CN119229019B - A human body model measurement and classification method based on human body 3D reconstruction - Google Patents
A human body model measurement and classification method based on human body 3D reconstructionInfo
- Publication number
- CN119229019B CN119229019B CN202411378342.7A CN202411378342A CN119229019B CN 119229019 B CN119229019 B CN 119229019B CN 202411378342 A CN202411378342 A CN 202411378342A CN 119229019 B CN119229019 B CN 119229019B
- Authority
- CN
- China
- Prior art keywords
- model
- data
- segmentation
- algorithm
- namely
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Abstract
本发明提供一种基于人体三维重建的人体模型测量分型方法，S1.数据采集，S2.数据预处理，S3.重建模型S4.模型分割S5身体关键数据测量，使用配套的模型测量工具，测量身体关键数据；S6.体型自动分型，使用聚类算法对提取的特征进行聚类分析。本发明三维扫描技术能在极短的时间内获取大量点云数据，形成高度逼真的三维模型，其数据采集精度高，能够捕捉到物体的细微特征。这大大提高了数据获取的效率，同时保证了数据的准确性，为后续的体型分型提供了坚实的基础。
The present invention provides a human body model measurement and typing method based on three-dimensional human body reconstruction. The method includes steps S1: data acquisition, S2: data preprocessing, S3: model reconstruction, S4: model segmentation, and S5: key body data measurement. Key body data are measured using a matching model measurement tool. S6: automatic body typing is performed using a clustering algorithm to perform cluster analysis on the extracted features. The present invention's three-dimensional scanning technology can acquire a large amount of point cloud data in a very short time, forming a highly realistic three-dimensional model. Its data acquisition is highly accurate, capable of capturing subtle features of an object. This significantly improves data acquisition efficiency while ensuring data accuracy, providing a solid foundation for subsequent body typing.
Description
Technical Field
The invention relates to the technical field of human body model measurement, in particular to a human body model measurement typing method based on human body three-dimensional reconstruction.
Background
The general body type calculation methods mainly include two methods:
Bmi Index method (Body Mass Index):
This is the stature judgment index commonly used in the current medical community. The BMI is calculated as weight in kilograms divided by height in meters squared.
According to the world health organization standard, BMI values of 18.5-24.9 are normal weight, 25-29.9 are overweight, and 30 or more are obese. But different countries and regions may be adapted according to circumstances.
It should be noted that the BMI index, although simple and easy to use, does not fully accurately reflect the individual's body shape and health, especially for people with large muscle mass, the BMI value may be high.
2. Body index method (t=w/(h)):
this is another body type calculation method based on height and weight. Where t is the body index, w is the weight (kg), and h is the height (meters).
According to the value of the body index, the body weight can be judged to be of a low body weight when t is less than 18, normal body weight when t is less than or equal to 18 and less than or equal to 25, overweight body weight when t is less than or equal to 25 and less than or equal to 27, and obesity when t is more than or equal to 27.
The body index method takes into account the square relationship of height and weight, possibly more accurate in some cases, than the BMI index.
Disclosure of Invention
The invention aims at overcoming the defects of the prior art and provides a human body model measurement typing method for three-dimensional reconstruction of a human body.
Both the BMI index method and the body index method are simple calculations based on height and weight, and although the body type of a person can be primarily determined, the body type and health condition of the person cannot be completely and accurately reflected. Therefore, in medical evaluation, comprehensive analysis is also required in combination with other indexes (such as waistline, body fat rate, muscle mass, etc.). By further dividing the body types into H-type, a-type, T-type, X-type,
1) H type (straight cylinder type)
Is characterized in that the shoulders, the waist and the buttocks are almost on the same straight line, the difference between the waistline and the hip circumference is small, and the whole body takes on a straight line shape.
Identification criteria:
the waist-Hip Ratio (Waist-Hip Ratio, WHR) is close to 1 (typically 0.8-1.0).
The difference between the shoulder width and the hip width is not large, and the waistline is obviously not narrow.
The BMI index may be in the normal range, but the body fat rate may be high, especially visceral fat.
2) A type (Pear type)
Is characterized in that the buttocks and thigh areas are plump, the waist is relatively thin, and the upper half is narrow. Fat is distributed mainly in the lower body.
Identification criteria:
the waist-to-hip ratio (WHR) is less than 0.8.
The hip circumference is obviously larger than the shoulder circumference, and the waistline is thinner.
Body fat rates are high, especially in the lower body (e.g., thigh, buttock area).
More commonly, women are often evaluated for waist-to-hip Ratio and waist-to-Height Ratio (Waist-Height Ratio, WHtR).
3) T type (inverted triangle)
Is characterized in that the shoulder is wider, the chest is developed, the waist is narrower, and the buttocks are smaller. Fat is mainly distributed in the upper body.
Identification criteria:
the shoulder-to-Hip Ratio (Shoulder-Hip Ratio) is greater than 1.
The upper body (particularly the shoulders and chest) is significantly more girth than the lower body (buttocks and thighs).
More commonly for men, lower body fat rates and higher muscle mass are often accompanied, especially in the upper half.
4) X type (sandglass type)
Is characterized in that the shoulder and the hip are wider, the waist is obviously thinner, and the typical hourglass shape is shown.
Identification criteria:
The waist-to-hip ratio (WHR) is typically around 0.7.
The waistline is significantly smaller than the shoulder and hip circumference.
The body fat rate is medium or slightly high, and the fat distribution is uniform.
Commonly found in women, is generally determined by the ratio of waist and hip circumference.
These body type classifications can more accurately evaluate and judge the body type of a person, and have the following advantages:
1) The diagnosis accuracy is improved, and by means of clear body type classification, a doctor can judge the body type characteristics of a patient more accurately, which is helpful for diagnosing health problems related to body types more accurately. For example, type H statures may be more prone to certain metabolic diseases, while type a statures may be associated with increased risk of cardiovascular disease.
2) Personalized therapy-different body types may respond differently to drugs, methods of treatment and nutritional needs. Through the body type classification, doctors can make more personalized treatment schemes for patients, thereby improving the treatment effect and the satisfaction degree of the patients.
3) Health risk assessment-body type classification helps doctors to more accurately assess the health risk of patients. For example, type X stature is generally considered a desirable body shape, while type a stature may be associated with increased risk of cardiovascular disease. By knowing the body shape of the patient, doctors can more specifically conduct health risk assessment and preventive measure formulation.
4) Health education and prevention body type classification may be the basis for health education and prevention strategies. Doctors can provide personalized health advice and prevention advice for patients according to the characteristics of different body types, and help the patients establish healthier life style and habit.
5) Facilitating the study explicit body type classification helps to facilitate the progress of medical studies. By comparing health differences and physiological characteristics among different body types, researchers can more deeply understand physiological mechanisms of human bodies and occurrence mechanisms of diseases, and make a greater contribution to medical development.
The invention uses a three-dimensional scanning mode to collect human body shape data, reconstructs a human body model in three dimensions, and automatically calculates human body shapes (H type, A type, T type and X type) by using the highly restored three-dimensional model. The method solves the problem that the traditional body type calculation method can not fully reflect the body type of the person, and realizes the rapid typing of the body type of the person.
1. Data acquisition
Selecting a proper scanning device according to actual needs. For example, a laser scanner or a structured light scanner may be selected if surface scanning is desired, and a CT scan or MRI may be selected if in-vivo structures are desired.
Data acquisition, namely ensuring that a tested person is placed in the center of the scanning equipment relatively static so as to acquire accurate scanning data. In the acquisition process, the tested person can be scanned at different angles or positions according to the requirement, so that more comprehensive data can be obtained.
2. Data preprocessing
Denoising, namely removing noise in the scanned data by filtering and other methods so as to improve the accuracy of subsequent processing.
Registration if multiple scanning devices or scans are used, different data sets need to be registered to ensure that they are in the same coordinate system.
And (3) repairing the data, namely repairing the missing or damaged part in the data, wherein interpolation and other methods can be used for repairing the data.
3. Reconstruction model
And selecting a reconstruction algorithm, namely selecting an appropriate reconstruction algorithm according to the data type and the requirement. For example, for point cloud data, a surface reconstruction algorithm, such as Marching Cubes algorithm, may be used, and for volume data, a voxelized reconstruction algorithm may be used.
Model reconstruction, which is to convert the preprocessed data into a three-dimensional model by using a selected algorithm. This may involve mesh generation, surface fitting, voxelization, etc.
Point cloud data is data commonly used for three-dimensional scanning and sensing equipment generation, and usually does not contain an explicit connection relationship. In order to construct a three-dimensional model, the common flow of point cloud data reconstruction is as follows:
1. Mesh generation for point cloud data or volume data, the core of the reconstruction process is to generate a polygonal mesh or triangular patch.
Using Marching Cubes algorithm, the specific procedure is as follows:
1. The three-dimensional space is divided into a series of regular cubes (Cube) and the vertices of each Cube are labeled as "inside" or "outside" depending on where the data point is located.
2. Within each cube, it is determined by looking up the boundary table which vertices need to be connected.
3. Generating triangular patches and connecting the triangular patches to form a grid. These triangular patches will eventually constitute the surface of the entire three-dimensional model.
2. Surface fitting, namely using a poisson reconstruction algorithm for data needing a smooth surface. The basic process is as follows:
1. Gradient fields and normal vectors are calculated for the point cloud.
2. These gradient field information are used to generate a smooth surface by global fitting.
3. A final triangular patch is generated to represent the model surface.
4. Model segmentation (optional)
If calculation of surface area and volume is required for a specific region, model segmentation may be performed. This may be done by manual segmentation or automatic segmentation algorithms.
1) Manual segmentation
Manual segmentation is a segmentation method that relies on manual manipulation, typically performed by a person skilled in the art using three-dimensional modeling software (e.g., blender, meshLab, etc.).
The steps are as follows:
And importing the complete three-dimensional model into modeling software.
The "segmentation tool" in the software is used to manually select the region to be segmented (e.g., selecting the head, torso, limbs, etc.).
The selected regions are independent from the overall model as separate model parts.
Surface area and volume calculations are performed separately for each sub-region.
2) Automatic segmentation algorithm
The automatic segmentation algorithm automatically segments the model according to specific rules or conditions through a computer program, so that manual operation is reduced, and efficiency is improved. Common automatic segmentation algorithms include mesh-based segmentation, volume-based segmentation, morphology-based segmentation, and the like.
