour curve. Then, the arc length is calculated along the cross-sectional contour curve to obtain the circumference dimensions such as bust, neckline, waist, abdomen, and hip. When a local mesh has minor breaks due to pose occlusion or image estimation errors, contour patching can be performed within a range where the distance between adjacent endpoints is less than a preset multiple of the average side length of the mesh, and the patched contour curve is smoothed. The preset multiple can be adjusted based on the average side length of the 3D body surface mesh model, the cross-sectional contour sampling density, and the allowable error of manual measurement. When the distance between adjacent endpoints exceeds the range corresponding to the preset multiple, the server does not perform automatic patching and triggers cross-sectional reconstruction or image verification.

[0041] After completing the circumference-type dimensional calculations, the server converts the 3D body surface mesh model into a weighted topological graph. Mesh vertices serve as graph nodes, and mesh edges serve as connecting edges. Edge weights are determined by the spatial distance between adjacent body surface vertices. Based on the garment measurement definition, the server selects the start and end points for different length-type dimensions. For example, the total shoulder width corresponds to the body surface path from the left acromion to the right acromion; the back garment length corresponds to the body surface path from the cervical spine point along the midline of the back to the garment's endpoint; the front garment length corresponds to the body surface path from the cervical spine point along the chest area to the hem's endpoint; and the left and right sleeve lengths correspond to the left and right shoulders, respectively. The acromion extends along the outer side of the arm to the wrist joint, while the trouser length and side crotch extend along the outer side of the leg to the ankle. In solving these paths, the server can use a shortest path algorithm to calculate the geodesic distance on the body surface and pre-define the path search area based on the clothing measurement items. For example, the shoulder width path is limited to the upper edge of the shoulder and back area, the sleeve length path to the outer side of the arm area, and the trouser length path to the outer side of the leg area. This ensures that length-related dimensions unfold along the body surface according to the measurement path, avoiding the shortest path crossing non-target areas such as the armpit, front of the torso, or inner side of the limbs. For items easily affected by posture, such as sleeve length, the ratio of sleeve length to user height can be set within a validation range of 0.33 to 0.38 for reasonableness assessment. This validation range can be adjusted based on historical measurement data of the target user group, clothing size range, and manual review results. When the calculation result exceeds the validation range, the server triggers acromion point, elbow joint point, and wrist joint point verification or geodesic path recalculation to reduce abnormal dimensions caused by arm posture deviation.

[0042] Geometric features are extracted from key anatomical regions such as the shoulders, chest, abdomen, buttocks, legs, and arms. For example, the acromion height difference can be used to determine whether the shoulders are flat, normal, or sloping; the forward protrusion of the chest and the chest cross-sectional shape can be used to determine chest shape; the maximum forward protrusion of the abdomen can be used to determine belly shape; the backward protrusion of the buttocks and the hip circumference cross-sectional shape can be used to determine hip shape; the lateral offset of the knee joint relative to the hip-ankle line can be used to determine leg shape; and the lateral curvature of the arm and the change in upper arm circumference can be used to determine arm shape. When the geometric features of the body surface approach the boundary of adjacent body type categories, or when the confidence score of the classification model output is lower than the preset classification confidence threshold, the server marks the corresponding body type parameters as pending review and updates the body type parameters based on the user's re-photograph results or the results of manual review. The server can combine rule threshold judgments with the output of the classification model. Rule thresholds are used to handle geometric differences with clear boundaries, while the classification model is used to fuse the combined relationships between multiple geometric features. The rule thresholds can be adjusted based on manually labeled body shape samples, historical measurement data, and clothing pattern adjustment results. For example, features such as the difference in acromion height, the amount of forward abdominal protrusion, the amount of backward hip protrusion, and the lateral offset of the knee joint relative to the hip-ankle line can all be set with corresponding threshold ranges. When the classification model output results are inconsistent with the rule threshold judgment results, the server can use the result with higher confidence as the output, or mark the corresponding body shape parameter as pending verification, ultimately forming structured body shape parameters such as chest shape, belly shape, shoulder shape, arm shape, leg shape, and hip shape.

[0043] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0044] In one embodiment, an automatic human body size calculation device is provided, which corresponds one-to-one with the automatic human body size calculation method described in the above embodiments. For example... Figure 7 As shown, the automatic human body size calculation device includes: Module 701 is used to construct a set of three-dimensional skeletal nodes of the human body based on the front and right side photos; The fitting module 702 is used to input the three-dimensional skeleton node set into the parameterized human morphology model and perform key point constraint optimization fitting to obtain a three-dimensional body surface mesh model. The calculation module 703 is used to calculate the girth dimension, length class dimension and body shape parameters based on the three-dimensional body surface mesh model.

[0045] It should be noted that the information interaction and execution process between the above-mentioned devices / units are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, and they will not be repeated here.

[0046] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0047] This application also provides a computer device, such as... Figure 8 As shown, the computer device includes: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor. When the processor executes the computer program, it implements the steps in any of the above method embodiments, or when the processor executes the computer program, it implements the functions of each module / unit in the above device embodiments.

[0048] For example, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete this application. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the computer device.

[0049] Those skilled in the art will understand that Figure 8 The computer device described is merely an example and does not constitute a limitation on the computer device. It may include more or fewer components than shown, or combine certain components, or different components. For example, the computer device may also include input / output devices, network access devices, buses, etc.

[0050] The aforementioned processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), or field-programmable gate arrays (FPGAs). Programmable Gate Array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0051] The memory can be an internal storage unit of the computer device, such as a hard drive or RAM. The memory can also be an external storage device of the computer device, such as a plug-in hard drive, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card. Furthermore, the memory can include both internal and external storage units of the computer device.

[0052] This application also provides a readable storage medium storing a computer program that, when executed by a processor, implements the steps described in the various method embodiments above.

[0053] This application provides a computer program product that, when run on an electronic device, enables the electronic device to perform the steps described in the various method embodiments above.

[0054] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of this application can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include at least: any entity or device capable of carrying computer program code to a photographing device / terminal device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electrical carrier signals or telecommunication signals.

[0055] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0056] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0057] In the embodiments provided in this application, it should be understood that the disclosed apparatus / devices and methods can be implemented in other ways. For example, the apparatus / device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0058] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0059] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for automatically calculating human body dimensions, characterized in that, include: Construct a set of three-dimensional skeletal nodes for the human body based on front and right side photos; The three-dimensional skeleton node set is input into the parameterized human morphology model for key point constraint optimization and fitting to obtain a three-dimensional body surface mesh model. The girth, length, and body shape parameters are calculated based on the three-dimensional body surface mesh model.

2. The method for automatically calculating human body dimensions as described in claim 1, characterized in that, The process of constructing a three-dimensional skeletal node set for the human body based on front and right side photos includes: The vertical pixel distance between the top of the head and the ankle in the frontal photo is calculated based on the user's height parameters to obtain the frontal scale factor. Based on the user's height parameters, the vertical pixel distance between the pixel coordinates of the top of the head and the pixel coordinates of the ankle joint in the right-side photo is converted to obtain the side scale factor. Based on the aforementioned frontal scale factor, the horizontal and vertical pixel coordinates of the first skeletal node in the frontal photograph are converted into the left-right and vertical components of the human body; based on the aforementioned side scale factor, the horizontal pixel coordinates of the second skeletal node in the right-side photograph are converted into the front-back components of the human body. The left-right direction component, the vertical direction component, and the front-back direction component of the human body are combined to obtain a three-dimensional skeleton node set.

3. The method for automatically calculating human body dimensions as described in claim 2, characterized in that, The horizontal and vertical pixel coordinates of the first skeletal node in the frontal photograph are converted into human body left-right and human body vertical components based on the frontal scale factor. Based on the aforementioned side scale factor, the horizontal pixel coordinates of the second skeletal node in the right-side image are converted into human anterior-posterior direction components, including: The frontal photo is input into the human keypoint detection model to detect skeletal nodes, and the horizontal and vertical pixel coordinates of the first skeletal node are obtained. The right side photo is input into the human keypoint detection model to detect skeletal nodes, and the horizontal and vertical pixel coordinates of the second skeletal node are obtained. Based on the aforementioned frontal scale factor, the horizontal pixel coordinates of the first skeletal node in the frontal photograph are converted to obtain the left-right direction component of the human body, and the vertical pixel coordinates of the first skeletal node in the frontal photograph are converted to obtain the vertical direction component. Based on the side scale factor, the horizontal pixel coordinates of the second skeletal node in the right side image are converted to obtain the front-back direction components of the human body.

4. The method for automatically calculating human body dimensions as described in claim 3, characterized in that, The step of inputting the three-dimensional skeleton node set into a parameterized human morphology model for key point constraint optimization and fitting to obtain a three-dimensional body surface mesh model includes: A joint objective function is constructed based on the keypoint alignment loss term, shape regularization term, and pose regularization term. The three-dimensional skeleton node set is then input into the parameterized human morphology model for iterative optimization and solution to obtain the optimal shape parameters and optimal pose parameters. Based on the optimal shape parameters and the optimal posture parameters, the parametric human morphology model is subjected to surface deformation and soft tissue thickness compensation to obtain a three-dimensional body surface mesh model.

5. The method for automatically calculating human body dimensions as described in claim 4, characterized in that, The step of deforming the parametric human morphology model based on the optimal shape parameters and the optimal pose parameters and applying soft tissue thickness compensation to obtain a three-dimensional body surface mesh model includes: The optimal shape parameters and the optimal posture parameters are input into the shape hybrid deformation function and skin deformation function of the parameterized human morphology model to perform surface deformation and obtain the initial mesh model. The body mass index (BMI) is calculated based on the user's weight and height parameters. Soft tissue thickness compensation proportional to the overscalar amount of the BMI is then applied to the vertices of the abdominal and hip regions in the initial mesh model along the direction of the body surface normal vector, resulting in a three-dimensional body surface mesh model.

6. The method for automatically calculating human body dimensions as described in claim 1, characterized in that, The calculation of girth dimensions, length class dimensions, and body shape parameters based on the three-dimensional body surface mesh model includes: The cross-sectional height corresponding to each girth is determined based on the vertical component of each anatomical node in the three-dimensional skeleton node set. A horizontal cutting plane is constructed at each cross-sectional height of the three-dimensional body surface mesh model and the cross-sectional contour curve is extracted. The arc length integral is performed on each cross-sectional contour curve to obtain the girth dimension. Using the vertices of the body surface in the three-dimensional body surface mesh model as the start and end points, the geodesic distance of each body surface measurement path is calculated on the topology map of the three-dimensional body surface mesh model to obtain the length-class dimension; The surface geometric features of each key anatomical region in the three-dimensional body surface mesh model are extracted respectively, and discrete classification is performed based on the surface geometric features to obtain body shape parameters.

7. An automatic human body size calculation device, characterized in that, The steps for implementing the automatic calculation method for human body size data as described in any one of claims 1 to 6 include: The building module is used to construct a set of three-dimensional skeletal nodes of the human body based on front and right side photos; The fitting module is used to input the three-dimensional skeleton node set into the parameterized human morphology model for key point constraint optimization fitting to obtain a three-dimensional body surface mesh model. The calculation module is used to calculate the girth dimension, length class dimension and body shape parameters based on the three-dimensional body surface mesh model.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the automatic human body size data estimation method as described in any one of claims 1 to 6.

9. A readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the automatic calculation method for human body size data as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, enables the implementation of the steps of the automatic calculation method for human body size data as described in any one of claims 1 to 6.

Images