CN111179418B - Three-dimensional human body measurement method and device without bare user - Google Patents
Three-dimensional human body measurement method and device without bare user Download PDFInfo
- Publication number
- CN111179418B CN111179418B CN201911410173.XA CN201911410173A CN111179418B CN 111179418 B CN111179418 B CN 111179418B CN 201911410173 A CN201911410173 A CN 201911410173A CN 111179418 B CN111179418 B CN 111179418B
- Authority
- CN
- China
- Prior art keywords
- human body
- dimensional
- triangle
- dimensional human
- representing
- 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
- 238000000691 measurement method Methods 0.000 title abstract description 5
- 230000000877 morphologic effect Effects 0.000 claims abstract description 70
- 238000013178 mathematical model Methods 0.000 claims abstract description 41
- 238000005259 measurement Methods 0.000 claims abstract description 31
- 239000011159 matrix material Substances 0.000 claims description 59
- 238000012549 training Methods 0.000 claims description 54
- 238000000034 method Methods 0.000 claims description 50
- 230000006870 function Effects 0.000 claims description 30
- 239000013598 vector Substances 0.000 claims description 25
- 230000008569 process Effects 0.000 claims description 14
- 210000000988 bone and bone Anatomy 0.000 claims description 12
- 239000004744 fabric Substances 0.000 claims description 11
- 230000005477 standard model Effects 0.000 claims description 8
- 230000009466 transformation Effects 0.000 claims description 7
- 238000001514 detection method Methods 0.000 claims description 6
- 230000005251 gamma ray Effects 0.000 claims description 6
- 238000011872 anthropometric measurement Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 abstract description 7
- 230000037237 body shape Effects 0.000 abstract description 5
- 238000004590 computer program Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 6
- 230000007547 defect Effects 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 230000004660 morphological change Effects 0.000 description 1
- 230000001144 postural effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
-
- A—HUMAN NECESSITIES
- A41—WEARING APPAREL
- A41H—APPLIANCES OR METHODS FOR MAKING CLOTHES, e.g. FOR DRESS-MAKING OR FOR TAILORING, NOT OTHERWISE PROVIDED FOR
- A41H1/00—Measuring aids or methods
- A41H1/02—Devices for taking measurements on the human body
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Graphics (AREA)
- Geometry (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biophysics (AREA)
- Textile Engineering (AREA)
- Processing Or Creating Images (AREA)
Abstract
The invention relates to a three-dimensional human body measurement method and a device without bare users, which are used for acquiring three-dimensional human body original point cloud data of a user object, solving morphological space parameters and posture space parameters related to the user object through a trained three-dimensional human body parameterized mathematical model, reconstructing a three-dimensional human body model of the net body size of the user object, and completing three-dimensional human body measurement through measuring the three-dimensional human body model of the net body size of the user object. According to the invention, the quasi-naked three-dimensional human body model is built by processing the three-dimensional human body original data of the normally-fitted user acquired by the three-dimensional human body scanner, the assistance of manual mark points is not needed, a vivid three-dimensional human body model result is built for the normally-fitted user, and the accurate measurement of the body shape and size of the chest circumference is realized.
Description
Technical Field
The invention relates to the technical fields of computer graphics and anthropometry, in particular to a three-dimensional anthropometry method and device without bare users.
Background
Three-dimensional anthropometry is a fundamental problem in anthropometry and the like, and the dimensions of various parts of the human body such as chest circumference and the like are measured by a tool. Human body measurement can be classified into manual measurement and automatic measurement according to the degree of human participation.
Manual measurement of human body: the measuring staff uses measuring tools such as a flexible tape ruler, an angle meter and the like to measure the human body in a contact mode according to the measuring standard, and directly measures the surface sizes of the human body such as the vertical direction, the transverse direction, the circumference and the like of the human body, such as the body size of the human body such as the height, the chest circumference, the waistline, the hip circumference, the crotch, the arm length, the leg length and the like. The contact measurement data are obtained manually by a measuring engineer, have low efficiency and poor objectivity, and have inconvenience and errors caused by a plurality of human factors, such as anisotropic contact measurement, fatigue measurement, subjective increase of release amount, and difference in measurement datum point position selection caused by different methods of different measuring engineers. Thus, the manual measurement of the human body is difficult to standardize, and a large error exists between the measurement result and the actual size data of the human body.
Automatic measurement of human body: the non-contact automatic measurement of three-dimensional human body surface by using three-dimensional human body scanner developed by optical measurement and computer technology. Thanks to the development of hardware devices such as three-dimensional anthropometric scanners, three-dimensional anthropometric measurements have evolved to the stage of directly acquiring (quasi-bare) three-dimensional anthropometric surface raw data with the hardware devices. Based on three-dimensional human body surface data, the solving process of the body shape and size and other measured values is approximately as follows: and identifying measurement reference feature points which are pre-configured on the quasi-naked user object, and completing measurement of the three-dimensional space geometric dimensions of the feature points corresponding to the chest circumference and the like. It follows that if the configuration or identification of the feature points deviates, the measurement result of the human body size has precision errors and even the reliability of the data. A more serious problem is that the requirement for bare users is unreasonable, which greatly limits the application of hardware devices such as three-dimensional body scanners in daily life.
In the prior art, CN 201310555513.4 and CN 2013102159963 are three-dimensional human body morphology and posture modeling methods proposed by the subject group. The above patent proposes an original three-dimensional body parametric modeling algorithm, but does not address the challenging problem of three-dimensional body measurement without the need for bare users.
The body parametric models proposed by CN 201710079335.0, CN 201710079314.9 and CN 201710079459.9 are essentially generalized descriptions of existing patents CN 201310555513.4 and CN 2013102159963, nor do they relate to the three-dimensional anthropometric problem of how to accomplish a normal wearing user.
Disclosure of Invention
In order to solve the defects in the prior art, the invention provides a three-dimensional human body measurement method and device without bare body of a user, and solves the problem that the prior art cannot support the three-dimensional human body measurement of a normal wearing user.
In order to achieve the above object, the present invention adopts the following technical scheme: a three-dimensional anthropometric method without requiring bare user, comprising the steps of:
the three-dimensional human body original point cloud data of the user object is obtained, morphological space parameters and posture space parameters related to the user object are solved through the trained three-dimensional human body parameterized mathematical model, meanwhile, a net body size three-dimensional human body model of the user object is reconstructed, and three-dimensional human body measurement is completed through measuring the net body size three-dimensional human body model of the user object.
Further, the three-dimensional human body parameterized mathematical model is:
M(β,θ；Φ):e＝B f (θ)S f (β)Q f (θ)e* (1)
S f (beta) is a morphological deformation matrix,B f (θ) is a rigid posture deformation matrix, Q f (θ) is a non-rigid pose deformation matrix; θ is a posture space parameter, β represents a morphology space parameter, and a constant Φ is a morphology and posture parameter learned in advance through data training; e is the side vector of triangle mesh in the pre-established three-dimensional human body standard model; e is the edge vector after deformation.
Further, the method comprises the steps of,
wherein |beta| is the number of morphological feature parameters, ψ b',f (0 is less than or equal to b '<|beta|) is the linear coefficient of the b' morphological characteristic parameter on the f triangle, and ψ |β|,f Is the offset of the f triangle on the morphological deformation matrix S, beta b' Representing the b' th morphological feature parameter;
wherein |bone| represents the number of rigid parts, w b,f Is the skin weight of the b-th rigid part to the f-th triangle, R (θ) b ) A rigid transformation matrix representing a b-th rigid component;
wherein, gamma 0,f The deformation matrix representing the situation that the rigid part of the human body is not deformed relatively is a unit matrix; gamma ray b,f Represents the b-th Rodrigas rotation vector θ b Linear coefficients on the f-th triangle.
Further, training the three-dimensional human parameterized mathematical model, the process comprising:
for the pre-established standard triangle grid T * Deforming to register with each sample k of the bare three-dimensional human body database, and accurately matching the deformed vertexes with the vertexes of the sample k to train the three-dimensional human bodyAnd (5) a volume parameterized mathematical model.
Further, in constant parameters of the three-dimensional human body parameterized mathematical model, each rigid part predesignates skin weights of each triangle; calculating a morphological constant parameter psi and a posture constant parameter gamma of the three-dimensional human body parameterized mathematical model;
the process of solving the constant parameters is to find the optimal constants ψ, Γ and find the optimal triangle mesh model for the human body in each training set such that the following objective functions are satisfied:
T k triangle mesh model for representing kth human body in training set, lambda 1 ，λ 2 ，λ 3 Is a weight parameter, pre-designated; θ k Representing the pose parameters of the kth human body in the training set, beta k Representing morphological parameters of a kth human body in the training set;representing a standard triangular mesh T * I=0, 1,2, representing the three edge numbers of the triangle; />Triangle mesh T representing kth human body in training set k The ith side of the f triangle; adj= { (f 1, f 2), f1 and f2 neighbors } represent T * F1, f2 represent two adjacent triangles; i is a unit 3-order matrix; v (V) k A set of all vertices of a triangular mesh representing a kth person in the training set; p (P) k Representing a set of all points of a point cloud model of a kth human body in a training set;
optimizing an objective function by using a trust domain method, completing alternate iterative computation of each type of variable through gradient descent until the computation of the objective function converges, solving constant parameters psi and Γ, and further obtaining a trained three-dimensional human body parameterized mathematical model.
Further, the morphological space parameter and the gesture space parameter related to the user object are solved, and simultaneously, the three-dimensional human model of the net body size of the user object is reconstructed, wherein the process is as follows:
skin detection is carried out on the three-dimensional image original data of the user, and two types of areas are identified: real skin and head point cloud P Skin Affiliated clothing point cloud P Cloth The method comprises the steps of carrying out a first treatment on the surface of the For P Skin The point cloud requires the deformed three-dimensional human body model T with the net body size to be matched with the three-dimensional human body model T; for P Cloth The point cloud requires that the deformed net-size three-dimensional human body model T is arranged on the inner side of the clothes and is close to the clothes, and is realized by using a Geman-McClure function rho (;
wherein e f,i The ith side of the f triangle of the triangle mesh representing the user object, and V represents all vertex sets of the triangle mesh of the user object; and (3) solving by adopting a dogleg method, and calculating a morphological variable beta and a posture variable theta of the user object and a deformed net body size three-dimensional human model T.
A three-dimensional anthropometric apparatus that does not require bare user, comprising:
the user data acquisition module is used for acquiring three-dimensional human body original point cloud data of a user object;
the solving module is used for solving morphological space parameters and posture space parameters related to the user object through the trained three-dimensional human parameterized mathematical model according to the three-dimensional human body origin cloud data of the user object, and reconstructing a net body size three-dimensional human body model of the user object;
and the measurement module is used for completing three-dimensional human body measurement by measuring the three-dimensional human body model of the net body size of the user object.
Further, the three-dimensional human body parameterized mathematical model is:
M(β,θ；Φ):e＝B f (θ)S f (β)Q f (θ)e* (1)
S f (beta) is a morphological deformation matrix, B f (θ) is a rigid posture deformation matrix, Q f (θ) is a non-rigid pose deformation matrix; θ is a posture space parameter, β represents a morphology space parameter, and a constant Φ is a morphology and posture parameter learned in advance through data training; e is the side vector of triangle mesh in the pre-established three-dimensional human body standard model; e is the edge vector after deformation;
wherein |beta| is the number of morphological feature parameters, ψ b',f B 'beta < |is 0.ltoreq.b' <|beta| which is the linear coefficient of the b 'th morphological characteristic parameter on the f-th triangle, and ψ is the linear coefficient of the b' th morphological characteristic parameter on the f-th triangle |β|,f Is the offset of the f triangle on the morphological deformation matrix S, beta b' Representing the b' th morphological feature parameter;
wherein |bone| represents the number of rigid parts, w b,f Is the skin weight of the b-th rigid part to the f-th triangle, R (θ) b ) A rigid transformation matrix representing a b-th rigid component;
wherein, gamma 0,f The deformation matrix representing the situation that the rigid part of the human body is not deformed relatively is a unit matrix; gamma ray b,f Represents the b-th Rodrigas rotation vector θ b Linear coefficients on the f-th triangle.
Further, in constant parameters of the three-dimensional human body parameterized mathematical model, each rigid part predesignates skin weights of each triangle; calculating a morphological constant parameter psi and a posture constant parameter gamma of the three-dimensional human body parameterized mathematical model;
the process of solving the constant parameters is to find the optimal constants ψ, Γ and find the optimal triangle mesh model for the human body in each training set such that the following objective functions are satisfied:
T k triangle mesh model for representing kth human body in training set, lambda 1 ，λ 2 ，λ 3 Is a weight parameter, pre-designated; θ k Representing the pose parameters of the kth human body in the training set, beta k Representing morphological parameters of a kth human body in the training set;representing a standard triangular mesh T * I=0, 1,2, representing the three edge numbers of the triangle; />Triangle mesh T representing kth human body in training set k The ith side of the f triangle; adj= { (f 1, f 2), f1 and f2 neighbors } represent T * F1, f2 represent two adjacent triangles; i is a unit 3-order matrix; v (V) k A set of all vertices of a triangular mesh representing a kth person in the training set; p (P) k Representing a set of all points of a point cloud model of a kth human body in a training set;
optimizing an objective function by using a trust domain method, completing alternate iterative computation of each type of variable through gradient descent until the computation of the objective function converges, solving constant parameters psi and Γ, and further obtaining a trained three-dimensional human body parameterized mathematical model.
Further, the morphological space parameter and the gesture space parameter related to the user object are solved, and simultaneously, the three-dimensional human model of the net body size of the user object is reconstructed, wherein the process is as follows:
skin detection is carried out on the three-dimensional image original data of the user, and two types of areas are identified: real skin and head point cloud P Skin Affiliated clothing point cloud P Cloth The method comprises the steps of carrying out a first treatment on the surface of the For P Skin The point cloud requires the deformed three-dimensional human body model T with the net body size to be matched with the three-dimensional human body model T; for P Cloth The point cloud requires that the deformed net-size three-dimensional human body model T is arranged on the inner side of the clothes and is close to the clothes, and is realized by using a Geman-McClure function rho (;
wherein e f,i The ith side of the f triangle of the triangle mesh representing the user object, and V represents the set of all vertices of the triangle mesh of the user object; and (3) solving by adopting a dogleg method, and calculating a morphological variable beta and a posture variable theta of the user object and a deformed net body size three-dimensional human model T.
The invention has the beneficial effects that:
according to the invention, the quasi-naked three-dimensional human body model is built by processing the three-dimensional human body original data of the normally-fitted user acquired by the three-dimensional human body scanner, the assistance of manual mark points is not needed, a vivid three-dimensional human body model result is built for the normally-fitted user, and the accurate measurement of the body shape and size of the chest circumference is realized.
Drawings
FIG. 1 is a flow chart of a measurement method in an embodiment of the invention.
Detailed Description
The invention is further described below with reference to the accompanying drawings. The following examples are only for more clearly illustrating the technical aspects of the present invention, and are not intended to limit the scope of the present invention.
Example 1:
as shown in fig. 1, a three-dimensional anthropometric method without bare user, comprising the steps of:
step 1, establishing a three-dimensional human body parameterized mathematical model representing a three-dimensional human body space;
based on the non-rigid deformation mathematical theory of the three-dimensional human body geometric space, the invention uses a main element analysis linearization method to complete the solution of a three-dimensional human body parameterization mathematical model equation, thereby simplifying the three-dimensional human body parameterization mathematical model into a quasi-linearization approximation solution equation;
assuming that a single three-dimensional human discretization is represented as a triangular mesh, including vertex v= { V 0 ,...,v |V|-1 Edge e= { E 0 ,...,e |E|-1 A (each edge e is a vector of two adjacent vertices) and a plane f= { F 0 ,...,f |F|-1 Each face f (i.e., a triangle) is a triangle formed by three adjacent vertices), v 0 Representing the first vertex, v |V|-1 Represents the |V| -1 vertex, e 0 Representing the first edge, e |E|-1 Represent the first |E|-1 Sides, f 0 Representing the first face, f |F|-1 Represent the first |F|-1 A plurality of faces; the three-dimensional human body model T= { V, E and F } of any human body is obtained based on T deformation, and V, E and F respectively represent the vertex, edge and face vector sets of the triangular mesh in the three-dimensional human body standard model; v, E, F represent the vertex, edge and face vector sets of the triangular mesh in the three-dimensional human model, respectively.
Three-dimensional human body parameterized mathematical model under triangle mesh descriptionExpressing the related parameters of human body to three-dimensional human body space +.>Mapping of->Wherein the variable beta represents morphological space parameters such as height, weight, three-dimensional and the like; the three-dimensional human body is abstracted into a joint model, and divided into |bone| (|bone|=19) rigid parts, b < th >The relative rotation of (1.ltoreq.b < |bone|) rigid members with respect to the parent rigid member uses a three-dimensional Rodrigues vector θ b Representing the three-dimensional Rodrigues vector θ of the root rigid component 0 Representing absolute rotation, θ, relative to a three-dimensional human standard model T # b And theta 0 The combination of the two represents a gesture space parameter theta (the theta dimension is 19 x 3 = 57), and the constant phi is a morphology and gesture parameter which is learned in advance through data training, and comprises a morphology constant parameter set psi= { psi b',f B' =0, 1,2, |β| -1, f=0, 1, |f| -1}, and a posture constant parameter set Γ= { γ b,f B=0, 1.|lane| -1, f=0, 1.|f| -1} and skin weight set w= { W b,f B=0, 1.|bond| -1, f=0, 1.|f| -1}, β|indicates the number of morphological spatial parameters, b indicates the number of rigid parts, b' indicates the number of morphological spatial parameters, ψ b',f Linear coefficient (9*1 matrix) representing the b' th morphological spatial parameter on the f-th triangle, gamma b,f Linear coefficients (9*3 matrix), w representing the b-th rode-g rotation vector on the f-th triangle b,f Representing the skin weight of the b-th rigid part to the f-th triangle. The three-dimensional human body parameterized mathematical model is a typical second-order partial differential equation with constraint conditions due to nonlinear factors such as non-rigid deformation. The f-th triangle side vector e passes through a non-rigid small-scale posture deformation matrix Q under the combined action of variables and constants f (θ), morphological deformation matrix S f (beta) and weighted skin Large Scale pose deformation matrix B f (theta) to obtain a deformed edge vector E, and finally reconstructing a geometric model T= { V, E, F }, Q of the three-dimensional grid through the poisson surface f ，S f ，B f Each being a 3*3 matrix.
The three-dimensional human parameterized mathematical model of the present invention may be formally represented as:
M(β,θ；Φ):e＝B f (θ)S f (β)Q f (θ)e* (1)
morphological deformation matrix S f (beta) describes the diversity of the three-dimensional human body morphological change on the f triangle, S f (beta) can be used as a line for morphological spatial parameter betaSex combination means:wherein |beta| is the number of morphological spatial parameters, ψ b',f (0.ltoreq.b '<|beta|) is the linear coefficient (9*1 matrix) of the b' th morphological spatial parameter on the f-th triangle, ψ |β|,f The offset of the f-th triangle form deformation matrix S (9*1 matrix), beta represents the form parameters such as height, weight, three circles and the like, beta b' Representing the b' th morphological space parameter, experiments show that the number of morphological space spaces |beta| is designated as 42, and that 1000 human training models are sufficient to calculate the morphological constant parameter ψ.
Gesture deformation matrix B f (θ)、Q f (θ) all delineate the diversity of the f-th triangle from three-dimensional body posture changes. Joint deformation method using linear skin, B f (θ) represents the large-scale deformation driven by the rigid skeleton member, i.e., |bone| (|bone|=19) transformation of the rigid members and summation of skin weight products in the human body,|bone| represents the number of rigid parts, w b,f Is an element in the set W, and represents the skin weight of the b rigid part to the f triangle, R (theta b ) A rigid transformation matrix representing the b-th rigid part, the rigid matrix being a 3*3 identity orthogonal matrix;
formally representing local small scale postural deformations in the body shape, gamma 0,f Indicating that the rigid body member is not deformed relatively (theta b A deformation matrix (9*1 matrix) with the zero vector of (1.ltoreq.b < |bone|) is an identity matrix; gamma ray b,f Represents the b-th Rodrigas rotation vector θ b The linear coefficients on the f-th triangle (9*3 matrix).
Step 2, training a three-dimensional human body parameterized mathematical model through a pre-established three-dimensional human body standard model to obtain a trained three-dimensional human body parameterized mathematical model;
for the pre-established standard triangle grid T * And (3) deforming to enable the deformed vertex to be registered with each sample k of the naked three-dimensional human body database, namely accurately matching the deformed vertex with the vertex of the sample k, so as to train the three-dimensional human body parameterized mathematical model.
In the constant parameters, the skin weight of each rigid part in W is manually specified in advance for each triangle. In the specific data training process, two other constant parameters ψ, Γ of a three-dimensional human body parameterized model are calculated.
The process of solving the constant parameters is to find the optimal constants ψ, Γ and find the optimal triangle mesh model for the human body in each training set such that the following objective functions are satisfied:
T k triangle mesh model for representing kth human body in training set, lambda 1 ，λ 2 ，λ 3 Is a weight parameter, pre-specified. θ k Representing the pose parameters of the kth human body in the training set, beta k Representing morphological parameters of the kth human body in the training set.Representing a standard triangular mesh T * I=0, 1,2, representing the three edge numbers of the triangle; />Triangle mesh T representing kth human body in training set k The ith side of the f-th triangle of (c). Adj= { (f 1, f 2), f1 and f2 neighbors } represent T * All adjacent triangle pairs in (1), f1, f2 represent two adjacent triangles, S f 1(),S f2 () The morphological deformation matrices of two adjacent triangles f1, f2 are respectively represented. I is a unit 3-order matrix. V (V) k Triangle mesh station for representing kth human body in training setThere is a collection of vertices. P (P) k Representing the set of all points of the point cloud model of the kth human body in the training set. The first term of equation (2) ensures the optimal triangle mesh T of the kth human body in the training set k Is composed of T * According to theta k ，β k The deformed model, the second term ensures that adjacent triangles have similar morphological deformation matrixes, the third term ensures that the gesture deformation matrixes keep rigidity as much as possible, and the fourth term ensures that the optimal triangle grid vertexes V of each training set human body k Vertex P with sample k k Precisely matches. The solution of the function involves the calculation of a typical second-order partial differential equation with constraint conditions, with ψ and Γ variables and multidimensional; optimizing an objective function (namely a trust domain dog leg method) by using a trust domain method, completing alternate iterative computation of each type of variable through gradient descent until the computation of the objective function converges, solving constant parameters psi and Γ, and further obtaining a trained three-dimensional human body parameterized mathematical model;
step 3, user instance processing: for a user object to be processed, three-dimensional human body original point cloud data of the user object are acquired, morphological space parameters beta and posture space parameters theta related to the user object are solved through a trained three-dimensional human body parameterized mathematical model, meanwhile, a net body size three-dimensional human body model of the user object is reconstructed, and the measurement of the body shape size of the chest circumference is completed through measuring the model.
In the solving process, firstly, skin detection is carried out on three-dimensional image original data of a user, and two types of areas are identified: real skin and head point cloud (i.e. set of points) P Skin Affiliated clothing point cloud P Cloth . For P Skin The point cloud requires the deformed three-dimensional human body model T with the net body size to be matched with the three-dimensional human body model T; for P Cloth The point cloud requires that the deformed net-size three-dimensional mannequin T be inside and nearer to the garment, and this is achieved using the Geman-McClure function ρ (·).
As in equation (2), equation (3) is also solved by the dopreg method, and the morphological and posture variables β and θ of the user object and the deformed net body size three-dimensional human model T are calculated. This also means that the invention directly completes the solution of the morphological variable β, unlike the common morphological variable indirect calculation scheme.
Example 2:
a three-dimensional anthropometric apparatus that does not require bare user, comprising:
the user data acquisition module is used for acquiring three-dimensional human body original point cloud data of a user object;
the solving module is used for solving morphological space parameters and posture space parameters related to the user object through the trained three-dimensional human parameterized mathematical model according to the three-dimensional human body origin cloud data of the user object, and reconstructing a net body size three-dimensional human body model of the user object;
and the measurement module is used for completing three-dimensional human body measurement by measuring the three-dimensional human body model of the net body size of the user object.
Further, the three-dimensional human body parameterized mathematical model is:
S f (beta) is a morphological deformation matrix, B f (θ) Large-scale pose deformation matrix for weighted skin, Q f (θ) is a non-rigid small scale pose deformation matrix; e is the side vector of triangle mesh in the pre-established three-dimensional human body standard model; e is the edge vector after deformation;
wherein |beta| is the number of morphological feature parameters, ψ b',f (0 is less than or equal to b '<|beta|) is the linear coefficient of the b' morphological characteristic parameter on the f triangle, and ψ |β|,f Is the f triangleThe shift amount of the morphological deformation matrix S, beta represents morphological characteristic parameters, beta b' Representing the b' th morphological feature parameter;
wherein |bone| represents the number of rigid parts, w b,f Is the skin weight of the b-th rigid part to the f-th triangle, R (θ) b ) A rigid transformation matrix representing a b-th rigid component;
wherein, gamma 0,f The deformation matrix representing the situation that the rigid part of the human body is not deformed relatively is a unit matrix; gamma ray b,f Represents the b-th Rodrigas rotation vector θ b Linear coefficients on the f-th triangle.
Further, in constant parameters of the three-dimensional human body parameterized mathematical model, each rigid part predesignates skin weights of each triangle; calculating a morphological constant parameter psi and a posture constant parameter gamma of the three-dimensional human body parameterized mathematical model;
the process of solving the constant parameters is to find the optimal constants ψ, Γ and find the optimal triangle mesh model for the human body in each training set such that the following objective functions are satisfied:
T k triangle mesh model for representing kth human body in training set, lambda 1 ，λ 2 ，λ 3 Is a weight parameter, pre-designated; θ k Representing the pose parameters of the kth human body in the training set, beta k Representing morphological parameters of a kth human body in the training set;representing a standard triangular mesh T * I=0, 1,2, representing the three edge numbers of the triangle; />Triangle mesh T representing kth human body in training set k The ith side of the f triangle; adj= { (f 1, f 2), f1 and f2 neighbors } represent T * F1, f2 represent two adjacent triangles; i is a unit 3-order matrix; v (V) k A set of all vertices of a triangular mesh representing a kth person in the training set; p (P) k Representing a set of all points of a point cloud model of a kth human body in a training set;
optimizing an objective function by using a trust domain method, completing alternate iterative computation of each type of variable through gradient descent until the computation of the objective function converges, solving constant parameters psi and Γ, and further obtaining a trained three-dimensional human body parameterized mathematical model.
Further, the morphological space parameter and the gesture space parameter related to the user object are solved, and simultaneously, the three-dimensional human model of the net body size of the user object is reconstructed, wherein the process is as follows:
skin detection is carried out on the three-dimensional image original data of the user, and two types of areas are identified: real skin and head point cloud P Skin Affiliated clothing point cloud P Cloth The method comprises the steps of carrying out a first treatment on the surface of the For P Skin The point cloud requires the deformed three-dimensional human body model T with the net body size to be matched with the three-dimensional human body model T; for P Cloth The point cloud requires that the deformed net-size three-dimensional human body model T is arranged on the inner side of the clothes and is close to the clothes, and is realized by using a Geman-McClure function rho (;
and (3) solving by adopting a dogleg method, and calculating a morphological variable beta and a posture variable theta of the user object and a deformed net body size three-dimensional human model T.
In summary, the three-dimensional anthropometric method provided by the invention has stable functions without bare users: because the solved result is in accordance with the three-dimensional human body parameterized mathematical model, even though the result is single-view incomplete data, the complete three-dimensional human body model result can be established. The method avoids the defect that the result is defective due to data defect in the traditional three-dimensional human body acquisition method, and reduces the influence caused by the original data defects such as holes, jitter, outliers and the like. More importantly, even if the user wears normally, the enhanced three-dimensional human body parameterization modeling method can still establish a three-dimensional human body result of the net body size, and accurate measurement of the three-dimensional human body size such as chest circumference is completed.
It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The foregoing is merely a preferred embodiment of the present invention, and it should be noted that modifications and variations could be made by those skilled in the art without departing from the technical principles of the present invention, and such modifications and variations should also be regarded as being within the scope of the invention.
Claims (7)
1. A three-dimensional anthropometric method without requiring bare user, comprising the steps of:
acquiring three-dimensional human body original point cloud data of a user object, solving morphological space parameters and posture space parameters related to the user object through a trained three-dimensional human body parameterized mathematical model, simultaneously reconstructing a net body size three-dimensional human body model of the user object, and measuring the net body size three-dimensional human body model of the user object to finish three-dimensional human body measurement;
the three-dimensional human body parameterized mathematical model is as follows:
S f (beta) is a morphological deformation matrix, B f (θ) is a rigid posture deformation matrix, Q f (θ) is a non-rigid pose deformation matrix; θ is a posture space parameter, β represents a morphology space parameter, and a constant Φ is a morphology and posture parameter learned in advance through data training; e isTriangle side vectors of triangle meshes in a pre-established three-dimensional human body standard model; e is the edge vector after deformation;
wherein |beta| is the number of morphological spatial parameters, ψ b′，f Is the linear coefficient of the b 'morphological space parameter on the f triangle, b' is more than or equal to 0 and less than or equal to beta, and psi |β|,f In-form deformation matrix S for the f-th triangle f Offset on (beta), beta b' Representing the b' th morphological spatial parameter;
wherein |bone| represents the number of rigid parts, w b,f Is the skin weight of the b-th rigid part to the f-th triangle, R (θ) b ) A rigid transformation matrix representing a b-th rigid component;
wherein, gamma 0,f The deformation matrix representing the situation that the rigid part of the human body is not deformed relatively is a unit matrix; gamma ray b,f Represents the b-th Rodrigas rotation vector θ b Linear coefficients on the f-th triangle.
2. A method of three-dimensional anthropometric measurement without requiring bare user of claim 1 wherein training a three-dimensional anthropometric mathematical model comprises:
for the pre-established standard triangle grid T * Deforming to register with each sample k of the bare three-dimensional human body database, and accurately matching the deformed vertexes with the vertexes of the sample k to train three-dimensional human body parameterization mathematicsAnd (5) a model.