Based on a mesh segmentation algorithm (Mesh Segmentation):
Grid pretreatment, namely, after a model is imported, preprocessing the grid of the model, such as removing redundant points, smoothing the grid and the like.
And determining a cutting plane by automatically determining the cutting plane through curvature change, geometric characteristics and the like according to the topological structure of the model. For example, joints of the body (shoulder, elbow, knee) are often natural segmentation points.
Region segmentation, namely dividing the model into a plurality of subareas such as a head, a trunk, limbs and the like according to a cutting plane.
Data extraction, the volume and surface area of each sub-region are calculated.
Based on the volumetric segmentation algorithm (Voxel-based Segmentation):
voxelization-the model voxelized into three-dimensional data consisting of a series of regular small cubes (voxels).
Region growing algorithm, namely, by selecting seed points, expanding the seed points to adjacent voxels by using the region growing algorithm to form segmented regions. For example, region growth may begin at the shoulder or knee joint.
The segmentation is completed by segmenting out different body regions, and then surface area and volume measurements can be made for each region.
Based on morphology segmentation algorithm:
Edge detection, namely, edge detection is carried out through the geometric outline (such as the curvature of the surface) of the model, and key points of the model such as shoulders, waists, knees and the like are identified.
Morphological operations, namely processing the model by using morphological operations such as expansion, corrosion and the like, and clearly dividing each region.
And (3) dividing the region into parts such as a head part, a trunk part, four limbs and the like.
Surface area and volume calculations further surface area and volume calculations are performed for each region using a specific measurement algorithm.
5. Body critical data measurement, using a matched model measurement tool, to measure body critical data.
After model segmentation, key data of the body (such as surface area, volume, length, circumference, etc.) are measured using a matched measuring tool. Common measuring tools include:
surface area and volume measurements:
the surface area and volume measurements are made for each segmented region by either a built-in tool in the three-dimensional modeling software (e.g., blender's "measurement tool" plug-in) or by writing custom scripts.
The surface area and volume of the closed surface can be calculated using the gaussian-blogging formula or directly by integration.
Length and circumference measurements:
and selecting key points and line segments by using the geometric properties of the model, and calculating the lengths and the circumferences of different parts of the body. For example, the length from the shoulder to the wrist may be measured, or the waist circumference calculated.
Accurate length and circumference data is calculated by setting a fixed measurement path or using an algorithm to find the shortest path along the curved surface.
6. Automatic parting of body type
And performing cluster analysis on the extracted features by using a cluster algorithm. The clustering algorithm comprises K-means clustering, hierarchical clustering and density clustering. These algorithms may divide the data points into different clusters or categories based on their similarity between them. In the clustering process, appropriate clustering parameters (such as the number of clusters, a similarity measurement method and the like) need to be selected to obtain the optimal clustering effect. And evaluating the clustering result to determine the accuracy and reliability of the typing. In addition, the clustering results are presented by visualization techniques (e.g., scatter plots, thermodynamic diagrams, etc.) to more intuitively understand the differences between different body types.
1.H type (straight cylinder type)
Is characterized in that the shoulders, the waist and the buttocks are almost on the same straight line, the difference between the waistline and the hip circumference is small, and the whole body takes on a straight line shape.
Identification criteria:
the waist-Hip Ratio (Waist-Hip Ratio, WHR) is close to 1 (typically 0.8-1.0).
The difference between the shoulder width and the hip width is not large, and the waistline is obviously not narrow.
The BMI index may be in the normal range, but the body fat rate may be high, especially visceral fat.
Type A (Pear type)
Is characterized in that the buttocks and thigh areas are plump, the waist is relatively thin, and the upper half is narrow. Fat is distributed mainly in the lower body.
Identification criteria:
the waist-to-hip ratio (WHR) is less than 0.8.
The hip circumference is obviously larger than the shoulder circumference, and the waistline is thinner.
Body fat rates are high, especially in the lower body (e.g., thigh, buttock area).
More commonly, women are often evaluated for waist-to-hip Ratio and waist-to-Height Ratio (Waist-Height Ratio, WHtR).
T type (inverted triangle)
Is characterized in that the shoulder is wider, the chest is developed, the waist is narrower, and the buttocks are smaller. Fat is mainly distributed in the upper body.
Identification criteria:
the shoulder-to-Hip Ratio (Shoulder-Hip Ratio) is greater than 1.
The upper body (particularly the shoulders and chest) is significantly more girth than the lower body (buttocks and thighs).
More commonly for men, lower body fat rates and higher muscle mass are often accompanied, especially in the upper half.
X type (hourglass type)
Is characterized in that the shoulder and the hip are wider, the waist is obviously thinner, and the typical hourglass shape is shown.
Identification criteria:
The waist-to-hip ratio (WHR) is typically around 0.7.
The waistline is significantly smaller than the shoulder and hip circumference.
The body fat rate is medium or slightly high, and the fat distribution is uniform.
Commonly found in women, is generally determined by the ratio of waist and hip circumference.
7. Visual display
And the three-dimensional rendering technology is used for visually displaying the calculation result so as to more intuitively understand and share the result. Three-dimensional rendering software or libraries may be used to present the model surface area and volume distribution, generating visual reports or animations.
A computer readable medium storing software comprising instructions executable by one or more computers, the instructions, by such execution, causing the one or more computers to perform operations comprising the flow of the system described above.
A computer system comprising:
One or more processors;
a memory storing instructions that, when executed by the one or more processors, cause the one or more processors to perform operations comprising the flow of the system described above.
Compared with the prior art, the invention has the advantages that:
1. the three-dimensional scanning technology can acquire a large amount of point cloud data in extremely short time to form a highly realistic three-dimensional model, and the data acquisition precision is high, so that the fine characteristics of an object can be captured. The method greatly improves the efficiency of data acquisition, ensures the accuracy of data, and provides a solid foundation for the subsequent body type parting.
2. The non-contact measurement is that the three-dimensional scanning technology adopts a non-contact measurement mode, and physical contact with a target object is not needed, so that the measurement can be performed on the premise of not damaging or polluting the object. This avoids errors caused by human factors and improves the accuracy and reliability of the measurement.
3. The three-dimensional scanning technology can work under various environmental conditions, including indoor, outdoor, no-light environment and the like, so that the three-dimensional scanning technology has wide application in various fields. In human body type parting, the characteristic of strong adaptability enables the technology to cope with various complex environments and ensures smooth data acquisition.
4. The intuitiveness and achievement diversity are that the acquired point cloud data not only contains space information, but also has color information and reflectivity value, and can truly reproduce object scenes. In human body type parting, the abundant information can help us to more comprehensively understand the morphological structure of the human body, and improve the parting accuracy. Meanwhile, various achievements can be output by one-time measurement, repeated measurement is not needed, and the working efficiency is improved.
5. The automatic processing and analysis of the three-dimensional scanning data can be realized by combining advanced computer vision and machine learning algorithms, so that the body type typing of the human body can be automatically carried out. The highly-automated processing mode greatly reduces the need of manual intervention and improves the parting efficiency and accuracy.
6. Personalized service, namely, human body type typing based on three-dimensional scanning data can provide personalized service according to specific morphological structures of individuals. For example, in the clothing industry, appropriate clothing can be customized according to individuals of different sizes, and in the fitness industry, personalized fitness plans and suggestions can be provided for individuals of different sizes.
Drawings
FIG. 1 is a data acquisition interface;
FIG. 2 is a data reconstruction interface;
FIG. 3 is a model segmentation interface;
fig. 4 is a measurement typing interface.
Detailed Description
As shown in fig. 1 to 4, a human body model measurement typing method for three-dimensional reconstruction of a human body.
Both the BMI index method and the body index method are simple calculations based on height and weight, and although the body type of a person can be primarily determined, the body type and health condition of the person cannot be completely and accurately reflected. Therefore, in medical evaluation, comprehensive analysis is also required in combination with other indexes (such as waistline, body fat rate, muscle mass, etc.). By further dividing the body types into H-type, a-type, T-type, X-type,
1) H type (straight cylinder type)
Is characterized in that the shoulders, the waist and the buttocks are almost on the same straight line, the difference between the waistline and the hip circumference is small, and the whole body takes on a straight line shape.
Identification criteria:
the waist-Hip Ratio (Waist-Hip Ratio, WHR) is close to 1 (typically 0.8-1.0).
The difference between the shoulder width and the hip width is not large, and the waistline is obviously not narrow.
The BMI index may be in the normal range, but the body fat rate may be high, especially visceral fat.
2) A type (Pear type)
Is characterized in that the buttocks and thigh areas are plump, the waist is relatively thin, and the upper half is narrow. Fat is distributed mainly in the lower body.
Identification criteria:
the waist-to-hip ratio (WHR) is less than 0.8.
The hip circumference is obviously larger than the shoulder circumference, and the waistline is thinner.
Body fat rates are high, especially in the lower body (e.g., thigh, buttock area).
More commonly, women are often evaluated for waist-to-hip Ratio and waist-to-Height Ratio (Waist-Height Ratio, WHtR).
3) T type (inverted triangle)
Is characterized in that the shoulder is wider, the chest is developed, the waist is narrower, and the buttocks are smaller. Fat is mainly distributed in the upper body.
Identification criteria:
the shoulder-to-Hip Ratio (Shoulder-Hip Ratio) is greater than 1.
The upper body (particularly the shoulders and chest) is significantly more girth than the lower body (buttocks and thighs).
More commonly for men, lower body fat rates and higher muscle mass are often accompanied, especially in the upper half.
4) X type (sandglass type)
Is characterized in that the shoulder and the hip are wider, the waist is obviously thinner, and the typical hourglass shape is shown.
Identification criteria:
The waist-to-hip ratio (WHR) is typically around 0.7.
The waistline is significantly smaller than the shoulder and hip circumference.
The body fat rate is medium or slightly high, and the fat distribution is uniform.
Commonly found in women, is generally determined by the ratio of waist and hip circumference.
These body type classifications can more accurately evaluate and judge the body type of a person, and have the following advantages:
1) The diagnosis accuracy is improved, and by means of clear body type classification, a doctor can judge the body type characteristics of a patient more accurately, which is helpful for diagnosing health problems related to body types more accurately. For example, type H statures may be more prone to certain metabolic diseases, while type a statures may be associated with increased risk of cardiovascular disease.
2) Personalized therapy-different body types may respond differently to drugs, methods of treatment and nutritional needs. Through the body type classification, doctors can make more personalized treatment schemes for patients, thereby improving the treatment effect and the satisfaction degree of the patients.
3) Health risk assessment-body type classification helps doctors to more accurately assess the health risk of patients. For example, type X stature is generally considered a desirable body shape, while type a stature may be associated with increased risk of cardiovascular disease. By knowing the body shape of the patient, doctors can more specifically conduct health risk assessment and preventive measure formulation.
4) Health education and prevention body type classification may be the basis for health education and prevention strategies. Doctors can provide personalized health advice and prevention advice for patients according to the characteristics of different body types, and help the patients establish healthier life style and habit.
5) Facilitating the study explicit body type classification helps to facilitate the progress of medical studies. By comparing health differences and physiological characteristics among different body types, researchers can more deeply understand physiological mechanisms of human bodies and occurrence mechanisms of diseases, and make a greater contribution to medical development.
The invention uses a three-dimensional scanning mode to collect human body shape data, reconstructs a human body model in three dimensions, and automatically calculates human body shapes (H type, A type, T type and X type) by using the highly restored three-dimensional model. The method solves the problem that the traditional body type calculation method can not fully reflect the body type of the person, and realizes the rapid typing of the body type of the person.
1. Data acquisition
Selecting a proper scanning device according to actual needs. For example, a laser scanner or a structured light scanner may be selected if surface scanning is desired, and a CT scan or MRI may be selected if in-vivo structures are desired.
Data acquisition, namely ensuring that a tested person is placed in the center of the scanning equipment relatively static so as to acquire accurate scanning data. In the acquisition process, the tested person can be scanned at different angles or positions according to the requirement, so that more comprehensive data can be obtained.
2. Data preprocessing
Denoising, namely removing noise in the scanned data by filtering and other methods so as to improve the accuracy of subsequent processing.
Registration if multiple scanning devices or scans are used, different data sets need to be registered to ensure that they are in the same coordinate system.
And (3) repairing the data, namely repairing the missing or damaged part in the data, wherein interpolation and other methods can be used for repairing the data.
3. Reconstruction model
And selecting a reconstruction algorithm, namely selecting an appropriate reconstruction algorithm according to the data type and the requirement. For example, for point cloud data, a surface reconstruction algorithm, such as Marching Cubes algorithm, may be used, and for volume data, a voxelized reconstruction algorithm may be used.
Model reconstruction, which is to convert the preprocessed data into a three-dimensional model by using a selected algorithm. This may involve mesh generation, surface fitting, voxelization, etc.
Point cloud data is data commonly used for three-dimensional scanning and sensing equipment generation, and usually does not contain an explicit connection relationship. In order to construct a three-dimensional model, the common flow of point cloud data reconstruction is as follows:
1. Mesh generation for point cloud data or volume data, the core of the reconstruction process is to generate a polygonal mesh or triangular patch.
Using Marching Cubes algorithm, the specific procedure is as follows:
1. The three-dimensional space is divided into a series of regular cubes (Cube) and the vertices of each Cube are labeled as "inside" or "outside" depending on where the data point is located.
2. Within each cube, it is determined by looking up the boundary table which vertices need to be connected.
3. Generating triangular patches and connecting the triangular patches to form a grid. These triangular patches will eventually constitute the surface of the entire three-dimensional model.
2. Surface fitting, namely using a poisson reconstruction algorithm for data needing a smooth surface. The basic process is as follows:
1. Gradient fields and normal vectors are calculated for the point cloud.
2. These gradient field information are used to generate a smooth surface by global fitting.
3. A final triangular patch is generated to represent the model surface.
4. Model segmentation (optional)
If calculation of surface area and volume is required for a specific region, model segmentation may be performed. This may be done by manual segmentation or automatic segmentation algorithms.
1) Manual segmentation
Manual segmentation is a segmentation method that relies on manual manipulation, typically performed by a person skilled in the art using three-dimensional modeling software (e.g., blender, meshLab, etc.).
The steps are as follows:
And importing the complete three-dimensional model into modeling software.
The "segmentation tool" in the software is used to manually select the region to be segmented (e.g., selecting the head, torso, limbs, etc.).
The selected regions are independent from the overall model as separate model parts.
Surface area and volume calculations are performed separately for each sub-region.
2) Automatic segmentation algorithm
The automatic segmentation algorithm automatically segments the model according to specific rules or conditions through a computer program, so that manual operation is reduced, and efficiency is improved. Common automatic segmentation algorithms include mesh-based segmentation, volume-based segmentation, morphology-based segmentation, and the like.
Based on a mesh segmentation algorithm (Mesh Segmentation):
Grid pretreatment, namely, after a model is imported, preprocessing the grid of the model, such as removing redundant points, smoothing the grid and the like.
And determining a cutting plane by automatically determining the cutting plane through curvature change, geometric characteristics and the like according to the topological structure of the model. For example, joints of the body (shoulder, elbow, knee) are often natural segmentation points.
Region segmentation, namely dividing the model into a plurality of subareas such as a head, a trunk, limbs and the like according to a cutting plane.
Data extraction, the volume and surface area of each sub-region are calculated.
Based on the volumetric segmentation algorithm (Voxel-based Segmentation):
voxelization-the model voxelized into three-dimensional data consisting of a series of regular small cubes (voxels).
Region growing algorithm, namely, by selecting seed points, expanding the seed points to adjacent voxels by using the region growing algorithm to form segmented regions. For example, region growth may begin at the shoulder or knee joint.
The segmentation is completed by segmenting out different body regions, and then surface area and volume measurements can be made for each region.
Based on morphology segmentation algorithm:
Edge detection, namely, edge detection is carried out through the geometric outline (such as the curvature of the surface) of the model, and key points of the model such as shoulders, waists, knees and the like are identified.
Morphological operations, namely processing the model by using morphological operations such as expansion, corrosion and the like, and clearly dividing each region.
And (3) dividing the region into parts such as a head part, a trunk part, four limbs and the like.
Surface area and volume calculations further surface area and volume calculations are performed for each region using a specific measurement algorithm.
5. Body critical data measurement, using a matched model measurement tool, to measure body critical data.
After model segmentation, key data of the body (such as surface area, volume, length, circumference, etc.) are measured using a matched measuring tool. Common measuring tools include:
surface area and volume measurements:
the surface area and volume measurements are made for each segmented region by either a built-in tool in the three-dimensional modeling software (e.g., blender's "measurement tool" plug-in) or by writing custom scripts.
The surface area and volume of the closed surface can be calculated using the gaussian-blogging formula or directly by integration.
Length and circumference measurements:
and selecting key points and line segments by using the geometric properties of the model, and calculating the lengths and the circumferences of different parts of the body. For example, the length from the shoulder to the wrist may be measured, or the waist circumference calculated.
Accurate length and circumference data is calculated by setting a fixed measurement path or using an algorithm to find the shortest path along the curved surface.
6. Automatic parting of body type
And performing cluster analysis on the extracted features by using a cluster algorithm. The clustering algorithm comprises K-means clustering, hierarchical clustering and density clustering. These algorithms may divide the data points into different clusters or categories based on their similarity between them. In the clustering process, appropriate clustering parameters (such as the number of clusters, a similarity measurement method and the like) need to be selected to obtain the optimal clustering effect. And evaluating the clustering result to determine the accuracy and reliability of the typing. In addition, the clustering results are presented by visualization techniques (e.g., scatter plots, thermodynamic diagrams, etc.) to more intuitively understand the differences between different body types.
1.H type (straight cylinder type)
Is characterized in that the shoulders, the waist and the buttocks are almost on the same straight line, the difference between the waistline and the hip circumference is small, and the whole body takes on a straight line shape.
Identification criteria:
the waist-Hip Ratio (Waist-Hip Ratio, WHR) is close to 1 (typically 0.8-1.0).
The difference between the shoulder width and the hip width is not large, and the waistline is obviously not narrow.
The BMI index may be in the normal range, but the body fat rate may be high, especially visceral fat.
Type A (Pear type)
Is characterized in that the buttocks and thigh areas are plump, the waist is relatively thin, and the upper half is narrow. Fat is distributed mainly in the lower body.
Identification criteria:
the waist-to-hip ratio (WHR) is less than 0.8.
The hip circumference is obviously larger than the shoulder circumference, and the waistline is thinner.
Body fat rates are high, especially in the lower body (e.g., thigh, buttock area).
More commonly, women are often evaluated for waist-to-hip Ratio and waist-to-Height Ratio (Waist-Height Ratio, WHtR).
T type (inverted triangle)
Is characterized in that the shoulder is wider, the chest is developed, the waist is narrower, and the buttocks are smaller. Fat is mainly distributed in the upper body.
Identification criteria:
the shoulder-to-Hip Ratio (Shoulder-Hip Ratio) is greater than 1.
The upper body (particularly the shoulders and chest) is significantly more girth than the lower body (buttocks and thighs).
More commonly for men, lower body fat rates and higher muscle mass are often accompanied, especially in the upper half.
X type (hourglass type)
Is characterized in that the shoulder and the hip are wider, the waist is obviously thinner, and the typical hourglass shape is shown.
Identification criteria:
The waist-to-hip ratio (WHR) is typically around 0.7.
The waistline is significantly smaller than the shoulder and hip circumference.
The body fat rate is medium or slightly high, and the fat distribution is uniform.
Commonly found in women, is generally determined by the ratio of waist and hip circumference.
7. Visual display
And the three-dimensional rendering technology is used for visually displaying the calculation result so as to more intuitively understand and share the result. Three-dimensional rendering software or libraries may be used to present the model surface area and volume distribution, generating visual reports or animations.
A computer readable medium storing software comprising instructions executable by one or more computers, the instructions, by such execution, causing the one or more computers to perform operations comprising the flow of the system described above.
A computer system comprising:
One or more processors;
a memory storing instructions that, when executed by the one or more processors, cause the one or more processors to perform operations comprising the flow of the system described above.
In the present embodiment of the present invention,
1. The three-dimensional scanning technology can acquire a large amount of point cloud data in extremely short time to form a highly realistic three-dimensional model, and the data acquisition precision is high, so that the fine characteristics of an object can be captured. The method greatly improves the efficiency of data acquisition, ensures the accuracy of data, and provides a solid foundation for the subsequent body type parting.
2. The non-contact measurement is that the three-dimensional scanning technology adopts a non-contact measurement mode, and physical contact with a target object is not needed, so that the measurement can be performed on the premise of not damaging or polluting the object. This avoids errors caused by human factors and improves the accuracy and reliability of the measurement.
3. The three-dimensional scanning technology can work under various environmental conditions, including indoor, outdoor, no-light environment and the like, so that the three-dimensional scanning technology has wide application in various fields. In human body type parting, the characteristic of strong adaptability enables the technology to cope with various complex environments and ensures smooth data acquisition.
4. The intuitiveness and achievement diversity are that the acquired point cloud data not only contains space information, but also has color information and reflectivity value, and can truly reproduce object scenes. In human body type parting, the abundant information can help us to more comprehensively understand the morphological structure of the human body, and improve the parting accuracy. Meanwhile, various achievements can be output by one-time measurement, repeated measurement is not needed, and the working efficiency is improved.
5. The automatic processing and analysis of the three-dimensional scanning data can be realized by combining advanced computer vision and machine learning algorithms, so that the body type typing of the human body can be automatically carried out. The highly-automated processing mode greatly reduces the need of manual intervention and improves the parting efficiency and accuracy.
6. Personalized service, namely, human body type typing based on three-dimensional scanning data can provide personalized service according to specific morphological structures of individuals. For example, in the clothing industry, appropriate clothing can be customized according to individuals of different sizes, and in the fitness industry, personalized fitness plans and suggestions can be provided for individuals of different sizes. .
While the invention has been described with reference to preferred embodiments, it is not intended to be limiting. Those skilled in the art will appreciate that various modifications and adaptations can be made without departing from the spirit and scope of the present invention. Accordingly, the scope of the invention is defined by the appended claims.
Claims (7)
1. A human body model measurement typing method based on three-dimensional reconstruction of human body is characterized by comprising the following steps:
S1, data acquisition
Data acquisition, namely ensuring that a tested person is relatively and statically placed in the center of scanning equipment, and scanning different angles or positions of the tested person according to requirements in the acquisition process;
s2, data preprocessing
Denoising, namely removing noise in the scanned data through filtering to improve the accuracy of subsequent processing;
Registration-using multiple scanning devices or scans, the registration of different data sets is required to ensure that they are in the same coordinate system;
Repairing the missing or damaged part in the data, and performing data repair by interpolation;
S3, reconstructing a model
Selecting a reconstruction algorithm, namely selecting an appropriate reconstruction algorithm according to the data type and the requirement;
Model reconstruction, namely converting the preprocessed data into a three-dimensional model by using a selected algorithm;
the construction of the three-dimensional model requires the reconstruction of point cloud data, and the specific method is as follows:
s3.1, grid generation:
S3.11, dividing the three-dimensional space into a series of regular cubes Cube, and marking the vertex of each Cube as 'inner' or 'outer', depending on the position of the data point;
s3.12, in each cube, determining which vertexes need to be connected by looking up a boundary table;
S3.13, generating triangular patches, and connecting the triangular patches to form a grid, wherein the triangular patches finally form the surface of the whole three-dimensional model;
S3.2, surface fitting, namely for data needing a smooth surface, using a poisson reconstruction algorithm:
S3.21, calculating a gradient field and a normal vector for the point cloud;
s3.22, using the gradient field information to generate a smooth curved surface through global fitting;
S3.23, generating a final triangular patch to represent the model surface;
s4, model segmentation
The specific part is subjected to calculation of surface area and volume, model segmentation is performed, and the calculation is completed through a manual segmentation or automatic segmentation algorithm:
s5, measuring body key data, namely measuring the body key data by using a matched model measuring tool, wherein the key data comprise surface area, volume, length and circumference;
S6, body types are automatically separated, wherein the body types comprise H type, A type, T type and X type, according to waist-hip ratio, shoulder width, hip width, waistline, shoulder circumference, hip circumference, height, waistline/height ratio and shoulder-hip ratio, body type related data are obtained through automatic calculation, and clustering analysis is carried out on the related data through a clustering algorithm.
2. The human body model measurement typing method based on the three-dimensional reconstruction of the human body according to claim 1, wherein the manual segmentation method in S4 is as follows:
The method comprises the steps of importing a complete three-dimensional model into modeling software, manually selecting a region to be segmented by using a segmentation tool in the software, wherein the selected region is independent from the whole model and is used as an independent model part;
Surface area and volume calculations are performed separately for each sub-region.
3. The human body model measurement typing method based on the human body three-dimensional reconstruction of claim 1, wherein the automatic segmentation algorithm method in S4 is as follows:
based on the mesh segmentation algorithm Mesh Segmentation:
grid pretreatment, namely, after a model is imported, the grids are pretreated,
Determining a cutting plane, namely automatically determining the cutting plane through curvature change and geometric characteristics according to the topological structure of the model;
Region segmentation, namely dividing a model into a plurality of subareas according to a cutting plane,
Data extraction, namely calculating the volume and the surface area of each sub-area;
or based on the volumetric segmentation algorithm Voxel-based Segmentation:
Voxelization, namely voxelization of the model and conversion of the model into three-dimensional data consisting of a series of regular cubes;
a region growing algorithm, namely expanding the seed points to adjacent voxels by using the region growing algorithm to form a segmented region;
dividing into different body regions, and measuring the surface area and volume of each region;
or based on morphological segmentation algorithms:
edge detection, namely carrying out edge detection through the geometric outline of the model, and identifying key points of the model;
Morphological operations, namely processing the model by using morphological operations such as expansion, corrosion and the like, and clearly dividing each region;
dividing the model into a head, a trunk and limbs;
surface area and volume calculations further surface area and volume calculations are performed for each region using a specific measurement algorithm.
4. The human body model measurement typing method based on human body three-dimensional reconstruction according to claim 1, wherein:
After model segmentation, key data of the body are measured using a matched measurement tool:
surface area and volume measurements:
Carrying out surface area and volume measurement on each divided area by a built-in tool in three-dimensional modeling software or writing a custom script;
Calculating the surface area and volume of the closed curved surface by using a Gaussian-Bohr formula, or directly calculating by an integration method;
Length and circumference measurements:
The geometric property of the model is used, the key points and the line segments are selected, the lengths and the circumferences of different parts of the body are calculated, a fixed measuring path is set or an algorithm is used for searching the shortest path along the curved surface, and accurate length and circumference data are calculated.
5. The method for measuring and typing a human body model based on three-dimensional reconstruction of a human body according to claim 1, wherein the clustering algorithm classifies data points into different clusters or categories according to similarity between the data points, and in the clustering process, proper clustering parameters are required to be selected to obtain the best clustering effect, and the clustering result is evaluated.
6. A computer readable medium storing software comprising instructions executable by one or more computers, the instructions by such execution causing the one or more computers to perform operations comprising the flow of the mannequin measurement typing method of any one of claims 1 to 5.
7. A computer system comprising:
One or more processors;
A memory storing instructions operable, when executed by the one or more processors, to cause the one or more processors to perform operations comprising the flow of the mannequin measurement typing method of any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202411378342.7A CN119229019B (en) | 2024-09-30 | A human body model measurement and classification method based on human body 3D reconstruction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202411378342.7A CN119229019B (en) | 2024-09-30 | A human body model measurement and classification method based on human body 3D reconstruction |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN119229019A CN119229019A (en) | 2024-12-31 |
| CN119229019B true CN119229019B (en) | 2025-10-10 |
Family
ID=
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107256565A (en) * | 2017-05-19 | 2017-10-17 | 安徽信息工程学院 | The measuring method and system of human body predominant body types parameter based on Kinect |
| CN110135078A (en) * | 2019-05-17 | 2019-08-16 | 上海凌笛数码科技有限公司 | A kind of human parameters automatic generation method based on machine learning |
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107256565A (en) * | 2017-05-19 | 2017-10-17 | 安徽信息工程学院 | The measuring method and system of human body predominant body types parameter based on Kinect |
| CN110135078A (en) * | 2019-05-17 | 2019-08-16 | 上海凌笛数码科技有限公司 | A kind of human parameters automatic generation method based on machine learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Dekker et al. | Building symbolic information for 3D human body modeling from range data | |
| Claes et al. | Computerized craniofacial reconstruction: conceptual framework and review | |
| CN1864074B (en) | Method for determining patient-related information, control device and magnetic resonance tomography apparatus | |
| CN101311967B (en) | Dummy body form establishment method and dummy body form based on body type of actual measurement for crowds | |
| Giachetti et al. | Robust automatic measurement of 3D scanned models for the human body fat estimation | |
| Bibin et al. | Whole-body pregnant woman modeling by digital geometry processing with detailed uterofetal unit based on medical images | |
| US20160180520A1 (en) | Quantitative method for 3-d joint characterization | |
| CN108597017A (en) | A kind of textured bone template construction method based on measurement parameter | |
| Khatam et al. | In-vivo quantification of human breast deformation associated with the position change from supine to upright | |
| Chiu et al. | Automated body volume acquisitions from 3D structured-light scanning | |
| Eulzer et al. | Temporal views of flattened mitral valve geometries | |
| Beylot et al. | 3D interactive topological modeling using visible human dataset | |
| Azouz et al. | Extracting main modes of human body shape variation from 3D anthropometric data | |
| Farinella et al. | Objective outcome evaluation of breast surgery | |
| Grimm et al. | Automatic patient pose estimation using pressure sensing mattresses | |
| CN119229019B (en) | A human body model measurement and classification method based on human body 3D reconstruction | |
| Chen et al. | Breast volume measurement by mesh projection method based on 3D point cloud data | |
| CN119229019A (en) | A human body model measurement and classification method based on human body three-dimensional reconstruction | |
| Jia et al. | 3D Personalized Human Modeling and Deformation Technology for Garment CAD. | |
| Drapikowski | Surface modeling—uncertainty estimation and visualization | |
| Cheong | Computational modeling of breast shape using spherical harmonics | |
| Khaleel et al. | 3D surface reconstruction of coronary artery trees for vessel locations’ detection | |
| Carletti et al. | Analyzing body fat from depth images | |
| Thelwell | Assessing Human Morphology using Statistical Shape Analysis | |
| JP3250040U (en) | A personalized joint skeletal and cartilage morphometry system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant |