JP2011180790A - Apparatus, method, program and system for generating three-dimensional shape model - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an apparatus, method, program and system for generating a three-dimensional shape model, to increase the accuracy of a new three-dimensional shape model to be generated with the use of statistical processing. <P>SOLUTION: A model generation part adds displacements T<SB>1</SB>-T<SB>5</SB>from polygon coordinates Ka<SB>1</SB>-Ka<SB>5</SB>constituting an initial homology model HMa generated with initial values of body shape parameters to polygon coordinates Kb<SB>1</SB>-Kb<SB>5</SB>constituting a target homology model HMb generated with target values of the body shape parameters, to polygon coordinates Ki<SB>1</SB>-Ki<SB>5</SB>of a reference homology model HMi corresponding to the initial values of the body shape parameters to whereby generate polygon coordinates Ks<SB>1</SB>-Ks<SB>5</SB>of a simulation model HMs. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a three-dimensional shape model generation device that generates a three-dimensional shape model that is a sample of a three-dimensional shape, and more particularly to a three-dimensional shape model generation device that generates a three-dimensional shape model using statistical processing, and uses the same. The present invention relates to a three-dimensional shape model generation method, a generation program, and a generation system.

  2. Description of the Related Art Conventionally, as described in Patent Document 1, for example, a technique for modeling a body shape, which is a body shape, based on a measurable body physical value is known. In the technique described in Patent Document 1, a body shape is modeled by (a) homology processing, (b) statistical processing, and (c) body model generation processing for a large number of polygon models having different body shapes.

  (A) In the homologation process, first, the values of physical quantities such as body weight and perimeter length, and three-dimensional scan data expressing the body shape by point groups are measured for a number of different bodies. Next, normalization processing for the point group is performed on each three-dimensional scan data, and a plurality of homologous models which are polygon models in which the number of data points and the topology are normalized are generated. Then, a homologous model associated with the physical quantity value is generated for each body as a sample for statistical processing.

(B) In statistical processing, for example, a homology model is classified for each attribute of the human body that affects the body shape, such as race, birthplace, age, gender, etc., and a population composed of a plurality of homologous models is classified for each human attribute. Generated. Next, a distribution axis required for statistical processing, that is, a component that gives a strong correlation between homologous models is extracted for each population. For example, principal component analysis, which is one of multivariate analyses, is performed for each population, and characteristic meanings are extracted as distribution axes, that is, principal component axes, from the meanings of the body shape of the population. Then, the correlation between the vertex coordinates of the polygon vertices constituting the homologous model and the coordinates of the polygon vertices on the multidimensional space defined by the plurality of principal component axes, that is, the principal component values is calculated as the first eigenvector matrix. According to such statistical processing, as shown in the following formula (1), each vertex coordinate C _i (i is an integer of 2 or more) constituting the homologous model is a principal component value P _j (j is 2 or more). Integer) and the first eigenvector matrix D _ij .

[C _i ] = [D _ij ] × [P _j ] ... Formula (1)
In statistical processing, multivariate analysis is also performed on the physical quantity values. That is, components showing a strong correlation between the values of physical quantities are extracted for each population. Then, the correlation between the value of the physical quantity associated with the homologous model and the value of the physical quantity in a multidimensional space composed of a plurality of principal component axes, that is, the principal component value is calculated as the second eigenvector matrix. According to such statistical processing, as shown in the following formula (2), the physical quantity value x _i (i is an integer of 2 or more) associated with the homologous model is the principal component value S _j (j is 2). It is possible to calculate from the above integer) and the second eigenvector matrix H _ij .

[Xi] = [ _Hij ] * [ _Sj ] ... Formula (2)
(C) In the body model generation process, as shown in Expression (3), a physical quantity target value B _pi for generating a three-dimensional body model and an inverse matrix of the second eigenvector matrix H _ij are used. Thus, the principal component value S _j corresponding to the target value B _pi is calculated. Next, as shown in Expression (4), a function Tm that associates the principal component value P _j of the homologous model with the principal component value S _{j of the} physical quantity, and the principal component value S according to the target value B _pi of the physical quantity. _j is used to calculate the principal component value P _j of the homologous model corresponding to the target value B _pi . Then, as shown in the equation (5), the inverse matrix of the first eigenvector matrix D _ij obtained as a result of the homologous model statistical process and the principal component value P of the homologous model HMD corresponding to the target value B _{pi j} is used to calculate the vertex coordinates C _{i of the} polygon corresponding to the target value B _pi , that is, a homologous model corresponding to the target value B _pi is generated.

[S _j ] = [H _ij ] ⁻¹ × [B _pi ] (3)
[P _j ] = Tm (S _j ) (4)
[C _i ] = [D _ij ] ⁻¹ × [P _j ]] Equation (5)
According to the above-described series of processing, the principal component value P _j necessary for generating the homology model can be estimated based on the target value B _pi of the physical quantity, and thus the homology that is not included in the population. Even a model can be generated from a target value B _pi of a physical quantity. That is, a three-dimensional body model corresponding to the target value _Bpi of the physical quantity can be generated.

JP 2008-171074 A

By the way, the above-described body modeling technique generates a 3D body model itself from the result of statistical processing based on the body shape difference between individuals and the target value B _pi of the physical quantity, and reproduces the body shape of the individual. Therefore, if the population used for statistical processing or the target value B _{pi of the} physical quantity changes, whatever the characteristics of the specific individual on the body shape will eventually be lost in the end. . In other words, in the modeling technique described above, since the individual body shape is just one sample constituting the statistical population, characteristics of the body shape within a specific individual are difficult to be reflected in the calculation results. It has become.

  On the other hand, the demand for the above-described body modeling technique is not limited to merely generating a three-dimensional body model corresponding to the value of a physical quantity, but is expanding to the simulation of the future body shape of the user. For example, the above-mentioned body modeling technology is expected as a technology that embodies the user's future body shape by modeling it according to lifestyle and managing health and providing fashion based on the 3D body model. ing. Here, the future body shape of the user indicates a change in the body shape within a specific individual, and is generally greatly different depending on the latest body shape of the user such as flesh and the shape of a heel. Therefore, if the future body shape of the individual user is accurately reproduced, the latest body shape of the user is greatly reflected in the simulation result, that is, the features on the body shape within the specific individual are accurately reproduced. That is indispensable.

  However, with the modeling technique described above, where the individual's body shape is only one sample, the characteristics of the body shape within a specific individual are hardly reflected in the simulation, so that it is insufficient for expressing the future body shape of the user. End up. Among the errors included in the simulation results, errors that are difficult to occur in human anatomy may intervene with respect to the user's body shape. FIG. 26 is a diagram showing an example of the error as described above. FIG. 26A shows the output result of the latest three-dimensional model of the user, and FIG. 26B shows that the height, which is one of the physical quantities, is the same as the latest value and is one of the physical quantities. The output result of the future three-dimensional figure model obtained when the target value of a certain weight is lighter than the latest value is shown.

For example, when body weight, one of the physical quantities, decreases, it is natural that the body shape itself becomes slender, and it is rare that the current body shape differs greatly depending on whether the thinned part is the chest part or the abdominal part. Absent. In addition, the peripheral lengths of the narrowed portions decrease in relation to each other, for example, the amount of decrease in the length of the navel circumference tends to be smaller than the amount of decrease in the chest circumference, and these are often reduced. Furthermore, even if the weight is reduced, if the simulation target is an adult male, the size of the arm length Lh, the leg length Lc, and the shoulder width Bs does not vary greatly in terms of human anatomy. Is. However, in the modeling technique in which the three-dimensional body model itself is generated only from the results of statistical processing, as shown in FIG. However, the whole body shape may be uniformly thinned, and the arm length Lh, leg length Lc, and shoulder width Bs may expand and contract.

  In view of the fact that generating the three-dimensional body model corresponding to the target value Bpi is the purpose of the modeling technique in the first place, it is difficult for the characteristics on the body shape in a specific individual to be reflected in the simulation results. Naturally. In other words, the problem that the characteristics of the body shape within a specific individual are difficult to be reflected in the simulation results is a problem that arises for the first time from the above-mentioned new request to model the future body shape of the user. This is a problem that cannot be solved by the technology that reproduces the body shape of a specific individual based on the body shape difference.

  The present invention has been made in view of such a situation, and an object of the present invention is to provide a three-dimensional shape model generation apparatus and a three-dimensional shape model generation device that improve the accuracy of a new three-dimensional body model generated using statistical processing. A method, a three-dimensional shape model generation program, and a three-dimensional shape model generation system are provided.

Hereinafter, means for solving the above-described problems and the effects thereof will be described.
The invention according to claim 1 is characterized in that a reference three-dimensional model is generated from three-dimensional shape data of a plurality of modeling objects, and a plurality of feature values are extracted by multivariate analysis of the plurality of reference three-dimensional models. An extraction unit, a value of a physical quantity measurable by the modeled object, and a plurality of feature value values of the reference 3D model of the modeled object are associated with the reference 3D model. A storage unit for storing; a correlation calculation unit for calculating a correlation between the values of the plurality of feature amounts associated with the reference three-dimensional model and the values of the physical amount associated with the reference three-dimensional model; The value of the plurality of feature values is calculated from the input value of the physical quantity based on the relative relationship calculated by the relative relationship calculation unit, and the plurality of feature values calculated based on the data stored in the storage unit New 3D model from values A model generation unit including a model generation unit that generates an initial three-dimensional model generated using an initial value as the input value and a target value as the input value. The gist is to generate a new three-dimensional model by adding the difference from the target three-dimensional model generated in this way to the reference three-dimensional model corresponding to the initial value of the physical quantity.

  The invention according to claim 2 is characterized in that the storage unit stores a plurality of physical quantity values for each of the modeling objects, and the correlation calculation unit is associated with the reference three-dimensional model. A correlation between the value of the quantity and the values of the plurality of physical quantities associated with the reference three-dimensional model, further calculates a correlation between the values of the plurality of physical quantities, and the model generation unit Based on the correlation between the physical quantities, a value of another physical quantity different from the physical quantity is calculated from the input value of the physical quantity, and a new value is calculated using the value of the other physical quantity and the input value of the physical quantity that are the results of the calculation. The gist is to generate a three-dimensional model.

According to a third aspect of the present invention, the storage unit stores a plurality of physical quantity values including the length between joints for each modeled object, and the correlation calculation unit corresponds to the reference three-dimensional model. The correlation between the plurality of attached feature values and the plurality of physical quantity values associated with the reference three-dimensional model is calculated, and the model generation unit calculates the initial values of the plurality of physical quantities. The gist is to generate an initial three-dimensional model and generate the target three-dimensional model from target values of the plurality of physical quantities in which the length between the joints is the same as the initial value.

  The invention according to claim 4 generates a reference three-dimensional model from three-dimensional shape data of a plurality of modeling objects, extracts a plurality of feature amounts by multivariate analysis of the plurality of reference three-dimensional models, and A physical quantity value measurable by the modeling object and a plurality of feature value values of the reference three-dimensional model of the modeling object are stored in association with the reference three-dimensional model; A correlation between the values of the plurality of feature values associated with the three-dimensional model and the values of the physical quantities associated with the reference three-dimensional model is calculated, and the input value of the physical quantity is calculated based on the calculated relative relationship A three-dimensional shape model generation method for calculating a value of the plurality of feature amounts from the generated data and generating a new three-dimensional model from the calculated values of the plurality of feature amounts based on the stored data, The initial value for the input value The difference between the generated initial three-dimensional model and the target three-dimensional model generated using the target value as the input value is added to the reference three-dimensional model corresponding to the initial value of the physical quantity, thereby further adding a new three-dimensional model. The gist is to generate a model.

  The invention according to claim 5 stores a plurality of physical quantity values for each modeled object, the plurality of feature quantity values associated with the reference three-dimensional model, and the reference three-dimensional model. The correlation between the plurality of physical quantities associated with each other is calculated, the correlation between the plurality of physical quantities is calculated, and the physical quantity is different from the input value of the physical quantity based on the correlation between the plurality of physical quantities. The gist is that a value of another physical quantity is calculated, and a new three-dimensional model is generated using the value of the other physical quantity and the input value of the physical quantity as a result of the calculation.

  The invention according to claim 6 stores a plurality of physical quantity values including a length between joints for each modeled object, and a plurality of feature quantity values associated with the reference three-dimensional model; , Calculating a correlation with the values of the plurality of physical quantities associated with the reference three-dimensional model, generating the initial three-dimensional model from the initial values of the plurality of physical quantities, and calculating the length between the joints The gist is to generate the target three-dimensional model from the target values of the plurality of physical quantities that have the same value as the initial value.

  According to the seventh aspect of the present invention, the computer generates a reference three-dimensional model from three-dimensional shape data of a plurality of modeling objects, and extracts a plurality of feature quantities by multivariate analysis of the plurality of reference three-dimensional models. A feature quantity extraction unit that associates the value of a physical quantity measurable with the modeled object and the value of the plurality of feature quantities of the three-dimensional model of the modeled object with the reference 3D model Storing a storage unit, a correlation calculation unit for calculating a correlation between the values of the plurality of feature amounts associated with the reference three-dimensional model and the values of the physical amount associated with the reference three-dimensional model, The value of the plurality of feature values is calculated from the input value of the physical quantity based on the relative relationship calculated by the relative relationship calculation unit, and the value of the plurality of feature values calculated based on the data stored in the storage unit New tertiary from A three-dimensional shape model generation program that functions as a model generation unit that generates a model, wherein the model generation unit is configured to generate an initial three-dimensional model generated using an initial value as the input value, and a target value as the input value. The gist is to add a difference from the target three-dimensional model generated by use to the reference three-dimensional model corresponding to the initial value of the physical quantity so as to function as a generation unit that generates a new three-dimensional model.

  The invention according to claim 8 causes the storage unit to function as a storage unit that stores a plurality of physical quantity values for each modeled object, and associates the correlation calculation unit with the reference three-dimensional model. Functions as a calculation unit that calculates a correlation between the plurality of feature quantity values obtained and the plurality of physical quantity values associated with the reference three-dimensional model, and further calculates a correlation between the plurality of physical quantity values Then, the model generation unit calculates a value of another physical quantity different from the physical quantity from the input value of the physical quantity based on the correlation of the plurality of physical quantities, and the value of the other physical quantity that is a result of the calculation And a function to generate a new three-dimensional model using the input value of the physical quantity.

  The invention according to claim 9 causes the storage unit to function as a storage unit that stores a plurality of physical quantity values including the length between joints for each modeling target, and The model generation by functioning as a calculation unit that calculates the correlation between the values of the plurality of feature amounts associated with the reference three-dimensional model and the values of the plurality of physical amounts associated with the reference three-dimensional model Generating the initial three-dimensional model from the initial values of the plurality of physical quantities, and calculating the target three-dimensional model from the target values of the plurality of physical quantities in which the length between the joints is the same as the initial value. The gist is to make it function as a generating unit for generating.

  The invention according to claim 10 is a three-dimensional shape model generation system including a three-dimensional shape model generation device and a terminal device connected via a network, wherein the three-dimensional shape model generation device includes a plurality of models. A feature extraction unit that generates a reference three-dimensional model from three-dimensional shape data of a target object and extracts a plurality of feature values by multivariate analysis of the reference three-dimensional model, and can be measured with the target model A storage unit for storing the physical quantity value and the plurality of feature value values of the reference three-dimensional model of the modeled object in association with the reference three-dimensional model; and the reference three-dimensional model. Based on the correlation calculated by the correlation calculation unit that calculates the correlation between the value of the plurality of feature values associated with the value of the physical quantity associated with the reference three-dimensional model, and the relative relationship calculated by the relative relationship calculation unit. And generating a new three-dimensional model from the calculated feature values based on the data stored in the storage unit based on the data stored in the storage unit. A difference between an initial three-dimensional model generated using an initial value as the input value and a target three-dimensional model generated using a target value as the input value, and the physical quantity A new three-dimensional model is generated by adding to the reference three-dimensional model corresponding to the initial value of the three-dimensional model, and the terminal device receives the input value of the physical quantity and the three-dimensional shape data corresponding to the input value. A transmission unit that transmits to the original shape model generation device, a reception unit that receives the 3D model generated by the model generation unit from the 3D shape model generation device, and outputs the 3D model received by the reception unit And summarized in that and a that output section.

  The invention according to claim 11 is characterized in that the storage unit stores a plurality of physical quantity values for each modeled object, and the correlation calculation unit is associated with the reference three-dimensional model. A correlation between the value of the quantity and the values of the plurality of physical quantities associated with the reference three-dimensional model, further calculates a correlation between the values of the plurality of physical quantities, and the model generation unit Based on the correlation between the physical quantities, a value of another physical quantity different from the physical quantity is calculated from the input value of the physical quantity, and a new value is calculated using the value of the other physical quantity and the input value of the physical quantity that are the results of the calculation. The gist is to generate a three-dimensional model.

  In the invention according to claim 12, the storage unit stores a plurality of physical quantity values including the length between joints for each modeled object, and the correlation calculation unit corresponds to the reference three-dimensional model. The correlation between the plurality of attached feature values and the plurality of physical quantity values associated with the reference three-dimensional model is calculated, and the model generation unit calculates the initial values of the plurality of physical quantities. The gist is to generate an initial three-dimensional model and generate the target three-dimensional model from target values of the plurality of physical quantities in which the length between the joints is the same as the initial value.

  According to the first, fourth, seventh, and tenth aspects of the invention, an initial three-dimensional model is generated from the initial value of the physical quantity based on the statistical processing result, and the statistical processing result is also obtained. Based on the physical quantity target value, a target three-dimensional model is generated. Then, a difference between the initial three-dimensional model and the target three-dimensional model is added to the reference three-dimensional model corresponding to the initial value, thereby generating a new three-dimensional model.

Here, in the new three-dimensional model obtained from statistical processing, that is, the initial three-dimensional model and the target three-dimensional model, the characteristic part unique to one sample is removed and the characteristic part related to the entire population is extracted. The For this reason, a characteristic part unique to one sample is hardly reflected in the difference between the initial three-dimensional model and the target three-dimensional model. That is, a shape that cancels a characteristic part of the reference three-dimensional model is less likely to appear in the difference between the initial three-dimensional model and the target three-dimensional model. As a result, even if the difference between the initial 3D model and the target 3D model is added to the reference 3D model corresponding to the initial value of the physical quantity, the characteristic part of the reference 3D model corresponding to the initial value is newly added. It is directly reflected in a simple three-dimensional model. Therefore, compared to the case where the reference three-dimensional model corresponding to the initial value of the physical quantity is just one sample in statistical processing, the characteristic part of the reference three-dimensional model is more reflected in the new three-dimensional model. Will be. As a result, it is possible to improve the accuracy of new 3D models generated using statistical processing. For example, 3D models corresponding to the initial values of physical quantities contain errors that deviate from human anatomy. It's hard to get it.

  According to the inventions of claim 2, claim 5, claim 8, and claim 11, a new three-dimensional model is generated from a plurality of physical quantity values, so that a new three-dimensional model is generated from one physical quantity value. Compared to the case of generating a model, a new three-dimensional model can be generated with higher accuracy. The difference between the initial three-dimensional model and the target three-dimensional model is also higher accuracy. Therefore, even the difference added to the reference three-dimensional model has higher accuracy, and as a result, the accuracy of the new three-dimensional model generated using statistical processing can be further improved.

  Furthermore, based on the correlation between a plurality of physical quantities, values of other physical quantities are calculated from the input values of the physical quantities. Therefore, a large number of physical quantity values in accordance with the characteristics of the shapes of the plurality of modeling objects are obtained from a small number of physical quantity values. Therefore, it is possible to give high accuracy to the input values themselves of a plurality of physical quantities, and it is possible to further improve the accuracy of a new three-dimensional model generated using statistical processing.

  According to the invention described in claim 3, claim 6, claim 9, and claim 12, a new three-dimensional model is generated in such a manner that the length between joints of the reference three-dimensional model is fixed. . Here, it is assumed that the modeled object is, for example, a human or a pet. In these modeling objects, when the growth of the skeleton stops, the length between the joints is necessarily fixed. In this regard, if a new three-dimensional model is generated in a manner in which the length between joints is fixed as described above, a three-dimensional model more suitable for the shape change characteristics of the modeled object is newly added. Can be obtained as a simple three-dimensional model.

The block diagram which shows the structure of a three-dimensional shape model production | generation system. The functional block diagram which shows the structure of a user terminal. The block diagram which shows the data structure of initial model setting data. The figure which shows an example of a body landmark with a three-dimensional body model. The block diagram which shows the structure of the target body model data in 1st embodiment. The functional block diagram which shows the structure of the body model generation server in 1st embodiment. The block diagram which shows the data structure of a homology model. The figure which shows the part utilized for a statistical process in a homology model. (A) (b) The distribution map which shows the process of the principal component analysis which is an example of multivariate analysis. (A) (b) The figure which shows the production | generation process of a simulation model. The flowchart which shows the processing content of the simulation of a body model. The flowchart which shows the flow of a homology model production | generation process. The flowchart which shows the flow of a statistical process. The flowchart which shows the flow of the simulation model production | generation process in 1st embodiment. The figure which shows the output result of the body shape simulation in 1st embodiment as a three-dimensional image. The block diagram which shows the structure of the target body model data in 2nd embodiment. The functional block diagram which shows the structure of the body shape model production | generation server in 2nd embodiment. The flowchart which shows the flow of the simulation model production | generation process in 2nd embodiment. (A) (b) The figure which shows the output result of the body shape simulation in 2nd embodiment as a three-dimensional image. The block diagram which shows the structure of the target body model data in 3rd embodiment. The figure which shows a joint position and a model skeleton with a homologous model. The functional block diagram which shows the structure of the body shape model production | generation server in 3rd embodiment. The block diagram which shows the data structure of joint position data. The flowchart which shows the flow of the homology model production | generation process in 3rd embodiment. The flowchart which shows the flow of the simulation model production | generation process in 3rd embodiment. (A) (b) The figure which shows the output image of the body model produced | generated by simulation of the conventional body model.

(First embodiment)
Hereinafter, a first embodiment of the present invention will be described with reference to FIGS. First, the configuration of the three-dimensional shape model generation system according to the present embodiment will be described with reference to FIGS. FIG. 1 is a configuration diagram showing a configuration of a three-dimensional shape model generation system, and FIG. 2 is a functional block diagram showing a configuration of a user terminal constituting the three-dimensional shape model generation system.

[3D shape model generation system]
As shown in FIG. 1, a body model generation server 1A, which is an example of a three-dimensional shape model generation apparatus, is connected to a user terminal 1B, which is a terminal device used by a user, with a communication network NT such as the Internet, a LAN, and a WAN. Connected through. Various data for simulating the future shape of the human body that is the modeling object, that is, the future shape as a three-dimensional shape model, is transmitted to the user via communication between the body shape model generation server 1A and the user terminal 1B. It is transmitted from the terminal 1B to the body model generation server 1A. Various data generated by the body model generation server 1A is transmitted from the body model generation server 1A to the user terminal 1B. The user terminal 1B is installed in a place corresponding to the use of the three-dimensional body model, such as a terminal device installed in a fitness club, a health equipment manufacturer, a housing manufacturer, a clothing manufacturer, or a food manufacturer. A plurality of body model generation servers 1A are connected.

  As shown in FIG. 2, the user terminal 1B is provided with a data processing unit 11 including various processors such as a CPU and a DMA controller. The data processing unit 11 reads, as data, a program for executing transmission and reception of various data used for the simulation of the three-dimensional body model, interprets it as an instruction, and executes it. The data processing unit 11 supervises, for example, storage processing, transmission processing, reception processing, output processing, and the like of various data used for the simulation of the three-dimensional body model according to the above program.

In addition, the user terminal 1B includes a transmission unit for communicating with the body model generation server 1A and a transmission / reception processing unit 12 as a reception unit. The user terminal 1B is provided with an input processing unit 13 including a three-dimensional scanner 13a and an interface between the three-dimensional scanner 13a and a computer. The input processing unit 13 inputs various data for executing the simulation of the three-dimensional body model to the data processing unit 11. Furthermore, the user terminal 1B is provided with a storage unit 14 composed of a hard disk drive, various memories, and the like. The storage unit 14 stores a user program 14d, which is an example of a program for executing a simulation of a three-dimensional body model, and various data input from the input processing unit 13. Furthermore, the user terminal 1B is provided with a body model output processing unit 15 as an output unit composed of an image processor such as a GPU and various displays. The body model output processing unit 15 outputs and displays the result of the simulation as a 3D image using the 3D body model received by the transmission / reception processing unit 12.

  Prior to the simulation of the three-dimensional body model, first, initial model setting data 14a, landmark data 14b, and target body model data 14c, which are prerequisites for the simulation, are temporarily stored in the storage area of the storage unit 14. Next, a request for starting the simulation is transmitted from the transmission / reception processing unit 12 to the body model generation server 1A together with the identifier ID for identifying the user. Further, the various data once stored in the storage area of the storage unit 14 are sequentially transmitted from the transmission / reception processing unit 12 to the body model generation server 1A. Then, a simulation based on the above-described various data is executed by the body model generation server 1A. The simulation result generated by the body model generation server 1A is transmitted from the body model generation server 1A to the transmission / reception processing unit 12, and then output as a three-dimensional image by the body model output processing unit 15.

  3 to 5 are diagrams showing data configurations of various data transmitted from the user terminal 1B to the body model generation server 1A. FIG. 3 is a block diagram showing the data structure of the initial model setting data, FIG. 4 is a view showing an example of a body landmark constituting the landmark data together with a three-dimensional body model, and FIG. 5 is a target body model data 14c. It is a block diagram which shows the data structure of these.

  As shown in FIG. 3, the initial model setting data 14a includes a header 141a that is data relating to the initial model setting data 14a itself, three-dimensional scan data 142a that is three-dimensional shape data, and incidental data 143a. Yes.

  The three-dimensional scan data 142a is data for executing a simulation of a three-dimensional body model based on the latest body shape of the user. The three-dimensional scan data 142a is measurement data obtained by measuring the latest body surface of the user with the three-dimensional scanner 13a, that is, three-dimensional shape data expressing the latest body shape of the user as a point cloud.

  When the coordinate data of the point group is included in the three-dimensional scan data 142a, the simulation of the three-dimensional body model is executed based on the latest body shape of the user. On the other hand, when the coordinate data of the point group of the three-dimensional scan data 142a is blank, the simulation of the three-dimensional body model is executed based on a body shape different from the latest body shape of the user.

The incidental data 143a is composed of human body attributes that affect the body shape and values of body shape parameters that serve as body shape indices. The attributes of the human body that influence the body shape include biological data such as race, age, and sex, and further data related to the living environment such as occupation, exercise history, and medical history. The body parameters used as the body shape index are height, weight, body mass index (= “weight” / “height” 2: BMI), circumference (neck circumference, chest circumference, waist circumference, navel circumference, upper arm circumference, calf circumference) , Around thighs, etc.). That is, the body shape parameter is a physical quantity that can be measured by the human body to be modeled. Further, the value of the body shape parameter is, for example, the latest measured value measured when acquiring the three-dimensional scan data 142a or before acquiring the three-dimensional scan data 142a. That is, the value of the body parameter constituting the incidental data 143a is a value that is an index of the latest body shape of the user, and is an initial value B _i1 to initial value B that is one of the input values of the simulation of the three-dimensional body model. _Treated as _ih .

  The landmark data 14b is data indicating three-dimensional coordinates of landmarks that are anatomical feature points. The landmark data 14b is used to extract vertices corresponding to the landmark from the point group constituting the three-dimensional scan data 142a. For example, as shown in FIG. 4, the landmark data 14b is data indicating the three-dimensional coordinates of the cervical point Pr1, the acromion point Pr2, the nipple point Pr3, the trochanter point Pr4, the tibial point Pr5, and the inferior causal point Pr6. It is. In order to extract the vertices corresponding to the cervical point Pr1, shoulder point Pr2, nipple point Pr3, trochanter point Pr4, tibial point Pr5, and epicarpal point Pr6 from the point group constituting the three-dimensional scan data 142a. The landmark data 14b is used.

As shown in FIG. 5, the target body model data 14c includes a header 141c that is data related to the target body model data 14c itself, and target body parameter data 142c used for simulation of the three-dimensional body model. . The target body parameter data 142c is the value of the body parameter set according to the purpose of the simulation. That is, the value of the body parameter constituting the target body model data 14c is a value that serves as an index of the future body shape of the user, and is one of the input values of the simulation of the three-dimensional body model, the target value B _p1 to the target. _Treated as value B _ph .

[Body Shape Generation Server 1A]
Next, the configuration of the body model generation server 1A will be described with reference to FIGS. FIG. 6 is a functional block diagram showing the configuration of the body model generation server, and FIG. 7 is a configuration diagram showing the data configuration of the homologous model. FIG. 8 is a diagram illustrating an extracted part of a homologous model used for statistical processing, and FIGS. 9A and 9B are distribution diagrams illustrating the contents of processing in the statistical processing. FIGS. 10A and 10B are diagrams for explaining the processing contents of the simulation of the three-dimensional model using a part of each homologous model.

  As shown in FIG. 6, the body model generation server 1 </ b> A includes a model generation processing control unit 21 configured with a CPU and various memories. The model generation processing control unit 21 reads out a simulation program for executing the simulation of the three-dimensional body model as data, interprets it as an instruction, and executes it. The body model generation server 1A is provided with a transmission / reception processing unit 22 for communicating with the user terminal 1B via the communication network NT. The body model generation server 1A is provided with a data processing unit 23 composed of a part of the CPU and various memories. The data processing unit 23 executes processing using various data in response to a command from the model generation processing control unit 21. Furthermore, the body model generation server 1A is provided with a storage unit 24 composed of a hard disk drive and various memories. The storage unit 24 stores a simulation program of the three-dimensional body model, and stores the results processed by the data processing unit 23.

  Then, the model generation processing control unit 21 controls various processes in the transmission / reception processing unit 22, the data processing unit 23, and the storage unit 24 according to the simulation program. In the body model generation server 1A, the model generation processing control unit 21, the transmission / reception processing unit 22, and the data processing unit 23 may be configured as an application server, and the storage unit 24 may be configured as a separate database server.

The data processing unit 23 includes a scan data processing unit 23a, a homologous model generation unit 23b constituting a feature amount extraction unit, a statistical processing unit 23c constituting a feature amount extraction unit and a correlation calculation unit, and a model generation unit 23d. It is configured. In the storage area of the storage unit 24, a simulation program 24p, a preprocessing model database 24a, a standard model database 24b, a homologous model database 24c, a statistical processing result database 24d, and a body model database 24e are constructed. . These databases are managed by each of the data processing units 23 that output data stored in each database.

  The scan data processing unit 23a uses the three-dimensional scan data 142a included in the initial model setting data 14a, and executes various pre-processing on the three-dimensional scan data 142a for executing a simulation of the three-dimensional body model. . More specifically, the scan data processing unit 23a uses the three-dimensional scan data 142a and the landmark data 14b to rotate, move, extract, and further detect the landmarks that constitute the three-dimensional scan data 142a. Associate and execute them to generate a preprocessing model. Then, the scan data processing unit 23a updates the preprocessing model database 24a in such a manner that the preprocessing model and the accompanying data 143a are linked to the identifier ID of the user.

  The homologous model generation unit 23b reads the preprocessing model associated with the user identifier ID from the preprocessing model database 24a, and obtains a polygon model in which the score, topology, data format, etc. are standardized, that is, as a reference three-dimensional model. The homology model HMD is generated from the pre-processing model. That is, the homology model generation unit 23b executes homology processing of the preprocessing model.

As shown in FIG. 7, the homologous model HMD is composed of a header 241c that is data relating to the homologous model HMD itself and body coordinate data 242c. The body coordinate data 242c includes polygon vertex K ₁ to polygon vertex K _n (n is an integer of 2 or more), vertex coordinates C ₁ to vertex coordinates C _n (n is an integer of 2 or more), and their polygon values. It is composed of a face F ₁ to a polygon face F _m (m is an integer of 2 or more) and a vertex array E ₁ to a vertex array E _m (m is an integer of 2 or more) that is an array of polygon vertices constituting them. .

  Then, the homologous model generation unit 23b updates the homologous model database 24c in such a manner that the homologous model HMD and the incidental data 143a are associated with the identifier ID of the user. That is, the homologous model generation unit 23b includes the attribute of the human body related to the user, the value of the latest body shape parameter of the user, and the homologous model HMD indicating the latest body shape of the user, which are linked to the identifier ID of the user. The homology model database 24c is updated in an attached manner. By such processing, a population composed of a number of homologous models HMD is configured for each human body attribute.

  The homologous model generation unit 23b uses the standard model received by the transmission / reception processing unit 22, and updates the standard model database 24b in a manner in which the standard model is associated with the attributes of the human body related to the standard model. The standard model is a polygon model indicating the body surface of the human body, and is a model in which the score, topology, and data format are standardized in advance, similar to the homologous model HMD. This standard model is prepared for each human attribute.

Further, as shown in FIG. 8, the homologous model generation unit 23b extracts the polygon model corresponding to the head of the human body from the homologous model HMD as the first synthesis target Ex1, and also extracts the part corresponding to both hands of the human body. The polygon model is extracted from the homologous model HMD as the second synthesis target Ex2. Further, the homologous model generation unit 23b extracts a polygon model corresponding to both legs of the human body as the third synthesis target Ex3 from the homologous model HMD, and extracts these first synthesis target Ex1 to the third synthesis target Ex3 from the homologous model HMD. The remaining part is extracted from the homology model HMD as a simulation target HMp. In the first synthesis target Ex1 to the third synthesis target Ex3, the simulation model HMs generated using the simulation target HMp is synthesized with portions corresponding to the user's human head, both human hands, and both human legs. It is used when it is done.

  Then, the homologous model generation unit 23b updates the homologous model database 24c in such a manner that the first synthesis target Ex1 to the third synthesis target Ex3 and the simulation target HMp are associated with the identifier ID of the user. By such processing, a population composed of a large number of simulation target HMp is configured for each attribute of the human body, like the population of the homologous model HMD.

  In addition, the homologous model generation unit 23b extracts a polygon model corresponding to the head of the human body from the standard model as the first synthesis target Ex1, and a polygon model corresponding to both hands of the human body as the second synthesis target Ex2. Extract from Further, the homologous model generation unit 23b extracts the polygon model corresponding to both feet of the human body as the third synthesis target Ex3, and extracts the remaining portions obtained by extracting the first synthesis target Ex1 to the third synthesis target Ex3 from the standard model. Extract as HMp. The first synthesis target Ex1 to the third synthesis target Ex3 are synthesized with the simulation model HMs generated using the simulation target HMp, with portions corresponding to a standard human head, both human hands, and both human legs. It is used when it is done. Then, the homologous model generation unit 23b updates the standard model database 24b in such a manner that the first synthesis target Ex1 to the third synthesis target Ex3 and the simulation target HMp are associated with the identifier ID of the user.

  When the simulation of the three-dimensional figure model is executed based on the user's homologous model HMD, that is, when the simulation is executed based on the latest figure of the user, the simulation target HMp in the homologous model database 24c is the third order. It is used as a reference homology model HMi that is an initial value of the original body model. On the other hand, when the simulation of the 3D body model is executed based on a 3D body model different from the user's homologous model HMD, the simulation target HMp in the standard model database 24b is the initial value of the 3D body model. It is used as a certain reference homology model HMi.

  The statistical processing unit 23c performs multivariate analysis using each simulation target HMp as a sample for each population, that is, for each human body attribute. More specifically, the statistical processing unit 23c reads each simulation target HMp constituting one population from the homologous model database 24c. Then, a component having a strong correlation between the simulation target HMp is extracted as a feature amount.

  For example, as shown in FIG. 9A, in the real space composed of the coordinate axes A, B, and C, the polygon vertices constituting each simulation target HMp are distributed in a state of weak correlation with respect to the coordinate axes A and B. is doing. The statistical processing unit 23c reads each simulation target HMp constituting such a population, and executes principal component analysis which is one of multivariate analyses. Then, as shown in FIG. 9B, a principal component having a strong correlation among the simulation target HMp is extracted. In other words, the principal component analysis, which is one of the multivariate analyses, is performed, and the characteristic meanings such as large size and slenderness among the meanings of the whole population are converted into the principal component axis S, the principal component axis T, and Extracted as principal component axis U. The number of principal components extracted by principal component analysis is not limited to three as illustrated in FIG. 9, but may be two, or four or more.

Then, the statistical processing unit 23c calculates the peripheral form related to the population by calculating the center point in the extracted principal component axis and changing the value for each principal component axis. The calculated peripheral form can be specified as a coordinate value in a coordinate space composed of a plurality of principal component axes, that is, a displacement amount (principal component value) from the center point. Then, the statistical processing unit 23c performs the simulation target HMp.
The correspondence relationship between the vertex value X ₁ to vertex value X _k (k is an integer of 2 or more) and the principal component value P ₁ to principal component value P _j (j is an integer of 2 or more) The calculation is performed as a relationship as shown in Expression (6), that is, as an eigenvector matrix A ₁ . In addition, the statistical processing unit 23c updates the homology model database 24c in a manner in which each simulation target HMp and the principal component values P ₁ to P _{j of the} simulation target HMp are linked. The statistical processing unit 23c includes a body attributes and eigenvector matrix A ₁ Field of the simulation object HMp used in the statistical processing to update the statistical processing result database 24d in a manner to be linked.

The statistical processing unit 23c executes a multiple regression process using the principal component values and body shape parameters as samples for each population, that is, for each human body attribute. More specifically, the statistical processing unit 23c sets the initial values B _i1 to the initial values B _{ih of the} body parameters related to the simulation target HMp and the simulation target HMp for all the simulation target HMp constituting one population. The corresponding principal component value P ₁ to principal component value P _j are read out from the homology model database 24c. The statistical processing unit 23c is configured so that the initial value B _i1 to the initial value B _{ih of} each body shape parameter is an objective variable (explained variable) and the principal component value P ₁ to the principal component value P _j are explanatory variables. A multiple regression process is executed every time. In this multiple regression process, the statistical processing unit 23c indicates the correlation between the initial values B _i1 to B _ih and the principal component values P ₁ to P _{j of} each body parameter as shown in Expression (7). Do relationship, that is, calculated as the correlation matrix A _2. The statistical processing unit 23c includes a body attributes were subject to statistical processing and the correlation matrix A ₂ updates the statistical processing result database 24d in a manner to be linked.

The model generation unit 23d reads the incidental data 143a associated with the user identifier ID from the homologous model database 24c. In addition, the model generation unit 23d extracts the initial value B _i1 to the initial value B _ih of the body shape parameters from the read incidental data 143a. Then, the model generation unit 23d uses the extracted initial value B _i1 to initial value B _{ih of} the body shape parameter and the inverse matrix of the correlation matrix A2 (see Expression (7)), and uses the initial value B _i1 to the initial value B _ih. The principal component values corresponding to are inversely calculated. Then, the model generation unit 23d uses the principal component value, which is the calculation result, as the initial principal component value P _i1 to the initial principal component value P _ij .

In addition, the model generation unit 23d extracts the body shape parameter target values B _p1 to B _ph from the target body model data 14c. The model generation unit 23d uses the extracted target value B _p1 to target value B _{ph of} the body shape parameter and the inverse matrix of the correlation matrix A ₂ (see Expression (7)), and uses the target value B _p1 to the target value B. Each principal component value corresponding to _ph is calculated. Then, the model generation unit 23d uses the principal component values that are the calculation results as the target principal component value P _p1 to the target principal component value P _pj .

The model generation unit 23d uses the initial principal component value P _i1 to the initial principal component value P _ij and the inverse matrix of the eigenvector matrix A ₁ (see Expression (6)), and uses the initial principal component value P _i1 to the initial component value. A polygon model corresponding to the principal component value P _ij is calculated. The model generation unit 23d uses the polygon model as an initial three-dimensional model (initial homologous model HMa).

The model generation unit 23d uses the target principal component value P _p1 to the target principal component value P _pj and the inverse matrix of the eigenvector matrix A ₁ (see Expression (6)), and uses the target principal component value P _p1 to the target A polygon model corresponding to the principal component value P _pj is calculated. The model generation unit 23d uses the polygon model as a target three-dimensional model (target homologous model HMb).

Further, the model generation unit 23d uses the initial homologous model HMa and the target homologous model HMb, and calculates a displacement amount from each polygon coordinate constituting the initial homologous model HMa to each polygon coordinate constituting the target homologous model HMb. Calculate as The model generating unit 23d, a displacement amount T _n is a calculation result using the reference homology model HMi is the initial value of the three-dimensional body shape model, based homology model HMi a polygon model obtained by adding the displacement amount T _n Calculation is performed as a simulation model HMs. And the model production | generation part 23d updates the body model database 24e in the aspect with which this simulation model HMs and user identifier ID are linked | related.

For example, as shown in FIG. 10 (a), the model generating unit 23d, a polygon coordinate Kb ₁ ~ polygon coordinates Kb constituting the target homology model HMb polygon coordinate Ka ₁ ~ polygon coordinate Ka ₅ constituting the initial homology model HMa displacement amount to _{5 T 1} ~ calculates the displacement amount _{T 5.} The model generating unit 23d, the operation result displacement amount _{T 1} ~ displacement _{T 5} and using a reference homology model HMi is, polygon coordinate Ki ₁ ~ polygon coordinates Ki ₅ a displacement amount _{T 1} constituting the reference homology model HMi The polygon coordinates Ks ₁ to the polygon coordinates Ks ₅ displaced by the displacement amount T ₅ are calculated as simulation models HMs. Further, the model generation unit 23d updates the body model database 24e in such a manner that the simulation model HMs and the identifier ID of the user are associated with each other.

  The model generation unit 23d reads the first synthesis target Ex1 to the third synthesis target Ex3 associated with the user identifier ID from the homologous model database 24c or the standard model database 24b. The model generation unit 23d reads the simulation model HMs associated with the user identifier ID from the body model database 24e. Further, the model generation unit 23d synthesizes the read first synthesis target Ex1 to the third synthesis target Ex3 and the read simulation model HMs, a part corresponding to the head of the human body, a part corresponding to both hands of the human body, and the human body A homologous model HMD composed of a portion corresponding to both feet and a simulation model HMs is generated. Then, the model generation unit 23d updates the body model database 24e in such a manner that the combined simulation model HMs and the user identifier ID are linked.

Next, the flow of simulation of a three-dimensional body model using the three-dimensional shape model generation system will be described with reference to FIGS. FIG. 11 is a flowchart showing the processing contents of the body model simulation, and FIG. 12 is a flowchart showing the flow of the homologous model generation processing. FIG. 13 is a flowchart showing the flow of statistical processing, and FIG. 14 is a flowchart showing the flow of simulation model generation processing. FIG. 15 is a diagram showing output results of the reference homology model HMi and the simulation model HMs displayed on the input / output screen of the user terminal.

  Prior to the execution of the simulation, the homologous model database 24c includes a population composed of a large number of individual homologous models HMD in advance for each attribute of the human body and a mother composed of a number of different simulation targets HMp. A group is composed of human attributes.

  First, a request for executing the body model simulation is transmitted from the user terminal 1B to the body model generation server 1A. The initial model setting data 14a, landmark data 14b, target body model data 14c, and standard model are also transmitted from the user terminal 1B to the body model generation server 1A. When the body model generation server 1A receives the request to execute the body model simulation and the various data, the model generation processing control unit 21 that supervises the processing of the processing units 23a, 23b, 23c, and 23d performs a simulation program. The simulation is executed as follows.

  That is, as shown in FIG. 11, whether or not the user's three-dimensional scan data 142a can be acquired in accordance with the request, in detail, whether the coordinate data of the point group is included in the three-dimensional scan data 142a. Determination of whether or not is performed by the model generation processing control unit 21 (step S11). Depending on whether or not the three-dimensional scan data 142a can be acquired, scan data processing (step S12), homologous model generation processing (step S20), statistical processing (step S30), and simulation model generation processing (step S40). The simulation model synthesis process (step S50) and the simulation model output process (step S60) are sequentially executed with different contents.

[Scan data processing]
When the model generation processing control unit 21 determines that the 3D scan data 142a can be acquired (step S11: YES), the scan data processing unit 23a executes various pre-processing for the 3D scan data 142a. To generate a pre-processing model. In addition, the scan data processing unit 23a updates the preprocessing model database 24a in such a manner that the preprocessing model and the incidental data 143a are linked to the identifier ID of the user, and ends the scan data processing in the simulation (step S12). ).

[Homologous model generation processing]
When the preprocess model is stored in the preprocess model database 24a, the homologous model generation unit 23b connects the standard model received from the user terminal 1B and the human body attribute related to the standard model in a manner associated with the standard model database 24b. Update.

When the standard model database 24b is updated, as shown in FIG. 12, the homologous model generation unit 23b obtains the preprocess model linked to the user identifier ID and the accompanying data 143a from the preprocess model database 24a. read out. The homologous model generation unit 23b extracts human body attributes and initial values B _i1 to B _{ih of} body parameters from the read incidental data 143a. Further, the homologous model generation unit 23b reads the standard model associated with the human body attribute from the standard model database 24b using the extracted human body attribute.

Then, the homologous model generation unit 23b generates a homologous model HMD including the polygon coordinates by displacing the polygon coordinates of the standard model so that the body surface indicated by the read preprocessing model matches the body surface indicated by the standard model. That is, the homogenization processing of the preprocessing model is executed. When the user's homologous model HMD is generated, the homologous model generating unit 23b relates to the homologous model HMD, initial values B _i1 to initial values B _{ih of} body parameters related to the homologous model HMD, and further to the homologous model HMD. The homology model database 24c is updated in such a manner that the attributes of the human body and the identifiers of the human body are associated with the user ID (step S21).

  When the homologous model HMD is stored in the homologous model database 24c, the homologous model generating unit 23b reads out the homologous model HMD linked to the user identifier ID from the homologous model database 24c, and from the homologous model HMD, The synthesis target Ex1 to the third synthesis target Ex3 are extracted (step S22). The homologous model generation unit 23b extracts from the homologous model HMD the simulation target HMp that is the remaining part obtained by removing the first synthesis target Ex1 to the third synthesis target Ex3 from the homologous model HMD (step S23). Then, the homologous model generation unit 23b updates the homologous model database 24c in such a manner that the first synthesis target Ex1 to the third synthesis target Ex3 and the simulation target HMp are linked to the identifier ID of the user, and generates a homologous model for the simulation. The process ends (step S24).

  If the model generation processing control unit 21 determines that the 3D scan data 142a cannot be acquired (step S11: NO), the homologous model generation unit 23b is associated with the identifier ID of the user. The attributes of the human body included in the incidental data 143a are read from the preprocessing model database 24a. Then, the homologous model generation unit 23b reads out the standard model associated with the attribute of the human body from the standard model database 24b. Next, the homologous model generation unit 23b extracts the first synthesis target Ex1 to the third synthesis target Ex3 from the read standard model (step S25). Further, the homologous model generation unit 23b extracts, from the standard model, the simulation target HMp, which is the remaining part obtained by removing the first synthesis target Ex1 to the third synthesis target Ex3 from the read standard model (step S26). Then, the homologous model generation unit 23b updates the standard model database 24b in such a manner that the first synthesis target Ex1 to the third synthesis target Ex3 and the simulation target HMp are linked to the identifier ID of the user, and generates a homologous model for the simulation. The process ends (step S27).

[Statistical processing]
When the homologous model generation process ends, the statistical processing unit 23c reads the auxiliary data 143a associated with the user identifier ID from the homologous model database 24c, and extracts the attributes of the human body included in the auxiliary data 143a. Next, the statistical processing unit 23c reads a plurality of simulation target HMp associated with the same attribute as the extracted human body attribute from the homologous model database 24c. That is, the statistical processing unit 23c reads a plurality of simulation targets HMp in a manner that includes the simulation target HMp indicating the latest body shape of the user.

Then, the statistical processing unit 23c performs principal component analysis on the plurality of read simulation target HMp, and the principal component that is the analysis result, the principal component value P ₁ to the principal component value P _{j of} each simulation target HMp, calculating the said eigenvector matrix _{a 1} (step S31). Furthermore, the statistical processing unit 23c reads from the homology model database 24c the initial values B _i1 to the initial values B _ih of the plurality of body shape parameters associated with the same attribute as the extracted human body attribute. Then, the statistical processing unit 23c calculates a correlation matrix A2 by executing a multiple regression process in which the read initial values B _i1 to B _{ih of} the body shape parameters are treated as objective variables and the principal component values as explanatory variables (step) 32).

The statistical processing unit 23c updates the homologous model database 24c in a manner in which each simulation target HMp and the principal component values P ₁ to P _{j of the} simulation target HMp are linked. The statistical processing section 23c, the human body attribute to be processed for statistical processing, the eigenvector matrix of A ₁ and the correlation matrix A ₂ updates the statistical processing result database 24d in a manner to be linked, statistics for the user The process ends (step S33).

[Simulation model generation processing]
When the statistical process ends, the model generation unit 23d reads the simulation target HMp linked to the user identifier ID from the homology model database 24c and sets the simulation target HMp as the reference homology model HMi. If the model generation processing control unit 21 determines that the three-dimensional scan data 142a cannot be acquired, the model generation unit 23d displays the simulation target HMp associated with the user identifier ID as the standard model database 24b. And the simulation target HMp is set as the reference homology model HMi (step S41).

Next, the model generation unit 23d reads the incidental data 143a associated with the user identifier ID from the homologous model database 24c, and extracts the initial values B _i1 to B _ih of the body shape parameters included in the incidental data 143a. . In addition, the model generation unit 23d extracts the target value B _p1 to the target value B _ph of the body shape parameter from the target body model data 14c received from the user terminal 1B (Step S42).

When the initial value B _i1 to the initial value B _ih of the body shape parameter and the target value B _p1 to the target value B _{ph of the} body shape parameter are extracted, the model generation unit 23d extracts the attributes of the human body included in the read incidental data 143a. The correlation matrix A2 associated with the attribute of the human body is read from the statistical processing result database 24d.

Then, the model generation unit 23d uses the initial value B _i1 to the initial value B _ih of the body shape parameter and the inverse matrix of the read correlation matrix A ₂ to perform the inverse operation of the above equation (7), thereby performing the initial principal component. A value P _i1 to an initial principal component value P _ij are calculated (step S43). Further, the model generation unit 23d uses the target parameter B _p1 to the target value B _ph of the body shape parameter and the inverse matrix of the read correlation matrix A ₂ to execute the inverse calculation of the above equation (7), thereby executing the target principal component. A value P _p1 to a target principal component value P _pj are calculated (step S44).

When the initial principal component value P _i1 to the initial principal component value P _ij and the target principal component value P _p1 to the target principal component value P _pj are calculated, the model generation unit 23d has the eigenvector matrix A ₁ associated with the attribute of the human body. Are read from the statistical processing result database 24d.

Then, the model generation unit 23d uses the initial principal component value P _i1 to the initial principal component value P _ij and the inverse matrix of the read eigenvector matrix A ₁ to execute the inverse operation of the above equation (6), thereby performing the polygon vertex The coordinate value of is calculated. By such processing, a homology model HMD corresponding to the initial value B _i1 to the initial value B _ih of the body shape parameter, that is, the initial homology model HMa is generated (step S45). Further, the model generation unit 23d uses the target principal component value P _p1 to the target principal component value P _pj and the inverse matrix of the read eigenvector matrix A ₁ to perform the inverse operation of the above equation (6), thereby executing the polygon vertex The coordinate value of is calculated. Through such processing, a homology model HMD corresponding to the body parameter target value B _p1 to target value B _ph , that is, the target homology model HMb is generated (step S46).

When the initial homologous model HMa and the target homologous model HMb are generated, the model generation unit 23d uses the initial homologous model HMa and the target homologous model HMb to generate the target homologous model HMb from each polygon coordinate constituting the initial homologous model HMa. A displacement amount T _n (n is an integer equal to or greater than 2) is calculated for each of the polygon coordinates constituting the same. Then the model generating unit 23d, the operation result and the displacement amount T _n is, three-dimensional body shape model initial value a is used and the reference homologous model HMi, reference homology model HMi polygon model plus displacement amount T _n in the Simulation model HM
Calculate as s. Then, the model generation unit 23d updates the body model database 24e in a manner in which the simulation model HMs and the user identifier ID are linked, and ends the simulation model generation process for the user.

  Here, in the homologous model HMD obtained from the statistical processing as described above, that is, the initial homologous model HMa and the target homologous model HMb, characteristic parts of the body shape such as flesh and wrinkles unique to the user are removed, and the entire population A characteristic part related to the is extracted. Therefore, it is difficult for the difference between the initial homology model HMa and the target homology model HMb to reflect a characteristic part unique to the user. That is, a shape that cancels a characteristic part of the reference homology model HMi is unlikely to appear in the difference between the initial homology model HMa and the target homology model HMb. Therefore, even if the difference between the initial homology model HMa and the target homology model HMb is added to the reference homology model HMi, characteristic features unique to the user such as flesh and wrinkles in the reference homology model HMi are generally included in the simulation model HMs. Will remain.

Therefore, compared to the case where the reference homology model HMi is only one sample in statistical processing, the characteristic part of the reference homology model HMi is directly reflected in the simulation model HMs. In other words, the shape of the body unique to the user is directly reflected in the simulation result. As a result, the accuracy of the simulation model HMs generated using statistical processing can be improved. For example, the human body structure is compared with the three-dimensional model corresponding to the target value B _p1 to the target value B _ph of the body shape parameter. Academic errors are difficult to include.

[Simulation model synthesis processing]
When the simulation model generation process ends, the model generation unit 23d reads the first synthesis target Ex1 to the third synthesis target Ex3 linked to the user identifier ID from the homologous model database 24c. Then, the model generation unit 23d combines the read first synthesis target Ex1 to third synthesis target Ex3 and the simulation model HMs (step S50). By such processing, a homologous model HMD composed of a portion corresponding to the head of the human body, a portion corresponding to both hands of the human body, a portion corresponding to both feet of the human body, and the simulation model HMs is generated.

  Here, the part corresponding to the head of the human body, the part corresponding to both hands of the human body, and the part corresponding to both feet of the human body are the torso of the human body with respect to body parameters such as height, weight, body mass index, and perimeter Generally, it is less likely to fluctuate compared to the abdomen. Therefore, even if the first synthesis target Ex1 to the third synthesis target Ex3 are not included in the sample as described above, the statistical processing is performed on the simulation target HMp including the trunk and the abdomen. Therefore, the accuracy of the result of the processing is sufficiently secured. Therefore, according to the statistical processing and the simulation model synthesis processing as described above, it is possible to reduce the data amount of the sample used for the statistical processing while suppressing loss of accuracy of the result of the statistical processing. .

  If the model generation processing control unit 21 determines that the three-dimensional scan data 142a cannot be acquired (step S11: NO), the model generation unit 23d stores the first ID associated with the user identifier ID. The first synthesis target Ex1 to the third synthesis target Ex3 are read from the standard model database 24b. Then, the model generation unit 23d combines the read first synthesis target Ex1 to third synthesis target Ex3 and the simulation model HMs (step S50). As a result, a homologous model HMD composed of a portion corresponding to the head of the human body, a portion corresponding to both hands of the human body, a portion corresponding to both feet of the human body, and the simulation model HMs is generated.

[Simulation model output processing]
When the simulation model synthesis process ends, the model generation unit 23d reads out the homologous model HMD associated with the user identifier ID from the homologous model database 24c. Then, the model generation unit 23d outputs the read homology model HMD and the homology model HMD obtained from the simulation model synthesis process to the user terminal 1B. That is, the model generation unit 23d outputs the homology model HMD including the reference homology model HMi and the homology model HMD including the simulation model HMs to the user terminal 1B. Thereby, in the body model output processing unit 15 of the user terminal 1B, as shown in FIG. 15, the 3D body image including the reference homology model HMi and the 3D body image including the simulation model HMs are input / output screens. Displayed on VW. That is, the three-dimensional body image showing the latest body shape of the user and the three-dimensional body image reflecting the latest body shape and the target value of the body shape parameter are displayed on the input / output screen VW.

  According to such processing, a three-dimensional body image showing the latest body shape and a future three-dimensional body image based on the latest body shape are provided to the user. Since these three-dimensional body images are based on a common reference homology model HMi, the arm length Lh, leg length Lc, and shoulder width Bs deviate from the viewpoint of human anatomy. It becomes difficult.

As described above, according to the first embodiment, the effects listed below can be obtained.
(1) An initial homologous model HMa is generated from the initial values B _i1 to B _{ih of} the body shape parameters based on the results of the statistical processing, and the target values B _p1 to the target values of the body shape parameters based on the results of the statistical processing. target homology model HMb is generated from B _ph. Then, a difference between the initial homology model HMa and the target homology model HMb is added to the reference homology model HMi, whereby a simulation model HMs, which is a new three-dimensional model, is generated. Therefore, compared to the case where the reference homology model HMi is a simple sample in statistical processing, a characteristic portion of the reference homology model HMi remains in the simulation model HMs. As a result, since the shape of the user-specific body shape is directly reflected in the simulation result, it is possible to improve the accuracy of the simulation model HMs generated using statistical processing.

(2) Principal component analysis is performed in such a manner that the reference homology model HMi is included in the population. Therefore, the initial principal component value P _i1 to the initial principal component value P _ij and the target principal component value P _p1 to the target principal component value P _pj obtained from the principal component analysis reflect the body shape of the reference homology model HMi. Will be. Therefore, as compared with the case where the statistical processing is executed in a mode in which the reference homology model HMi is not included, the difference between the initial homology model HMa and the target homology model HMb is also a characteristic part of the body shape of the reference homology model HMi. Will be reflected.

(3) Since the simulation model HMs is generated from a plurality of body parameter values, the simulation model HMs can be generated with higher accuracy compared to the case where the simulation model HMs is generated from one body parameter value. . That displacement is a difference between the initial homology model HMa and goal homology model HMb T _n also becomes higher accuracy. Therefore, in addition to the reference homology model HMi being handled as an object to be changed, the difference added to the reference homology model HMi is of higher accuracy. As a result, it is possible to further improve the accuracy of the body model generated using statistical processing.

(4) A simulation target HMp, which is the remaining portion obtained by removing the first synthesis target Ex1 to the third synthesis target Ex3 from the homologous model HMD, is generated, and statistical processing and simulation model generation processing using the simulation target HMp are performed. Executed. Then, after the simulation model HMs is generated, the simulation model HMs and the first synthesis target Ex1 to the third synthesis target Ex3 are synthesized. Therefore, it is possible to reduce the data amount of the sample used for statistical processing and simulation model generation processing.
(Second embodiment)
Hereinafter, a second embodiment of the present invention will be described with reference to FIGS. FIG. 16 is a block diagram showing the data structure of the target body model data, and FIG. 17 is a functional block diagram showing the structure of the body model generation server of the second embodiment. FIG. 18 is a flowchart showing the flow of the simulation model generation process, and FIGS. 19A and 19B are diagrams showing the output results of the initial homologous model and the simulation model displayed on the input / output screen of the user terminal. .

  In the second embodiment, a correlation determination unit 23e is added to the data processing unit 23 of the body model generation server 1A of the first embodiment. Then, the determination as to whether or not the correlation between the body shape parameters is given is performed by the correlation determination unit 23e, and the target value of the body shape parameter is changed according to the determination. That is, when the correlation determination unit 23e determines whether to correlate between the body parameters, the correlation between the body parameters is added to the target value for generating the target homology model HMb. Therefore, below, the changed point from 1st embodiment is demonstrated in detail.

  As shown in FIG. 16, the target body model data 14c includes a header 141c that is data relating to the target body model data 14c itself, target body parameter data 142c used for simulation of the three-dimensional body model, and simulation conditions. It consists of a single piece of correlation selection data 143c.

The target body parameter data 142c is a body parameter value set according to the purpose of the body simulation. The value of the body parameter constituting the target body model data 14c is a value that is an index of the future body shape of the user, and is treated as a target value B _p1 that is one of the input values of the simulation of the three-dimensional body model. It is.

  The correlation selection data 143c is used to determine whether or not a correlation between body parameters is given to each body parameter value. More specifically, the correlation selection data 143c includes an identifier indicating the non-correlation parameter PC1, an identifier indicating the reference correlation parameter PC2, and an identifier indicating the partial correlation parameter PC3. This is data for selecting the reference correlation body shape parameter PC2 and the partial correlation body shape parameter PC3 from a plurality of body shape parameters.

  The uncorrelated body parameter PC1 is a body parameter selected from a plurality of body parameters, and is a body parameter to which no correlation between body parameters is given in the simulation of a three-dimensional body model. This non-correlated body parameter PC1 is a body parameter having a weak correlation with other body parameters, and is a body parameter in which the initial value of the body parameter is set to a target value.

The reference correlation body shape parameter PC2 is one body shape parameter selected from a plurality of body shape parameters excluding the uncorrelated body shape parameter PC1, and is a body shape parameter in which the target value B _p1 is set in the target body shape parameter data 142c. is there. The reference correlation body shape parameter PC2 is used as a reference for correlation between body shape parameters in the simulation of the three-dimensional body shape model.

The partial correlation body shape parameter PC3 is a body shape parameter obtained by removing the non-correlation body shape parameter PC1 and the reference correlation body shape parameter PC2 from a plurality of body shape parameters. The partial correlation body shape parameter PC3 is a correlation target to which a correlation with the reference correlation body shape parameter PC2 is given in the simulation of the three-dimensional body shape model. In the simulation of the three-dimensional body shape model, the correlation between the reference correlation body shape parameter PC2 and the partial correlation body shape parameter PC3 is given to each partial correlation body shape parameter PC3.

  For example, when a height that is weakly correlated with other body shape parameters is selected as the non-correlated body shape parameter PC1, body shape parameters other than height are set as the correlated body shape parameters. When the body weight is selected from the correlation body shape parameters as the reference correlation body shape parameter PC2, a correlation between the body weight and the body shape parameter other than the body weight is given to each body shape parameter other than the body weight, that is, each partial correlation body shape parameter PC3. When the correlation selection data 143c is blank, the simulation of the three-dimensional body model is executed in such a manner that no correlation is given between the body parameter values.

[Body Shape Generation Server 1A]
As shown in FIG. 17, the data processing unit 23 of the body model generation server 1A includes a scan data processing unit 23a, a homologous model generation unit 23b, a statistical processing unit 23c, a model generation unit 23d, and a correlation determination unit. 23e. The model generation processing control unit 21 controls various processes in the scan data processing unit 23a, the homologous model generation unit 23b, the statistical processing unit 23c, the model generation unit 23d, and the correlation determination unit 23e according to the simulation program.

  The correlation determination unit 23e determines whether or not to correlate between body shape parameters based on the correlation selection data 143c received by the transmission / reception processing unit 22. That is, the correlation determination unit 23e determines whether or not to select the uncorrelated body parameter PC1, the reference correlated body parameter PC2, and the partial correlated body parameter PC3 from a plurality of body parameters. For example, in the correlation selection data 143c, it is assumed that the uncorrelated body parameter PC1 is height and the reference correlated body parameter PC2 is weight. Then, the correlation determination unit 23e determines that the height is the uncorrelated body parameter PC1, the weight is the reference correlation body parameter PC2, and the body shape parameters other than the height and the weight are the partial correlation body parameter PC3. On the other hand, when the correlation selection data 143c is blank, the correlation determination unit 23e determines that no correlation is given between the body shape parameters.

The statistical processing unit 23c executes, for each population, multiple regression processing for estimating the value of one body parameter from the values of a plurality of body parameters. More specifically, a multiple regression process in which the uncorrelated body parameter PC1 determined by the correlation determination unit 23e is treated as an objective variable and a correlated body parameter other than the uncorrelated body parameter PC1 is treated as an explanatory variable. Run for each group. Then, the statistical processing unit 23c calculates the correlation between the reference correlation body shape parameter PC2 and each partial correlation body shape parameter PC3 as a relationship as shown in Expression (8). I.e. calculated as statistical functions representing the values Bp2~ value _{B pj} of partial correlations body shape parameters PC3 with the value _{B p1} reference correlation forms parameter PC2. In addition, the statistical processing unit 23c updates the statistical processing result database 24d in a manner in which each correlation body parameter and a statistical function applied to the correlation body parameter are associated with each population.

For example, the statistical processing unit 23c calculates a multiple regression equation in which the height, which is the uncorrelated body parameter PC1, is treated as an objective variable, and the correlated body shape parameter composed of the reference correlated body shape parameter and the partial correlated body shape parameter is treated as an explanatory variable. Create for each population. Then, the statistical processing unit 23c extracts partial regression coefficients a ₁ to partial regression coefficients a _h (h is an integer of 2 or more), which are coefficients related to each correlator shape parameter, and determines the bias related to body weight, which is the reference correlator shape parameter PC2. dividing the partial regression coefficient a2~ partial regression coefficients _{a h} according to the partial correlation body shape parameters PC3 regression coefficient _{a 1.} Use by such processing, the correlation between each partial correlation forms parameter PC3 the reference correlation body shape parameters PC2 is, the correlation coefficient _{(a 2 / a 1, a} 3 / a 1, ..., a h / a 1) a It is calculated as a statistical function.

The model generation unit 23d reads the incidental data 143a associated with the user identifier ID from the homologous model database 24c. In addition, the model generation unit 23d extracts the initial value B _i1 to the initial value B _ih of the body shape parameters from the read incidental data 143a. Then, the model generation unit 23d uses the extracted initial value B _i1 to initial value B _{ih of} the body shape parameter and the inverse matrix of the correlation matrix A ₂ (see Expression (7)), and uses the initial value B _i1 to the initial value B. Each principal component value corresponding to _ih is inversely calculated. The model generation unit 23d uses the principal component values as the initial principal component value P _i1 to the initial principal component value P _ij .

Further, the model generation unit 23d extracts the target value of the body shape parameter, that is, the target value B _p1 of the reference correlation body shape parameter, from the target body shape model data 14c. Furthermore, the model generation unit 23d extracts human body attributes from the read incidental data 143a, and each correlation coefficient (a ₂ / a ₁ , a ₃ / a ₁ ,..., A associated with the human body attributes is extracted. _h / a ₁ ) is read from the statistical processing result database 24d. Then, the model generation unit 23d calculates a target value B _p2 to a target value B _{ph of} each partial correlation body parameter, that is, a correlation correction value, using the target value B _p1 of the reference correlation body shape parameter and each correlation coefficient.

  Next, the flow of simulation of a three-dimensional body model using the above three-dimensional shape model generation system will be described with reference to FIGS. 18, 19A and 19B. In the second embodiment, the simulation model generation process and the simulation model output process of the first embodiment are mainly changed. Therefore, the following mainly describes the simulation model generation process and the simulation model output process, which are the changes.

First, prior to the execution of scan data processing, target body model data 14c including correlation selection data 143c is transmitted from the user terminal 1B to the body model generation server 1A. At this time, as shown in FIG. 19A, the body parameter selection table VWs provided on the input / output screen VW includes the body parameter target in the body parameter column selected as the reference correlation body parameter PC2. The value B _p1 is displayed. In addition, “0” is displayed as the value of the body parameter in the body parameter column selected as the partial correlation body parameter PC3. In addition, the body parameter selected as the uncorrelated body parameter is not displayed in the selection table VWs itself.

In FIG. 19A, the body weight is selected as the reference correlation body shape parameter PC2, and the target value B _p1 “60.5” is displayed in the body weight column. As in the first embodiment, depending on whether or not the three-dimensional scan data 142a can be acquired, scan data processing (step S12), homologous model generation processing (step S20), statistical processing (step S30), The simulation model generation process (step S40), the simulation model synthesis process (step S50), and the simulation model output process (step S60) are sequentially executed with different contents.

[Simulation model generation processing]
When the statistical processing ends, the model generation unit 23d reads the simulation target HMp linked to the user identifier ID from the homologous model database 24c, as shown in FIG. 18, and uses the simulation target HMp as the reference homologous model HMi. Set as. If the model generation processing control unit 21 determines that the three-dimensional scan data 142a cannot be acquired, the model generation unit 23d displays the simulation target HMp associated with the user identifier ID as the standard model database 24b. And the simulation target HMp is set as the reference homology model HMi (step S41).

Next, the model generation unit 23d reads the incidental data 143a associated with the user identifier ID from the homologous model database 24c, and extracts the initial values B _i1 to B _ih of the body shape parameters included in the incidental data 143a. . Further, the model generation unit 23d extracts the target value B _p1 of the reference correlation body shape parameter PC2 from the target body shape model data 14c received from the user terminal 1B (Step S42).

When the model generation unit 23d extracts the initial value B _i1 to the initial value B _ih and the target value B _{p1 of the} body shape parameter, the correlation determination unit 23e uses the body shape parameter based on the correlation selection data 143c received by the transmission / reception processing unit 22. It is determined whether or not a correlation is given to the value of. That is, the correlation determination unit 23e determines whether or not to select the uncorrelated body parameter PC1, the reference correlated body parameter PC2, and the partial correlated body parameter PC3 from the plurality of body parameters (step S421).

  If the correlation determination unit 23e determines that the correlation is to be given (step S421: YES), the statistical processing unit 23c reads the incidental data 143a associated with the user identifier ID from the homologous model database 24c. The attributes of the human body included in the incidental data 143a are extracted. Next, the statistical processing unit 23c reads the values of a plurality of body shape parameters associated with the same attribute as the extracted human body attribute from the homologous model database 24c.

  The statistical processing unit 23c then performs multiple regression in which the uncorrelated body parameter PC1 determined by the correlation determination unit 23e is treated as an objective variable, and a correlated body parameter other than the uncorrelated body parameter PC1 is treated as an explanatory variable. The process is executed for the read values of the plurality of body shape parameters. Further, the statistical processing unit 23c calculates a correlation coefficient (see equation (8)) indicating the correlation between the reference correlation body parameter PC2 and the partial correlation body parameter PC3 (step S422). Then, the statistical processing unit 23c updates the statistical processing result database 24d in such a manner that the correlation body parameters used in the multiple regression processing and the correlation coefficient applied to the correlation body parameters are associated with each population.

When the statistical processing unit 23c calculates the correlation coefficient, the model generation unit 23d extracts the human body attribute from the read incidental data 143a and correlates the correlation coefficient (a ₂ / a ₁ , a ₃ / a ₁ ,..., a _h / a ₁ ) are read from the statistical processing result database 24d. Then, the model generation unit 23d calculates a target value B _p2 to a target value B _{ph of} each partial correlation body parameter, that is, a correlation correction value by using the target value B _p1 of the reference correlation body shape parameter and each correlation coefficient (step S423).

When the correlation correction value is calculated for each partial correlation body parameter PC3, the model generation unit 23d determines the initial value B _i1 to the initial value B _{ih of} each body parameter and the correlation matrix A ₂ (see Expression (7)). And the initial principal component value P _i1 to the initial principal component value P _ij corresponding to the initial value B _i1 to the initial value B _ih are inversely calculated (step S43).

Next, the model generation unit 23d handles the initial values of the uncorrelated body parameters as target values, and the target values B _p1 to B _{ph of} all body parameters and the inverse matrix of the correlation matrix A ₂ (see Expression (7)). And the target principal component value P _p1 corresponding to the target value B _p1 to the target value B _ph
The target principal component value P _pj is inversely calculated (step S44).

If the correlation determination unit 23e determines that no correlation is given (step S421: NO), the model generation unit 23d treats the initial value of the uncorrelated body parameter PC1 as a target value and performs partial correlation. The initial value of the body shape parameter PC3 is handled as the target value (step S424). Then, the model generation unit 23d uses the target value B _p1 to the target value B _{ph of} all body parameters and the inverse matrix of the correlation matrix A ₂ (see Expression (7)), and the target value B _p1 to the target value B _The target principal component value P _p1 to the target principal component value P _pj corresponding to _ph are inversely calculated.

When the initial principal component value P _i1 to the initial principal component value P _ij and the target principal component value P _p1 to the target principal component value P _pj are calculated, the model generation unit 23d performs the initial principal component value as in the first embodiment. Using the value P _i1 to the initial principal component value P _ij and the inverse matrix of the eigenvector matrix A ₁ , the coordinate value of the polygon vertex is calculated by performing the inverse calculation of the above equation (6). That is, the initial homology model HMa corresponding to the initial value B _i1 to the initial value B _ih of the body shape parameter is generated (step S45). Further, the model generation unit 23d uses the target principal component value P _p1 to the target principal component value P _pj and the inverse matrix of the eigenvector matrix A ₁ and executes the inverse operation of the above equation (6) to thereby obtain the coordinate value of the polygon vertex. Is calculated. That is, the target homologous model HMb corresponding to the target value B _p1 to the target value B _ph of the body shape parameter is generated (step S46).

When the initial homologous model HMa and the target homologous model HMb are generated, the model generation unit 23d, like the first embodiment, each polygon that configures the target homologous model HMb from each polygon coordinate that configures the initial homologous model HMa. It calculates the displacement amount T _n to coordinate. Then the model generating unit 23d, the operation result and the displacement amount T _n is, three-dimensional body shape model initial value a is used and the reference homologous model HMi, reference homology model HMi polygon model plus displacement amount T _n in the Are calculated as simulation models HMs. Then, the model generation unit 23d updates the body model database 24e in a manner in which the simulation model HMs and the user identifier ID are linked, and ends the simulation model generation process for the user.

  Here, for example, when the weight, which is one of the body shape parameters, decreases, it is natural that the body shape itself becomes thin. In particular, the current body shape greatly determines whether the thinned part is the chest part or the abdominal part. There are many differences. In addition, the peripheral length of the narrowed portion decreases in relation to each other, for example, these decrease in many cases so that the amount of decrease in the length of the navel is less than the amount of decrease in the chest circumference.

In this regard, according to the configuration as described above, a correlation is given in advance between the target value B _p1 to the target value B _ph used for generating the simulation model HMs. For this reason, as long as the target value of the reference correlation body shape parameter PC2 is input, a target value corresponding to the shape characteristic of the population is set for other body shape parameters other than the reference correlation body shape parameter PC2. For example, even if it is a request | requirement which produces | generates simulation model HMs only based on the value of a body weight, the target value according to the attribute of the human body to which a user belongs is set with respect to the body shape parameters other than a body weight. As a result, the accuracy of the simulation model HMs is improved as compared with the case where the target value when generating the simulation model HMs is only the body weight or the case where the target value not conforming to the attribute of the human body to which the user belongs is used. It becomes possible to make it.

[Simulation model output processing]
When the simulation model synthesis process ends in the same manner as in the first embodiment, the model generation unit 23d reads out the homologous model HMD associated with the user identifier ID from the homologous model database 24c. Then, the model generation unit 23d outputs the read homologous model HMD and the homologous model HMD obtained from the simulation model synthesis process to the user terminal 1B. The model generation unit 23d outputs the target value of the partial correlation body parameter PC3 used for the simulation generation process to the user terminal 1B. Thereby, in the body model output processing unit 15 of the user terminal 1B, as shown in FIG. 19A, the three-dimensional body image including the reference homology model HMi is displayed on the input / output screen VW together with the selection table VWs. . Next, in the body model output processing unit 15 of the user terminal 1B, as shown in FIG. 19 (b), the three-dimensional body image including the simulation model HMs and the partial correlation body parameter PC3 used for the simulation generation process are displayed. The selection table VWs including the target value is displayed on the input / output screen VW.

  According to such processing, as in the first embodiment, a three-dimensional body image showing the latest body shape and a future three-dimensional body image based on the latest body shape are provided to the user. Even though the three-dimensional body image is generated by inputting only the value of the reference correlation body shape parameter PC2, the correlation with the reference correlation body shape parameter PC2 is given to the other body shape parameters, so the arm length Lh The sizes of the leg length Lc and the shoulder width Bs are difficult to deviate in terms of human anatomy. In addition, the value of the correlated body parameter is provided to the user along with the future three-dimensional body image based on the latest body shape. Therefore, it is possible to provide the user with specific values such as the perimeter length that is difficult to confirm from the three-dimensional body shape image.

As described above, according to the second embodiment, in addition to the effects (1) to (4), the effects listed below can be obtained.
(5) Based on the correlation with the reference correlation body parameter PC2, target values of all the partial correlation body parameters PC3 are calculated. Therefore, the value of the body parameter that matches the shape characteristic of the population to which the user's body shape belongs is obtained only from the target value B _p1 of the reference correlation body parameter PC2. Therefore, it is possible to give high accuracy to the target values themselves of the plurality of body parameters, so that it is possible to further improve the accuracy of the target homologous model HMb and eventually the simulation model HMs generated using statistical processing.
(Third embodiment)
Hereinafter, a third embodiment of the present invention will be described with reference to FIGS. FIG. 20 is a configuration diagram showing the data configuration of the target body model data of the third embodiment, and FIG. 21 is a diagram showing joint positions and skeletons together with homologous models. FIG. 22 is a functional block diagram showing the configuration of the body model generation server, and FIG. 23 is a configuration diagram showing the data configuration of the skeleton parameter data. FIG. 24 is a flowchart showing the flow of homology model generation processing, and FIG. 25 is a flowchart showing the flow of simulation model generation processing.

  In the third embodiment, a skeleton fixation determination unit 23f is added to the data processing unit 23 of the body model generation server 1A of the second embodiment. Then, the determination as to whether or not the initial homology model HMa and the target homology model HMb are the same, that is, the determination as to whether or not to fix the skeleton is performed by the skeleton fixation determination unit 23f, and according to the determination This is a form in which the conditions for generating the target homology model HMb are changed. Therefore, below, the changed point from 2nd embodiment is demonstrated in detail.

  As shown in FIG. 20, the target body model data 14c includes a header 141c, target body parameter data 142c, and correlation selection data 143c, as in the second embodiment, and skeleton parameter data that is one of the simulation conditions. 144c is further provided.

The skeleton parameter data 144c is data relating to the skeleton that governs the shape of the human body as a whole, and is composed of skeleton parameter values. As shown in FIG. 21, the skeletal parameters are the length between the shoulder joint J1 and the elbow joint J2, the length between the elbow joint J2 and the sacroiliac joint J3, the sacroiliac joint J3, and the hip joint J4. The length between the joints, which is one of the physical quantities of the human body, such as the length between the hip joint J4 and the knee joint J5. This skeletal parameter data 144c is a value of the body shape parameter set according to the purpose of the simulation. That is, the value of the skeleton parameter constituting the target body model data 14c is a value that serves as an index of the future body shape of the user, and is treated as a target value that is one of the input values of the simulation of the three-dimensional body model.

  When the length between joints is included in the skeleton parameter data 144c, the simulation of the three-dimensional body model is executed based on the target value of the skeleton parameter data 144c. On the other hand, when the skeleton parameter data 144c is blank, the simulation of the three-dimensional body model is executed in a manner that fixes the skeleton.

[Body Shape Generation Server 1A]
As shown in FIG. 22, the data processing unit 23 of the body model generation server 1A is similar to the second embodiment in that the scan data processing unit 23a, the homologous model generation unit 23b, the statistical processing unit 23c, and the model generation 23d and a correlation determination unit 23e, and further includes a skeleton fixation determination unit 23f that determines whether to fix the skeleton based on the skeleton parameter data 144c. The model generation processing control unit 21 performs various processes in the scan data processing unit 23a, the homologous model generation unit 23b, the statistical processing unit 23c, the model generation unit 23d, the correlation determination unit 23e, and the skeleton fixation determination unit 23f according to the simulation program. Supervise processing.

  The skeleton fixation determination unit 23f determines whether to fix the skeleton based on the skeleton parameter data 144c received by the transmission / reception processing unit 22. That is, the skeleton fixation determination unit 23f determines whether or not the initial homology model HMa and the target homology model HMb have the same skeleton, that is, whether or not the skeleton is fixed in the simulation of the three-dimensional shape model.

The homology model generation unit 23b generates joint position data BD using the homology model HMD or the standard model generated by the homology model generation unit 23b. As shown in FIG. 23, the joint position data BD includes a header 243c, which is data related to the joint position data BD itself, and joint coordinates 244c. The joint coordinates 244c are the joint positions J1 to Jn (n is an integer of 2 or more) and their coordinate values, such as the shoulder joint J1, the elbow joint J2, the sacroiliac joint J3, the hip joint J4, and the knee joint J5. It is comprised by coordinate _Cj1 -joint coordinate _Cjn (n is an integer greater than or equal to 2). The homologous model generation unit 23b uses the coordinate values of a plurality of polygon vertices located around each of the joint positions J1 to Jn, and an arithmetic expression set in advance to obtain the joint position, and uses the joint coordinates C _j1 to joint coordinates C _jn are calculated. Then, the homologous model generation unit 23b updates the homologous model database 24c in such a manner that the joint position data BD is associated with the identifier ID of the user. That is, the homologous model generation unit 23b updates the homologous model database 24c in such a manner that data relating to the current skeleton of the user is associated with the identifier ID of the user.

  When the simulation of the three-dimensional body model is executed based on the user's homologous model HMD, the homologous model generating unit 23b uses the polygon vertex of the homologous model HMD generated by the homologous model generating unit 23b to Data BD is generated. The homologous model generation unit 23b calculates the length between the joints from the joint coordinates 244c included in the joint position data BD. The length between the joints calculated by the homologous model generation unit 23b is handled as the value of the skeleton parameter in the current body shape of the user, that is, the initial value of the skeleton parameter in the simulation.

On the other hand, when the simulation of the three-dimensional figure model is executed based on a three-dimensional figure model different from the user's homologous model HMD, the homologous model generation unit 23b reads and reads the standard model from the standard model database 24b. The joint position data BD is generated using the polygon vertices of the standard model. The homologous model generation unit 23b calculates the length between the joints from the joint coordinates 244c included in the joint position data BD. The length between the joints calculated by the homologous model generation unit 23b is handled as the value of the skeletal parameter in the standard body shape, that is, the initial value of the skeletal parameter in the simulation.

  The statistical processing unit 23c performs a multiple regression process using the principal component values, body shape parameters, and skeleton parameters of the simulation model HMs as samples for each population, that is, for each human body attribute. More specifically, the statistical processing unit 23c reads the initial values of the body parameters, the initial values of the skeletal parameters, the principal component values, and the same from the homologous model database 24c for all the simulation target HMp constituting one population. The statistical processing unit 23c executes multiple regression processing for each human body attribute in such a manner that the initial value of the body shape parameter and the initial value of the skeletal parameter are objective variables (explained variables) and the principal component value is an explanatory variable. By this multiple regression processing, the statistical processing unit 23c calculates the correlation between the initial value of the body shape parameter and the initial value of the skeleton parameter and the principal component value as a relationship corresponding to the above equation (7), that is, the correlation matrix A2. . Further, the statistical processing unit 23c updates the statistical processing result database 24d in such a manner that the attribute of the human body subjected to statistical processing and the correlation matrix A2 are associated with each other.

The model generation unit 23d reads the incidental data 143a associated with the user identifier ID from the homologous model database 24c. In addition, the model generation unit 23d extracts initial values of body shape parameters from the read incidental data 143a. The model generation unit 23d reads out the initial value of the skeleton parameter associated with the user identifier ID from the homologous model database 24c. Then, the model generation unit 23d uses the initial value of the body shape parameter, the initial value of the skeletal parameter, and the inverse matrix of the correlation matrix A ₂ (see Expression (7)), and inversely calculates each principal component value according to the initial value. To do. The model generation unit 23d uses the principal component value, which is the calculation result, as the initial principal component value.

  The model generation unit 23d extracts the target value of the body shape parameter from the target body shape parameter data 142c of the target body shape model data 14c, and extracts the target value of the skeleton parameter from the skeleton parameter data 144c of the target body shape model data 14c.

Then, when the simulation without fixing the skeleton is executed, the model generation unit 23d obtains the target value of the extracted body parameter, the target value of the extracted skeleton parameter, and the inverse of the correlation matrix A ₂ (see Expression (7)). Using the matrix, each principal component value corresponding to the target value is calculated. Then, the model generation unit 23d uses the principal component value that is the calculation result as the target principal component value.

  On the other hand, when the simulation for fixing the skeleton is executed, the model generation unit 23d handles the read initial value of the skeleton parameter as the target value of the skeleton parameter. Then, using the target value of the body parameter, the target value (initial value) of the skeleton parameter, and the inverse matrix of the correlation matrix A2 (see Expression (7)), each principal component value is calculated according to the target value. . Then, the model generation unit 23d uses the principal component value that is the calculation result as the target principal component value.

Then, as in the second embodiment, the model generation unit 23d uses the initial principal component value and the inverse matrix of the eigenvector matrix A ₁ (see Equation (6)) to calculate a polygon model corresponding to the initial principal component value. Calculate. The model generation unit 23d uses the polygon model as the initial homologous model HMa. The model generation unit 23d calculates a polygon model corresponding to the target principal component value using the target principal component value and the inverse matrix of the eigenvector matrix A ₁ (see Expression (6)). The model generation unit 23d uses the polygon model as the target homologous model HMb.

  Next, the flow of simulation of a three-dimensional body model using the above three-dimensional shape model generation system will be described with reference to FIGS. In the third embodiment, the homologous model generation process, statistical process, and simulation model generation process of the second embodiment are mainly changed. Therefore, in the following description, the homologous model generation process, statistical process, and simulation model generation process, which are the changes, will be mainly described.

[Homologous model generation processing]
As shown in FIG. 24, the preprocessing model associated with the user identifier ID and the accompanying data 143a are read from the preprocessing model database 24a. In addition, the homologous model generation unit 23b extracts human body attributes and initial values B _i1 to B _{h of} body parameters from the read incidental data 143a. Further, the homologous model generation unit 23b reads the standard model associated with the human body attribute from the standard model database 24b using the extracted human body attribute.

  Then, the homologous model generation unit 23b generates a homologous model HMD including the polygon coordinates by displacing the polygon coordinates of the standard model so that the body surface indicated by the read preprocessing model matches the body surface indicated by the standard model. To do. The homologous model generation unit 23b generates joint position data BD corresponding to the polygon vertex of the homologous model HMD based on the generated homologous model HMD. When the homologous model HMD and the joint position data BD are generated, the homologous model generation unit 23b generates initial values of the body parameters related to the homologous model HMD, the joint position data BD, and the homologous model HMD, and attributes of the human body. The homology model database 24c is updated in such a manner that they are linked to the identifier ID of the user (step S21).

  When the homologous model HMD and the joint position data BD are stored in the homologous model database 24c, the homologous model generating unit 23b reads from the homologous model HMD linked to the user identifier ID as in the second embodiment. The first synthesis target Ex1 to the third synthesis target Ex3 are extracted (step S22). The homologous model generation unit 23b extracts from the homologous model HMD the simulation target HMp that is the remaining part obtained by removing the first synthesis target Ex1 to the third synthesis target Ex3 from the homologous model HMD (step S23). Then, the homologous model generation unit 23b updates the homologous model database 24c in such a manner that the first synthesis target Ex1 to the third synthesis target Ex3 and the simulation target HMp are linked to the identifier ID of the user, and generates a homologous model for the simulation. The process ends (step S24).

  When the model generation processing control unit 21 determines that the 3D scan data 142a cannot be acquired, the homologous model generation unit 23b reads the human body from the incidental data 143a associated with the user identifier ID. Extract attributes. Then, the homologous model generation unit 23b reads out the standard model associated with the attribute of the human body from the standard model database 24b.

  Then, the homologous model generation unit 23b generates joint position data BD corresponding to the polygon vertex of the standard model based on the read standard model. When the standard model and the joint position data BD are generated, the homologous model generation unit 23b includes the standard model, the joint position data BD, the initial values of the body parameters related to the standard model, and the attributes of the human body. The homology model database 24c is updated in a manner associated with the user identifier ID (step S211).

Next, as in the second embodiment, the homologous model generation unit 23b extracts the first synthesis target Ex1 to the third synthesis target Ex3 from the standard model (step S221). In addition, the homologous model generation unit 23b extracts the simulation target HMp, which is the remaining part obtained by removing the first synthesis target Ex1 to the third synthesis target Ex3 from the standard model, from the standard model (step S231). Then, the homologous model generation unit 23b updates the standard model database 24b in such a manner that the first synthesis target Ex1 to the third synthesis target Ex3 and the simulation target HMp are linked to the identifier ID of the user, and generates a homologous model for the simulation. The process ends (step S27).

[Statistical processing]
When the homologous model generation process is completed, the statistical processing unit 23c performs the principal component analysis on the plurality of simulation target HMp, as in the second embodiment, and the principal component that is the analysis result and each simulation target HMp. main component value, the eigenvector matrix a ₁ and further statistical processing unit 23c calculates the the extracted human attributes plurality of body shape parameters associated with the same attribute as the initial value and skeletal parameter initial value homology model of Read from database 24c. The statistical processing section 23c, an initial value and the initial value is dependent variable skeletal parameters read body shape parameters, the principal component values by performing multiple regression process for calculating a correlation matrix A ₂ to be handled as explanatory variables.

The statistical processing unit 23c updates the homologous model database 24c in such a manner that each simulation target HMp and the principal component value of the simulation target HMp are linked. The statistical processing section 23c, the human body attribute to be processed for statistical processing, the eigenvector matrix of A ₁ and the correlation matrix A ₂ updates the statistical processing result database 24d in a manner to be linked, statistics for the user End the process.

[Simulation model generation processing]
When the statistical process ends, the model generation unit 23d sets the simulation target HMp associated with the user identifier ID as the reference homologous model HMi, as shown in FIG. 25, as in the second embodiment. When the model generation processing control unit 21 determines that the three-dimensional scan data 142a cannot be acquired, the model generation unit 23d uses the simulation target HMp associated with the user identifier ID as the reference homologous model HMi. (Step S41).

  Next, the model generation unit 23d extracts the initial value of the body shape parameter from the incidental data 143a associated with the identifier ID of the user. The model generation unit 23d extracts the target value of the reference correlation body shape parameter PC2 from the target body shape model data 14c received from the user terminal 1B (step S42). When the model generation unit 23d extracts the initial value and the target value of the body shape parameter, the correlation determination unit 23e determines whether or not to correlate the body parameter value based on the correlation selection data 143c. (Step S421).

  If the correlation determination unit 23e determines that the correlation is to be given (step S421: YES), the statistical processing unit 23c extracts the human body attribute from the incidental data 143a associated with the user identifier ID. . In addition, the statistical processing unit 23c reads the values of a plurality of body shape parameters associated with the same attribute as the extracted human body attribute from the homologous model database 24c. The statistical processing unit 23c then performs multiple regression in which the uncorrelated body parameter PC1 determined by the correlation determination unit 23e is treated as an objective variable, and a correlated body parameter other than the uncorrelated body parameter PC1 is treated as an explanatory variable. Processing is executed to calculate a correlation coefficient (step S422).

When the statistical processing unit 23c calculates the correlation coefficient, the model generation unit 23d calculates the target value of each partial correlation body parameter, that is, the correlation correction value, using the target value of the reference correlation body parameter and each correlation coefficient. (Step S423). If the correlation determination unit 23e determines that no correlation is given (step S421: NO), the model generation unit 23d treats the initial value of the uncorrelated body parameter PC1 as the target value and also uses the partial correlation body shape. The initial value of parameter PC3 is handled as a target value (step S424).

  When the correlation correction value is calculated for each partial correlation body parameter PC3, the homology model generation unit 23b uses the polygon vertices of the reference homology model HMi linked to the identifier ID of the user, and uses the joint position data BD. Is generated. The homologous model generation unit 23b calculates the length between joints, that is, the initial value of the skeleton parameter, from the joint coordinates 244c included in the joint position data BD. When the model generation processing control unit 21 determines that the 3D scan data 142a cannot be acquired, the homologous model generation unit 23b reads the standard model from the standard model database 24b, and reads the polygon of the standard model that has been read. The joint position data BD is generated using the vertices. The homologous model generation unit 23b calculates the length between the joints, that is, the initial value of the skeletal parameter from the joint coordinates 244c included in the joint position data BD (step S431).

  When the homologous model generation unit 23b calculates the initial value of the skeleton parameter, the skeleton fixation determination unit 23f determines whether to fix the skeleton based on the skeleton parameter data 144c received by the transmission / reception processing unit 22 (step S432). .

When the skeleton fixation determination unit 23f determines that the skeleton is to be fixed (step S432: YES), the model generation unit 23d determines the initial value of the body shape parameter, the initial value of the skeleton parameter, the correlation matrix A ₂ (formula Using the inverse matrix (see (7)), the initial principal component value corresponding to each initial value is inversely calculated (step S433). Next, the model generation unit 23d uses the target value of each body parameter, the initial value of the skeletal parameter, with the initial value of the uncorrelated body parameter as the target value, and the inverse matrix of the correlation matrix A ₂ (see Expression (7)). The target principal component values corresponding to these values are inversely calculated (step S441).

When the skeleton fixation determining unit 23f determines that the skeleton is not fixed (step S432: NO), the model generation unit 23d determines the initial value of the body shape parameter, the initial value of the skeleton parameter, the correlation matrix A ₂ (formula Using the inverse matrix (see (7)), the initial principal component value corresponding to each initial value is inversely calculated (step S434). Next, the model generation unit 23d uses the target value of each body parameter, the target value of the skeletal parameter, with the initial value of the uncorrelated body parameter as the target value, and the inverse matrix of the correlation matrix A ₂ (see Equation (7)). The target principal component values corresponding to these values are inversely calculated (step S442).

When the initial principal component values and target main component value is calculated, the model generating unit 23d, like the second embodiment, using the inverse matrix of the initial principal component values and eigenvector matrix A _1, the formula (6) The coordinate value of the polygon vertex is calculated by executing the inverse operation of That is, the initial value of the body shape parameter, the initial value of the skeleton parameter, and the initial homologous model HMa corresponding to them are generated (step S45). Further model generating unit 23d, using the inverse matrix of the target main component value and the eigenvector matrix A _1, calculates the coordinates of the polygon vertices by performing the inverse operation of the equation (6). That is, the target value of the body shape parameter, the initial value or target value of the skeleton parameter, and the target homology model HMb corresponding to them are generated (step S46).

When the initial homologous model HMa and the target homologous model HMb are generated, the model generation unit 23d, like the second embodiment, each polygon that configures the target homologous model HMb from each polygon coordinate that configures the initial homologous model HMa. It calculates the displacement amount T _n to coordinate. Then the model generating unit 23d, the operation result and the displacement amount T _n is, three-dimensional body shape model initial value a is used and the reference homologous model HMi, reference homology model HMi polygon model plus displacement amount T _n in the Are calculated as simulation models HMs. Then, the model generation unit 23d updates the body model database 24e in a manner in which the simulation model HMs and the user identifier ID are linked, and ends the simulation model generation process for the user.

  Here, for example, when the target of the three-dimensional body model is an adult, the length between the joints is necessarily fixed. If new simulation models HMs are generated in a manner in which the skeleton is fixed as described above, it is possible to obtain a three-dimensional body model suitable for the characteristics of adult body shape changes as a new three-dimensional model. Become.

As described above, according to the third embodiment, in addition to the effects (1) to (5), the effects listed below can be obtained.
(6) The length between joints, that is, the initial value of the skeleton parameter is calculated based on the polygon vertex of the reference homology model HMi. When the skeleton fixation determination unit 23f determines that the skeleton is to be fixed, the model generation unit 23d uses the initial value of the body shape parameter and the initial value of the skeleton parameter, and generates a simulation model HMs corresponding to these. . Therefore, a three-dimensional body model suitable for the characteristics of the body shape to be modeled is obtained as a new three-dimensional model for the modeling target to which the skeleton should be fixed.

The embodiment described above can be implemented with the following modifications.
In the third embodiment, whether or not the skeletons of the initial homologous model HMa and the target homologous model HMb are made the same after the determination as to whether or not to give correlation to each partial correlation body parameter C3 is made. Is determined. After this change is made and it is determined whether or not the skeletons of the initial homologous model HMa and the target homologous model HMb are the same, it is determined whether or not a correlation is given to each partial correlation parameter C3. It may be configured. In addition, it is a configuration in which only the determination as to whether or not the skeletons of the initial homologous model HMa and the target homologous model HMb are the same is performed without determining whether or not the correlation is given to each partial correlation body parameter C3. There may be. Even if it is these structures, the effect similar to said (6) is acquired.

  In the second embodiment, the uncorrelated body parameter PC1 is set based on the correlation selection data 143c. This may be modified such that the identifier indicating the uncorrelated body parameter PC1 is omitted from the correlation selection data 143c, and the uncorrelated body parameter PC1 is determined in advance by the statistical processing unit 23c. For example, the statistical processing unit 23c may be configured in advance so that multiple regression processing in which height is handled as an objective variable and other body parameters are handled as explanatory variables is executed for each population. Even with such a configuration, the effect according to the above (5) can be obtained.

  In the second embodiment, each body shape is based on a multiple regression process in which the uncorrelated body parameter PC1 is treated as an objective variable and a correlated body parameter other than the uncorrelated body parameter PC1 is treated as an explanatory variable. Correlation is given to the parameter. By changing this, the configuration may be such that the correlation is given to each body parameter by a conversion table or the like that converts the target value of the reference correlation body parameter PC2 into the partial correlation body parameter PC3.

  -The structure by which the skeleton parameter in 3rd embodiment is contained in the partial correlation body parameter C2 in said 2nd embodiment may be sufficient. In other words, the length between the joints may be included as one of the body shape parameters.

In the above embodiment, a series of processes consisting of scan data processing, homologous model generation processing, statistical processing, simulation model generation processing, simulation model synthesis processing, and simulation model output processing is performed as a single request from the user terminal 1B. It was assumed that they were executed in order. By changing this, for example, a request for scan data processing may be different from a request for homologous model generation processing, statistical processing, simulation model generation processing, simulation model synthesis processing, and simulation model output processing. Furthermore, each of scan data processing, homologous model generation processing, statistical processing, simulation model generation processing, simulation model synthesis processing, and simulation model output processing is started each time by a request from the user terminal 1B corresponding to the processing. It may be a configuration. Even in such a configuration, the effects according to the above (1) to (6) can be obtained. In addition, a plurality of simulations with different body shape parameters and skeleton parameters can be more easily realized with respect to one result of each processing, for example, one preprocessing model and one homologous model HMD.

  In the above embodiment, the simulation based on the homologous model HMD linked to the user identifier ID and the simulation based on the standard model depending on whether or not the coordinate data of the point group is included in the three-dimensional scan data 142a Is switched. For example, a configuration in which the simulation based on the homologous model HMD linked to the identifier ID of the user and the simulation based on the standard model are switched by a request from the user terminal 1B may be used.

  A sample used for multivariate analysis is not limited to a plurality of simulation target HMp having the same human body attribute, and may be a homologous model HMD having the same human body attribute, for example. That is, the multivariate analysis may be performed in such a manner that at least one of the first synthesis target Ex1 to the third synthesis target Ex3 and the simulation target HMp are samples. With such a configuration, in addition to the effects according to the above (1) to (3), (5), and (6), the simulation model synthesis process can be omitted.

  In the above embodiment, the standard model database 24b is constructed based on the standard model transmitted from the user terminal 1B and the human body attributes related to the standard model. By changing this, prior to the start of simulation, the standard model database 24b may be configured in advance so that the standard model and the attributes of the human body related to the standard model are linked.

  In the above embodiment, the multivariate analysis is executed in such a manner that the reference homology model HMi is included in the population. By changing this, the multivariate analysis may be executed in such a manner that the reference homology model HMi is not included in the population. That is, the sample to be subjected to the multivariate analysis may be a plurality of homologous models HMD having human body attributes common to the user. Similarly, the multiple regression process using the principal component value and the body shape parameter as a sample is a sample to be subjected to the multiple regression process, which is a principal component value of a plurality of homologous models HMD whose human body attributes are common to the user, and Any body parameters relating to the homologous model HMD may be used.

  In the above embodiment, the principal component analysis is used as an example of multivariate analysis. However, the present invention is not limited to this. For example, factor analysis or cluster analysis may be used as an example of multivariate analysis.

  In the above embodiment, it is assumed that the modeled object is a human body, but the present invention is not limited to this, and the modeled object may be a domestic animal such as a pet. That is, any object can be used as long as it can acquire 3D shape data and can generate a reference 3D model from the 3D shape data.

A, B, C ... coordinate axes, S, T, U ... principal component axes, A ₁ ... eigenvector matrix, A ₂ ... correlation matrix, BD ... joint position data, C ₁ , C _i , C _n ... vertex coordinates, F ₁ , F _m ... polygon surface, K _i , _Kn ... polygon vertex, NT ... communication network, Pi, P _j ... principal component values, T ₁ , T ₅
, T _n ... displacement amount, B _i1 , B _ih ... initial value, B _p1 , B _ph ... target value, Ex1 ... first synthesis target, Ex2 ... second synthesis target, Ex3 ... third synthesis target, HMp ... simulation target , HMs ... simulation model, Ka ₁ , Ka ₅ , Kb ₁ , Kb ₅ , Ki ₁ , Ki ₅ , Ks ₁ , Ks ₅ ... polygon coordinates, PC1 ... uncorrelated body shape parameter, PC2 ... reference correlator body shape parameter, PC3 ... bias Correlated body parameters, P _i1 , P _ij ... initial principal component values, P _p1 , P _pj ... target principal component values, 1A ... body shape model generation server, 1B ... user terminal, 11 ... data processing unit, 12 ... transmission / reception processing unit , 13 ... input processing unit, 13a ... three-dimensional scanner, 14 ... storage unit, 14a ... initial model setting data, 14b ... landmark data, 14c ... target body model 14d ... user program, 15 ... body model output processing unit, 21 ... model generation processing control unit, 22 ... transmission / reception processing unit, 23 ... data processing unit, 23a ... scan data processing unit, 23b ... homologous model generation unit, 23c ... Statistical processing unit, 23d ... Model generation unit, 23e ... Correlation determination unit, 23f ... Skeletal fixation determination unit, 24a ... Pre-processing model database, 24b ... Standard model database, 24c ... Homologous model database, 24d ... Statistical processing result Database, 24e ... body model database, 24p ... simulation program, 142a ... three-dimensional scan data, 142c ... target body parameter data, 143a ... incidental data, 143c ... correlation selection data, 144c ... skeletal parameter data, 242c ... body coordinate data.

Claims (12)

A feature quantity extraction unit that generates a reference three-dimensional model from three-dimensional shape data of a plurality of modeling objects and extracts a plurality of feature quantities by multivariate analysis of the plurality of reference three-dimensional models;
A storage unit that stores values of physical quantities measurable by the modeled object and values of the plurality of feature values of the reference three-dimensional model of the modeled object in association with the reference three-dimensional model When,
A correlation calculation unit that calculates a correlation between the values of the plurality of feature amounts associated with the reference three-dimensional model and the values of the physical amount associated with the reference three-dimensional model;
The value of the plurality of feature values is calculated from the input value of the physical quantity based on the relative relationship calculated by the correlation calculation unit, and the value of the plurality of feature values calculated based on the data stored in the storage unit A three-dimensional shape model generation device comprising a model generation unit for generating a new three-dimensional model from
The model generation unit
The difference between the initial three-dimensional model generated using the initial value for the input value and the target three-dimensional model generated using the target value for the input value is used as a reference three-dimensional model corresponding to the initial value of the physical quantity. A three-dimensional shape model generation device characterized by generating a new three-dimensional model by adding. The storage unit
Storing a plurality of physical quantity values for each of the modeling objects;
The correlation calculation unit is
The correlation between the values of the plurality of feature amounts associated with the reference three-dimensional model and the values of the plurality of physical amounts associated with the reference three-dimensional model is calculated, and further, the values of the plurality of physical amounts are calculated. Calculate the correlation,
The model generation unit
Based on the correlation between the plurality of physical quantities, calculates a value of another physical quantity different from the physical quantity from the input value of the physical quantity, and uses the value of the other physical quantity and the input value of the physical quantity as a result of the calculation. The three-dimensional shape model generation apparatus according to claim 1, wherein a new three-dimensional model is generated. The storage unit
Storing values of a plurality of physical quantities including a length between joints for each modeling object;
The correlation calculation unit is
Calculating a correlation between the values of the plurality of feature quantities associated with the reference three-dimensional model and the values of the plurality of physical quantities associated with the reference three-dimensional model;
The model generation unit
Generating the initial three-dimensional model from the initial values of the plurality of physical quantities, and generating the target three-dimensional model from the target values of the plurality of physical quantities in which the length between the joints is the same value as the initial value. The three-dimensional shape model generation apparatus according to claim 1 or 2. Generating a reference three-dimensional model from three-dimensional shape data of a plurality of modeling objects, and extracting a plurality of features by multivariate analysis of the plurality of reference three-dimensional models;
A physical quantity value measurable by the modeled object and a plurality of feature value values of the reference three-dimensional model of the modeled object are stored in association with the reference three-dimensional model;
Calculating a correlation between the values of the plurality of feature quantities associated with the reference three-dimensional model and the values of the physical quantities associated with the reference three-dimensional model;
A value of the plurality of feature values is calculated from the input value of the physical quantity based on the calculated relative relationship, and a new three-dimensional model is calculated from the value of the calculated feature value based on the stored data. A three-dimensional shape model generation method for generating,
The difference between the initial three-dimensional model generated using the initial value for the input value and the target three-dimensional model generated using the target value for the input value is used as a reference three-dimensional model corresponding to the initial value of the physical quantity. A method for generating a three-dimensional shape model, wherein a new three-dimensional model is generated by adding the three-dimensional model. Storing a plurality of physical quantity values for each of the modeling objects;
The correlation between the values of the plurality of feature amounts associated with the reference three-dimensional model and the values of the plurality of physical amounts associated with the reference three-dimensional model is calculated, and further, the values of the plurality of physical amounts are calculated. Calculate the correlation,
Based on the correlation between the plurality of physical quantities, calculates a value of another physical quantity different from the physical quantity from the input value of the physical quantity, and uses the value of the other physical quantity and the input value of the physical quantity as a result of the calculation. The three-dimensional shape model generation method according to claim 4, wherein a new three-dimensional model is generated. Storing values of a plurality of physical quantities including a length between joints for each modeling object;
Calculating a correlation between the values of the plurality of feature quantities associated with the reference three-dimensional model and the values of the plurality of physical quantities associated with the reference three-dimensional model;
Generating the initial three-dimensional model from the initial values of the plurality of physical quantities, and generating the target three-dimensional model from the target values of the plurality of physical quantities in which the length between the joints is the same value as the initial value. The three-dimensional shape model generation method according to claim 4 or 5. Computer
A feature amount extraction unit that generates a reference three-dimensional model from three-dimensional shape data of a plurality of modeling objects and extracts a plurality of feature amounts by multivariate analysis of the plurality of reference three-dimensional models;
A storage unit that stores a value of a physical quantity measurable by the modeled object and a value of the plurality of feature values of the three-dimensional model of the modeled object in association with the reference three-dimensional model;
A correlation calculation unit for calculating a correlation between the values of the plurality of feature quantities associated with the reference three-dimensional model and the values of the physical quantities associated with the reference three-dimensional model;
The value of the plurality of feature values is calculated from the input value of the physical quantity based on the relative relationship calculated by the correlation calculation unit, and the value of the plurality of feature values calculated based on the data stored in the storage unit A 3D shape model generation program that functions as a model generation unit that generates a new 3D model from
The model generator is
The difference between the initial three-dimensional model generated using the initial value for the input value and the target three-dimensional model generated using the target value for the input value is used as a reference three-dimensional model corresponding to the initial value of the physical quantity. A program for generating a three-dimensional shape model, which is added to function as a generation unit that generates a new three-dimensional model. The storage unit
By functioning as a storage unit that stores a plurality of physical quantity values for each modeled object,
The correlation calculation unit is
The correlation between the values of the plurality of feature amounts associated with the reference three-dimensional model and the values of the plurality of physical amounts associated with the reference three-dimensional model is calculated, and further, the values of the plurality of physical amounts are calculated. Let it function as a calculation unit that calculates the correlation,
The model generator is
Based on the correlation between the plurality of physical quantities, calculates a value of another physical quantity different from the physical quantity from the input value of the physical quantity, and uses the value of the other physical quantity and the input value of the physical quantity as a result of the calculation. The three-dimensional shape model generation program according to claim 7, wherein the program functions as a generation unit that generates a new three-dimensional model. The storage unit
Function as a storage unit that stores a plurality of physical quantity values including the length between joints for each modeled object;
The correlation calculation unit is
Function as a calculation unit that calculates the correlation between the values of the plurality of feature amounts associated with the reference three-dimensional model and the values of the plurality of physical amounts associated with the reference three-dimensional model;
The model generator is
Generating the initial three-dimensional model from the initial values of the plurality of physical quantities and generating the target three-dimensional model from the target values of the plurality of physical quantities in which the length between the joints is the same as the initial value The three-dimensional shape model generation program according to claim 7 or 8, wherein the three-dimensional shape model generation program according to claim 7 or 8 is used. A three-dimensional shape model generation system comprising a three-dimensional shape model generation device and a terminal device connected via a network,
The three-dimensional shape model generation device includes:
A feature quantity extraction unit that generates a reference three-dimensional model from three-dimensional shape data of a plurality of modeling objects and extracts a plurality of feature quantities by multivariate analysis of the plurality of reference three-dimensional models;
A storage unit that stores values of physical quantities measurable by the modeled object and values of the plurality of feature values of the reference three-dimensional model of the modeled object in association with the reference three-dimensional model When,
A correlation calculation unit that calculates a correlation between the values of the plurality of feature amounts associated with the reference three-dimensional model and the values of the physical amount associated with the reference three-dimensional model;
The value of the plurality of feature values is calculated from the input value of the physical quantity based on the relative relationship calculated by the correlation calculation unit, and the value of the plurality of feature values calculated based on the data stored in the storage unit And a model generation unit for generating a new three-dimensional model from
The model generation unit
The difference between the initial three-dimensional model generated using the initial value for the input value and the target three-dimensional model generated using the target value for the input value is used as a reference three-dimensional model corresponding to the initial value of the physical quantity. To create a new 3D model by adding
The terminal device
A transmission unit that transmits the input value of the physical quantity and the three-dimensional shape data corresponding to the input value to the three-dimensional shape model generation device;
A receiving unit that receives the three-dimensional model generated by the model generating unit from the three-dimensional shape model generating device;
An output unit that outputs a three-dimensional model received by the receiving unit. The storage unit
Storing a plurality of physical quantity values for each of the modeling objects;
The correlation calculation unit is
The correlation between the values of the plurality of feature amounts associated with the reference three-dimensional model and the values of the plurality of physical amounts associated with the reference three-dimensional model is calculated, and further, the values of the plurality of physical amounts are calculated. Calculate the correlation,
The model generation unit
Based on the correlation between the plurality of physical quantities, calculates a value of another physical quantity different from the physical quantity from the input value of the physical quantity, and uses the value of the other physical quantity and the input value of the physical quantity as a result of the calculation. A new three-dimensional model is generated. The three-dimensional shape model generation system according to claim 10. The storage unit
Storing values of a plurality of physical quantities including a length between joints for each modeling object;
The correlation calculation unit is
Calculating a correlation between the values of the plurality of feature quantities associated with the reference three-dimensional model and the values of the plurality of physical quantities associated with the reference three-dimensional model;
The model generation unit
Generating the initial three-dimensional model from the initial values of the plurality of physical quantities, and generating the target three-dimensional model from the target values of the plurality of physical quantities in which the length between the joints is the same value as the initial value. The three-dimensional shape model generation system according to claim 10 or 11.