3. A three-dimensional anthropometric method without requiring bare users according to claim 2, wherein, in the constant parameters of the three-dimensional anthropometric mathematical model, each rigid part is pre-assigned skin weights for each triangle; calculating a morphological constant parameter psi and a posture constant parameter gamma of the three-dimensional human body parameterized mathematical model;
the process of solving the constant parameters is to find the optimal constants ψ, Γ and find the optimal triangle mesh model for the human body in each training set such that the following objective functions are satisfied:
T k triangle mesh model for representing kth human body in training set, lambda 1 ，λ 2 ，λ 3 Is a weight parameter, pre-designated; θ k Representing the pose parameters of the kth human body in the training set, beta k Representing morphological parameters of a kth human body in the training set;representing a standard triangular mesh T * I=0, 1,2, representing the three edge numbers of the triangle; />Triangle mesh T representing kth human body in training set k The ith side of the f triangle; adj represents T * F1, f2 represent two adjacent triangles; i is a unit 3-order matrix; v (V) k A set of all vertices of a triangular mesh representing a kth person in the training set; p (P) k Representing a set of all points of a point cloud model of a kth human body in a training set; |f| represents the number of faces in the grid;
optimizing an objective function by using a trust domain method, completing alternate iterative computation of each type of variable through gradient descent until the computation of the objective function converges, solving constant parameters psi and Γ, and further obtaining a trained three-dimensional human body parameterized mathematical model.
4. A method of three-dimensional anthropometric measurement without the need for bare users according to claim 3, wherein the morphological spatial parameters and the pose spatial parameters associated with the user object are solved, while a net-dimensional three-dimensional anthropometric model of the user object is reconstructed by the steps of:
skin detection is carried out on the three-dimensional image original data of the user, and two types of areas are identified: real skin and head point cloud P Skin Affiliated clothing point cloud P Cloth ；
Wherein ρ (·) is a Geman-McClure function, e f,i The ith side of the f triangle of the triangle mesh representing the user object, and V represents the set of all vertices of the triangle mesh of the user object; and (3) solving by adopting a dogleg method, and calculating a morphological space parameter beta and a posture space parameter theta of the user object and a deformed net body size three-dimensional human model.
5. A three-dimensional anthropometric apparatus that does not require bare user, comprising:
the user data acquisition module is used for acquiring three-dimensional human body original point cloud data of a user object;
the solving module is used for solving morphological space parameters and posture space parameters related to the user object through the trained three-dimensional human parameterized mathematical model according to the three-dimensional human body origin cloud data of the user object, and reconstructing a net body size three-dimensional human body model of the user object;
the measurement module is used for completing three-dimensional human body measurement by measuring a three-dimensional human body model of the net body size of the user object;
the three-dimensional human body parameterized mathematical model is as follows:
S f (beta) is a morphological deformation matrix, B f (θ) is a rigid posture deformation matrix, Q f (θ) is a non-rigid pose deformation matrix; θ is a posture space parameter, β represents a morphology space parameter, and a constant Φ is a morphology and posture parameter learned in advance through data training; e is the side vector of triangle mesh in the pre-established three-dimensional human body standard model; e is the edge vector after deformation;
wherein |beta| is the number of morphological spatial parameters, ψ b′，f Is the linear coefficient of the b 'morphological space parameter on the f triangle, b' is more than or equal to 0 and less than or equal to beta, and psi |β|,f In-form deformation matrix S for the f-th triangle f Offset on (beta), beta b' Representing the b' th morphological spatial parameter;
wherein |bone| represents the number of rigid parts, w b,f Is the skin weight of the b-th rigid part to the f-th triangle, R (θ) b ) A rigid transformation matrix representing a b-th rigid component;
wherein, gamma 0,f The deformation matrix representing the situation that the rigid part of the human body is not deformed relatively is a unit matrix; gamma ray b,f Represents the b-th Rodrigas rotation vector θ b Linear coefficients on the f-th triangle.
6. A three-dimensional anthropometric apparatus without requiring bare users according to claim 5 wherein, of the constant parameters of the three-dimensional anthropometric parameterization mathematical model, each rigid member is pre-assigned skin weights for each triangle; calculating a morphological constant parameter psi and a posture constant parameter gamma of the three-dimensional human body parameterized mathematical model;
the process of solving the constant parameters is to find the optimal constants ψ, Γ and find the optimal triangle mesh model for the human body in each training set such that the following objective functions are satisfied:
T k triangle mesh model for representing kth human body in training set, lambda 1 ，λ 2 ，λ 3 Is a weight parameter, pre-designated; θ k Representing the pose parameters of the kth human body in the training set, beta k Representing morphological parameters of a kth human body in the training set;representing a standard triangular mesh T * I=0, 1,2, representing the three edge numbers of the triangle; />Triangle mesh T representing kth human body in training set k The ith side of the f triangle; adj represents T * F1, f2 represent two adjacent triangles; i is a unit 3-order matrix; v (V) k A triangle mesh all vertex set representing the kth human body in the training set; p (P) k Representing all point sets of a point cloud model of a kth human body in a training set; |f| represents the number of faces in the grid;
optimizing an objective function by using a trust domain method, completing alternate iterative computation of each type of variable through gradient descent until the computation of the objective function converges, solving constant parameters psi and Γ, and further obtaining a trained three-dimensional human body parameterized mathematical model.
7. The apparatus of claim 6, wherein the method for solving the morphological spatial parameters and the gesture spatial parameters associated with the user object and reconstructing the net-size three-dimensional human model of the user object comprises the steps of:
skin detection is carried out on the three-dimensional image original data of the user, and two types of areas are identified: real skin and head point cloud P Skin Affiliated clothing point cloud P Cloth The method comprises the steps of carrying out a first treatment on the surface of the For P Skin The point cloud requires the deformed three-dimensional human body model T with the net body size to be matched with the three-dimensional human body model T; for P Cloth The point cloud requires that the deformed net-size three-dimensional human body model T is arranged on the inner side of the clothes and is close to the clothes, and is realized by using a Geman-McClure function rho (;
wherein e f,i The ith side of the f triangle of the triangle mesh representing the user object, and V represents the set of all vertices of the triangle mesh of the user object; and (3) solving by adopting a dogleg method, and calculating a morphological space parameter beta and a posture space parameter theta of the user object and a deformed net body size three-dimensional human model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911410173.XA CN111179418B (en) | 2019-12-31 | 2019-12-31 | Three-dimensional human body measurement method and device without bare user |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911410173.XA CN111179418B (en) | 2019-12-31 | 2019-12-31 | Three-dimensional human body measurement method and device without bare user |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111179418A CN111179418A (en) | 2020-05-19 |
| CN111179418B true CN111179418B (en) | 2024-02-27 |
Family
ID=70652336
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911410173.XA Active CN111179418B (en) | 2019-12-31 | 2019-12-31 | Three-dimensional human body measurement method and device without bare user |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111179418B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112233223A (en) * | 2020-09-29 | 2021-01-15 | 深圳市易尚展示股份有限公司 | Automatic human body parametric model deformation method and device based on three-dimensional point cloud |
| CN112700540A (en) * | 2021-01-12 | 2021-04-23 | 北京著手健康科技有限公司 | Three-dimensional human body model construction method, device, equipment and readable storage medium |
| CN112991503B (en) * | 2021-04-22 | 2021-07-27 | 腾讯科技（深圳）有限公司 | Model training method, device, equipment and medium based on skin weight |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108629831A (en) * | 2018-04-10 | 2018-10-09 | 清华大学 | 3 D human body method for reconstructing and system based on parametric human body template and inertia measurement |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108694740A (en) * | 2017-03-06 | 2018-10-23 | 索尼公司 | Information processing equipment, information processing method and user equipment |
-
2019
- 2019-12-31 CN CN201911410173.XA patent/CN111179418B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108629831A (en) * | 2018-04-10 | 2018-10-09 | 清华大学 | 3 D human body method for reconstructing and system based on parametric human body template and inertia measurement |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111179418A (en) | 2020-05-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111179418B (en) | Three-dimensional human body measurement method and device without bare user | |
| Brett et al. | A method of automated landmark generation for automated 3D PDM construction | |
| US20120020573A1 (en) | Image analysis systems using non-linear data processing techniques and methods using same | |
| US20020003546A1 (en) | Virtual shape generation method and device using the same | |
| CN105608737B (en) | A kind of human foot three-dimensional rebuilding method based on machine learning | |
| JP6741305B2 (en) | Vertebral posture estimation device | |
| Ballester et al. | Data-driven three-dimensional reconstruction of human bodies using a mobile phone app | |
| CN104021547A (en) | Three dimensional matching method for lung CT | |
| CN102208109A (en) | Different-source image registration method for X-ray image and laser image | |
| CN103822581B (en) | A kind of irregularly shaped object volume measuring method based on compressed sensing | |
| CN108509908A (en) | A kind of pupil diameter method for real-time measurement based on binocular stereo vision | |
| CN111127641B (en) | Three-dimensional human body parameterized modeling method with high-fidelity facial features | |
| Kaashki et al. | Deep learning-based automated extraction of anthropometric measurements from a single 3-D scan | |
| Rouwane et al. | Adjusting fictitious domain parameters for fairly priced image-based modeling: Application to the regularization of Digital Image Correlation | |
| Bouaziz et al. | Dynamic 2D/3D registration for the Kinect | |
| Wang | Computing length-preserved free boundary for quasi-developable mesh segmentation | |
| Danckaers et al. | Evaluation of 3D body shape predictions based on features | |
| Azouz et al. | Extracting main modes of human body shape variation from 3D anthropometric data | |
| CN114463433A (en) | In-vivo SLAM method and system based on fusion of optical fiber strain sensor and monocular camera | |
| CN113689478A (en) | Alignment method, device and system of measuring equipment | |
| Yahara et al. | Estimation of anatomical landmark positions from model of 3‐dimensional foot by the FFD method | |
| Knyaz et al. | Photogrammetric techniques for paleoanthropological objects preserving and studying | |
| Gu et al. | 3D female upper body modelling based on 2D images | |
| Novikov et al. | Creating digital models of paleontological sample by photogrammetry and computed tomography | |
| Ahmad et al. | Longitudinal scoliotic trunk analysis via spectral representation and statistical analysis |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |