CN104021589A - Three-dimensional fitting simulating method - Google Patents

Abstract

The invention provides a three-dimensional fitting simulating method. The method comprises the following steps: generating a human model according to the height, bust-waist-hip measurements and facial features of an object; producing garment models according to garment pictures and data in a garment database, so as to form a garment model base; and matching the human model with the garment models by adding accessory models, adding scene models, and showing in a dynamic manner. The three-dimensional fitting simulating method has the beneficial effects that a three-dimensional platform is applied to a virtual website platform, and clients can browse dynamic real effects that the clients wear the selected garments in multi-angle and all-directional manners, so that the problems of goods return and the like due to the dissatisfaction of the clients to the bought garments are solved, and certain unnecessary intermediate links are greatly reduced; the method is high in authenticity and strong in persuasion in product exhibition, and the pleasure and the participation of customers during on-line shopping can be increased.

Description

A kind of three-dimensional simulation fitting method

Technical field

The invention belongs to garment industry, relate to a kind of three-dimensional simulation fitting method.

Background technology

Along with the develop rapidly of network technology, internet is changing our live and work mode just rapidly.Also changing the life style of enterprise." shopping at network " this new term is also accepted by business circles, economical circles and ordinary people gradually.This shopping way has not only adapted to purchaser's one-stop shopping demand of the supply of material fast, and met that consumer saves time, expense and easy to operate demand.Therefore, traditional shopping way is just replaced by increasing shopping online.But contrast the product of other kind, slow than in the imagination of the speed of development of Apparel Online Shopping, subject matter is that consumer has been accustomed to the buying pattern directly perceived of trying on, touching, online purchase clothes are existed to certain misgivings, and people are very big for the difference of build and sensation in virtual and actual environment.Ordinary consumer is buying before clothes, and custom is first through with it and has a try, a side but look at fit degree, the opposing party and also want to understand outward appearance and the texture after dressing.Separately have investigation to show, consumer is timid in purchase the reason of clothing on the net in, and the 85%th, because can not try on, 78% thinks and can not touch dress material or specifically see clothes, has bought easily and has regretted.Because user buys clothes on the net, suffer from and can not try on, sometimes take clothes at a glance, although very beautiful, be but not suitable for user and wear.Now existing on the market what sell that clothing website adopts is all the displaying of photo, and this is that a kind of two-dimension picture shows do not have interactive displaying, client to can't see own actual wearing effect.

Summary of the invention

The object of this invention is to provide a kind of three-dimensional simulation fitting method, solve above-mentioned one or multiple of the prior art.

The invention provides a kind of three-dimensional simulation fitting method, comprise the steps:

(1), according to the height of object, measurements of the chest, waist and hips size and facial characteristics, generate person model;

(2) by garment image and data in garment data, make dress form, form dress form storehouse;

(3) person model and dress form interpolation accessory model are arranged in pairs or groups, add model of place, represent with dynamic-form.

A kind of three-dimensional simulation fitting method provided by the invention, applies to three-dimensional platform under virtual site platform, make client can multi-angle, browse the dynamic true effect of oneself putting on selected clothes in all directions.Thereby solved client to the clothes of buying the problem such as the dissatisfied or return of goods, some unnecessary links in the middle of greatly having dwindled.On displaying merchandise, there is high fidelity, strong cogency, also can increase enjoyment and the property of participation of Consumers ' Online Shopping.

In some embodiments, set up person model, step is as follows: adopt B-spline curves curved surface to carry out human body surface-rendering; Human body feature point is corresponded in three-dimensional cartesian coordinate system, set up the 3 D human body coordinate of characteristics of human body's value; Set up characteristics of human body's curve model according to the eigenwert of corresponding human body; Auxiliary other moulding curves and the characteristic curve of characteristics of human body's curve model constructs three dimensional character model jointly.

In some embodiments, the step of input height, measurements of the chest, waist and hips size and facial characteristics generation person model also comprises as follows: person model is carried out to personalization and regulate; Build three-dimensional virtual human body deformation of free control network, human body is divided into 16 regions; Deformation of free control network is carried out to local modulation.

In some embodiments, deformation of free is a rectangular parallelepiped control framework being made up of three-dimensional vertices, and the object that will be out of shape embeds in control framework, sets up the relation between distortion framework and object to be deformed, and algorithm is as follows:

(1) in parcel is treated the control framework of changing object, build local O '-STU coordinate system, X is a certain summit in O-XYZ coordinate system, and the change type of X and local O '-STU coordinate system is:

X＝O'+sS+tT+uU

$s=\frac{T\×U\·(X-O^{\′})}{T\×U\·S}$

$t=\frac{S\×U\·(X-O^{\′})}{S\×U\·T}$

$u=\frac{S\×T\·(X-O^{\′})}{S\×T\·U}$

In formula:

O' is the true origin in O '-STU space;

The axial vector that S, T, U are local coordinate system, and for s, t, u arbitrarily in control framework, all meet 0<s<1,0<t<1,0<u<1;

(2) build and control network P _{i, j, k}, respectively along S, T, U direction of principal axis is divided into l, m, n is interval, and the position of each summit in local coordinate system is defined as

$P_{i,j,k}=O^{\&prime;}+\frac{i}{l}S+\frac{j}{l}T+\frac{k}{n}U$

In formula, i=0,1 ..., l; J=0,1 ..., m; K=0,1 ..., n.

(3) according to after the relation between distortion framework and object to be deformed, change P _{i, j, k}position cause distortion, the local coordinate that has an O' is (s, u, t), X _ffdfor any point in local O '-STU coordinate system, local O '-STU coordinate is tied to the conversion type of O-XYZ coordinate system and is

$X_{\mathrm{ffd}}=\underset{i=0}{\overset{l}{\mathrm{\&Sigma;}}}\underset{j=0}{\overset{m}{\mathrm{\&Sigma;}}}\underset{k=0}{\overset{n}{\mathrm{\&Sigma;}}}{P}_{i,j,k}{B}_{\mathrm{il}}\left(s\right)B_{\mathrm{jm}}\left(t\right)B_{\mathrm{kn}}\left(u\right)$

In formula, B _il(s), B _jm(t), B _kn(u) be respectively l, m, m Bezier basis function.

In some embodiments, the facial characteristics that person model relates to is that user is by taking photograph, directly uploading local picture or utilize instrument and equipment scanning face to generate, step is as follows: user's browser of entering the Web page, take pictures, upload local photo or carry out face scanning by on-the-spot scanning device with camera, obtain a face face picture.

In some embodiments, the modeling procedure of model of place is as follows:

(1) automatically generate or import landform, data texturing corresponding to landform is set;

(2) by vector quantization data importing ground material object or by importing third-party three-dimensional modeling data, complete the generation of part ground object model in scene;

(3) by increasing particIe system, in scene, realize playing up of special-effect, make scene seem truer.

Brief description of the drawings

Fig. 1 is a kind of three-dimensional simulation fitting method of the present invention realization flow figure;

Fig. 2 is that a kind of three-dimensional simulation fitting method of the present invention is set up mask process flow diagram;

Fig. 3 is the local O '-STU coordinate system of a kind of three-dimensional simulation fitting method of the present invention.

Embodiment

The invention provides a kind of three-dimensional simulation fitting method, comprise the steps:

(1), according to the height of object, measurements of the chest, waist and hips size and facial characteristics, generate person model 1;

(2) by garment image and data in garment data, make dress form 2, form dress form storehouse;

(3) person model 1 and dress form 2 are added to accessory model 3 and arrange in pairs or groups, add model of place 4, represent with dynamic-form.

First user has the process of a registration, need registered user name and password, then will input self height, body weight and measurements of the chest, waist and hips size, after the login afterwards of these data, can directly demonstrate, do not need to repeat input, and these data also can revise, because some people's build can be always not constant.

As shown in Figure 1, a kind of three-dimensional simulation fitting method of the present invention, in step S1, set up person model 1, dress form 2, accessory model 3 and model of place 4, dress form 2, accessory model 3 and model of place 4 form respectively clothes storehouse, accessory storehouse and scene library; In step S2, person model 1, dress form 2, accessory model 3 and model of place 4 are imported to graphical interaction software Unity3D; In step S3, all models carry out alternately, in step S4, after model completes alternately, carrying out virtual display in graphical interaction software Unity3D.

First use B-spline surface method to build three dimensional character model 1, B-spline curves equation is defined as: curve, the local shape of curve is subject to the control of respective vertices very directly perceived.First adopt B-spline curves curved surface to carry out human body surface-rendering, using human body front waist mid point as true origin, set up three-dimensional cartesian coordinate system; Human body feature point is corresponded in three-dimensional cartesian coordinate system, set up the 3 D human body coordinate of characteristics of human body's value; Set up characteristics of human body's curve model, and characteristics of human body is worth corresponding human body, the data point that the matched curve of the size table of useful person model 1 and model sample point and each characteristic portion obtains is come reverse reference mark, then comes with B-spline curves curved surface the reference mark that matching is tried to achieve; In concrete modeling process, human body is divided into some patchs, obtain the three dimensional character model with several characteristic curvees by B-spline curves matching, the method of recycling interpolation, is aided with other moulding curves, jointly constructs three dimensional character model 1 with characteristic curve.

In addition, by program code, height and measurements of the chest, waist and hips data are converted into the three-dimensional data that can be modeling software identification, input the person model 1 that different height and measurements of the chest, waist and hips data will produce different size, this can carry out personalization to person model 1 and regulate, select deformation of free to carry out Region control adjusting, deformation of free is a flexible plastic distortion framework, deformation of free is less demanding for model, do not need to carry out the model construction in early stage, only need common triangle mesh model.The basic thought of deformation of free is to build a rectangular parallelepiped control framework being made up of three-dimensional control vertex, and the object that will be out of shape embeds in control framework, and mobile reference mark makes frame deformation, and the object of embedding also can be out of shape thereupon.Specific algorithm step is as follows:

(1) in parcel is treated the control framework of changing object, build local O '-STU coordinate system (as shown in Figure 3), X is a certain summit in O-XYZ coordinate system, and the change type of X and local O '-STU coordinate system is:

X＝O'+sS+tT+uU

$s=\frac{T\&times;U\&CenterDot;(X-O^{\&prime;})}{T\&times;U\&CenterDot;S}$

$t=\frac{S\&times;U\&CenterDot;(X-O^{\&prime;})}{S\&times;U\&CenterDot;T}$

$u=\frac{S\&times;T\&CenterDot;(X-O^{\&prime;})}{S\&times;T\&CenterDot;U}$

In formula: O' is the true origin in O '-STU space; The axial vector that S, T, U are local coordinate system, and for s, t, u arbitrarily in control framework, all meet 0<s<1,0<t<1,0<u<1.

(2) build and control network P _{i, j, k}, respectively along S, T, U direction of principal axis is divided into l, m, n is interval, and the position of each summit in local coordinate system is defined as

$P_{i,j,k}=O^{\&prime;}+\frac{i}{l}S+\frac{j}{l}T+\frac{k}{n}U$

In formula, i=0,1 ..., l; J=0,1 ..., m; K=0,1 ..., n.

(3) when after the relation of having set up between this distortion framework and object to be deformed, can be by changing P _{i, j, k}position cause distortion, the local coordinate of supposing to have an O' is (s, u, t), X _ffdfor any point in local O '-STU coordinate system, its transform mode that is tied to O-XYZ coordinate system by local O '-STU coordinate is as follows:

$X_{\mathrm{ffd}}=\underset{i=0}{\overset{l}{\mathrm{\&Sigma;}}}\left[\begin{array}{c}l\\ i\end{array}\right]{(1-s)}^{l-i}{s}^i\left[\begin{array}{c}\end{array}\underset{j=0}{\overset{m}{\mathrm{\&Sigma;}}}\left[\begin{array}{c}m\\ j\end{array}\right]{(1-t)}^{m-j}{t}^j\left[\underset{k=0}{\overset{n}{\mathrm{\&Sigma;}}}\begin{array}{c}\end{array}\left[\begin{array}{c}n\\ k\end{array}\right]{(1-u)}^{n-k}{u}^k{P}_{i,j,k}\right]\right]$

Can simplify and be expressed as

$X_{\mathrm{ffd}}=\underset{i=0}{\overset{l}{\mathrm{\&Sigma;}}}\underset{j=0}{\overset{m}{\mathrm{\&Sigma;}}}\underset{k=0}{\overset{n}{\mathrm{\&Sigma;}}}{P}_{i,j,k}{B}_{\mathrm{il}}\left(s\right)B_{\mathrm{jm}}\left(t\right)B_{\mathrm{kn}}\left(u\right)$

In formula, B _il(s), B _jm(t), B _kn(u) be respectively l, m, m Bezier basis function.

Virtual portrait model partition based on H-Anim normalization is head, neck, trunk, belly, right upper arm, right forearm, the right hand, left upper arm, left forearm, left hand, right thigh, right leg, right crus of diaphragm, left thigh, left leg and 16 regions of left foot, carries out Region control.Concrete steps:

(1) build three-dimensional virtual human body deformation of free control network, human body is divided into above-mentioned 16 regions.

(2) deformation of free control network is carried out to local modulation.

If every part is Part _p, by calculating this part of three dimensional point cloud Point[], build respectively local coordinate system.As follows to single part concrete grammar:

1) from the Part of person model 1 _pfor starting point, p=1;

2) with Part _pthree dimensional point cloud Point[], the maximal value Max that obtains in X-direction (Point[] .x), the maximal value Max that obtains in Y direction (Point[] .y), the maximal value Max that obtains in Z-direction (Point[] .z), obtain in X-axis, Y-axis, the minimum M in of Z-direction (Point[] .x), Min (Point[] .y), Min (Point[] .z), O'(Min (Point[] .x), Min (Point[] .y), Min (Point[] .z)), S (Max (Point[] .x)-Min (Point[] .x), 0, 0), T (0, Max (Point[] .y)-Min (Point[] .y), 0), U (0,0, Max (Point[] .z)-Min (Point[] .z)),

3) change control framework P _{i, j, k}, again use formula calculate at new P _{i, j, k}the coordinate on lower summit;

4) repeating step 1, and p+1, until P=16.

As shown in Figure 2, set up after person model 1, in order to reach effect more true to nature, add high human simulation face mask, user is by taking photograph, directly upload local picture or utilize instrument and equipment scanning face face to generate, specific as follows, in step 10, user's browser of entering the Web page, in step S11, take pictures by camera, in step S12, user uploads local photo, in step S13, carry out face scanning by on-the-spot scanning device, obtain a face face picture, in step S14, the picture obtaining by different modes imports server, in step S15, start daemon software, implementation procedure is as follows: use active shape model (Active Shape Model, ASM) algorithm detects 65 unique points on face automatically, select the master pattern of a general human face model as native system face D facial modelling, and on master pattern, calibrate grid feature summit corresponding to 65 unique points going out with image detection for grid adaptation, the unique point detecting on photo is by projective transformation in three dimensions, and as x and the y value of unique point on the personalized model that is about to generate, depth value z adopts the experience z value of master pattern corresponding point, the 3D unique point of demarcating is before moved to above-mentioned projected position, all the other 3D net points are by radial basis function (Radial Basis Function, RBF) interpolation algorithm carries out interpolation adjustment, and master pattern is just adapted to for personalized human face mask like this, from photo, take out the texture of face facial parts as face mask corresponding part, calculate the texture coordinate on summit on personalized mask, personalized geometric model is carried out to pinup picture, in step S16, finally generate face mask.

ASM algorithm is a kind of based on statistical active appearance models, (a) first choosing lineup's face full face manually demarcates, the unique point that some is evenly distributed is chosen in the position such as profile and pupil at face mask and eyebrow, eyes, nose, face, each like this shape can be described by one group of calibration point, all use 65 profiles that remove to mark face and face for every piece image, form some distributed model.(b), due to the difference of size and the shape etc. of each image, after artificial mark, need the normalization of aliging of the figure of each image labeling.Alignment is not changing under the prerequisite of nominal shape exactly, by rotation, zooming and panning operation, the position of two width images is approached as far as possible, shape alignment procedure is as follows: first choose a comparatively desirable shape vector as initial sample, other all vector is alignd with it; Afterwards, the average shape vector after the alignment calculating being done to normalization processes.And using normalized average shape vector as sample, then by the average shape vector alignment therewith of other shape vector through last alignment; Repeat this process, until the average shape of adjacent twice vector difference is less than a certain particular value, alignment procedure finishes.And people's face shape of manually demarcating is above normalized to operation, in elimination picture, face is due to different angles, distance is far and near, the non-shape that the extraneous factors such as posture changing cause is disturbed, thereby make invocation point distributed model more effective, in simple terms, be exactly by suitable translation series of points distributed model, rotation, scale transformation, on the basis that does not change a distributed model, snap to same some distributed model, thereby change under the rambling state of raw data obtaining, reduce the interference of non-shape factor, after normalization, can obtain a Nature face model.(c) utilize local gray level model, each unique point for Nature face model is carried out iterative search along the normal direction of profile, also be the mahalanobis distance that each point within the scope of this is calculated to the average normalization gray scale derivative vector of its normalization gray scale derivative vector and this point, therefrom the candidate point of selected distance minimum is as optimal match point.Each unique point in model is carried out to the search of optimal match point, obtaining so a new shape vector is X '.In search procedure, first in lower resolution image, carry out signature search, then rising to step by step high one-level image in different resolution searches for, under lower resolution, owing to having ignored some details of image and the scope of search is larger comparatively speaking, once can obtain the matching result comparatively desirable to overall profile.Under high-resolution, can reduce hunting zone, thereby obtain the comparatively desirable result of local matching, in each iteration, constantly adjust parameter, make whole model constantly adapt to the Reasonable Shape that target shape can keep again face.

In addition, the garment image providing by clothes business and data, make dress form 2, forms dress form storehouse.Adopt method of geometry to set up dress form 2, the dress form 2 that Geometric Modeling Method is set up is simple in structure, and the size of dress form 2 and also easily amendment of size.Then in dress form 2, also need to carry out cloth simulation, certainly, first cloth simulation will be understood the physical characteristics of fabric, comprises the physical characteristics after the physical characteristics of fabric when static and fabric are subject to external force and deform in the time of motion.Easily realize and spring-Mass Model is simple, operation efficiency is higher, is a kind of simulate fabric formative method.

Accessory model 3, as knapsack, wrist-watch etc., adopts conventional modeling means, as polygon modeling, carries out modeling.

Same clothes have different effects under different scenes, set up multiple models of place 4, form scene library as indoor, outdoor etc., and the selection that user can be various, to buy the clothes of admiring most.Set up scene step as follows:

(1) automatically generate or import landform, data texturing corresponding to landform is set.Normally graphic data is digitizing elevation model data, and data texturing corresponding to landform be the image data in this scene normally.

(2) by vector quantization data importing buildings, trees etc., complete the generation of part ground object model in scene.Add the texture that imports buildings, make them seem enough true.Be often referred to buildings, trees etc. and have some common features, it is the object that repeatability is high, in real-time virtual system, they should use the least possible face, so this class model mainly creates three-dimensional modelling effect with the transparent pinup picture of two-dimentional models coupling in the time setting up, replace detail model with pinup picture, and be to rotate around viewpoint the setup of attribute of model, so no matter from which angle browse, can see whole object, thereby be effective object true to nature.

(3), for some Special complex models, can adopt and import third-party three-dimensional modeling data.As 3DS data etc.

(4), in order to make scene seem truer, can play up scene as illumination, material, texture, atomization, increase sky etc. according to sense of reality technology.

(5) by increasing particIe system, in scene, realize playing up of the special-effect such as fireworks, snowflake, make scene seem truer.

As shown in Figure 1, person model 1, dress form 2, accessory model 3 and model of place 4 are imported to graphical platform Unity3D, stress model and attribute information.Then graphical platform Unity3D carries out alternately person model 1, dress form 2, accessory model 3 and model of place 4, first being alternately and wearing process of dress form 2 and person model 1, in clothes modeling process, on the each pieces of cloth of dress form 2, drawing some directed line segments serves as a mark, on person model 1, also draw mark at correspondence position simultaneously, by these marks and mapping algorithm, system by clothing sleeve through on person model 1.

In the present invention, wear process and do not need manual assigned tags.Notation and dress form 2 are bound together, use the coordinate sequence from reference position to final position summit to specify the oriented broken line serving as a mark, these coordinates are oriented broken line end points relative positions on pieces of cloth, are shone upon and can under three-dimensional situation, be picked up very easily these points by local coordinate system.Each coordinate corresponding on person model 1 is showed with coordinate sequence simultaneously.All these information are stored in dress form 2 files, when reading model data, also will be written into together.Utilize mapping algorithm, dependence modeling mark is mapped to two-dimentional clothes pieces of cloth the three dimensions at virtual portrait model 1 place.

After clothes have worn, need to implement towing adjustment to clothes according to fitting, towing adjustment is the process of an interactive fine setting, is that dress form 2 is done to finer adjustment.According to the needs of fitting, user can by towing adjust the sleeve of clothes mention some or collar pull open some etc.The realization that towing is adjusted is on the basis of system surfaces towing algorithm, to introduce towing fixed point mechanism to complete.According to the style difference of clothes, for clothes are specified different fixed points, the rationality of clothes deformation when these fixed points are used for guaranteeing towing adjustment.

In order to allow, dress form 2 surfaces are more smooth has authenticity more, implements a formation process and makes each dihedral angle in dress form 2 more approach 180 degree.

Dress form 2 covers are through after virtual portrait model 1, use graphical platform Unity3D, and virtual portrait model 1 imports in model of place 4, and initial position is set, and add script control to person model 1.Such as, in streetscape, personage needs walking, can use role controller, action is set corresponding with the directionkeys on keyboard, uses keyboard just can realize the walking of virtual portrait different directions.So just form personage's roaming scence, no longer just 1 rotation of virtual portrait model waits simple motion.

In addition, when registered members, system can point out user to input height, body weight and measurements of the chest, waist and hips size, the data of input in the time selecting before clothing that system can be registered according to user like this, select a person model 1 similar to user, then by edit-modify person model 1, to obtain more accurate model.

Shopping at network has not been singly the behavior of a certain types of populations, and therefore, virtual fitting technology also will meet the demand of special population.In order be to be better customer service, online fitting room can also be sorted out client like data class, in registered members, add the input at age, can inquire age bracket from background data base, when then client purchases clothing on the net, can send out pushed information, as recommend some clothing collocation of such client, inform collocation that other client of such this class of client likes etc., provide more selection to client.

On fitting website, not only can try on a dress, also can add some accessories, as shoes, jewellery etc.When fitting, the thing of taking a fancy to arbitrarily can be arranged in pairs or groups, this is in market shopping, to be difficult to realize.After having tried clothes, just can settle a purchase on the net at once.For working clan, have between network fitting, can save a lot of time, remove the fatigue of going window-shopping and time-consuming from.

A kind of three-dimensional simulation fitting method provided by the invention, applies to three-dimensional platform under virtual site platform, makes client can see in real time the true effect of oneself putting on selected clothes.Thereby solved client to the clothes of buying the problem such as the dissatisfied or return of goods, some unnecessary links in the middle of greatly having dwindled.On displaying merchandise, there is high fidelity, strong cogency, also can increase enjoyment and the property of participation of Consumers ' Online Shopping.

Claims (6)

1. a three-dimensional simulation fitting method, wherein, comprises the steps:

(1), according to the height of object, measurements of the chest, waist and hips size and facial characteristics, generate person model;

(2) by garment image and data in garment data, make dress form, form dress form storehouse;

(3) person model and dress form interpolation accessory model are arranged in pairs or groups, add model of place, represent with dynamic-form.

2. three-dimensional simulation fitting method according to claim 1, wherein, described generation person model step is as follows:

(1) adopt B-spline curves curved surface to carry out human body surface-rendering;

(2) human body feature point is corresponded in three-dimensional cartesian coordinate system, set up the 3 D human body coordinate of characteristics of human body's value;

(3) set up characteristics of human body's curve model according to the eigenwert of corresponding human body;

(4) auxiliary other moulding curves and the characteristic curve of characteristics of human body's curve model constructs three dimensional character model jointly.

3. three-dimensional simulation fitting method according to claim 1, wherein, the step that input height, measurements of the chest, waist and hips size and facial characteristics generate described person model also comprises as follows: described person model is carried out to personalization and regulate; Build three-dimensional virtual human body deformation of free control network, human body is divided into 16 regions; Deformation of free control network is carried out to local modulation.

4. three-dimensional simulation fitting method according to claim 3, wherein, described deformation of free is a rectangular parallelepiped control framework being made up of three-dimensional vertices, and the object that will be out of shape embeds in control framework, set up the relation between described distortion framework and object to be deformed, algorithm is as follows:

(1) in parcel is treated the control framework of changing object, build local O '-STU coordinate system, X is a certain summit in O-XYZ coordinate system, and the change type of X and local O '-STU coordinate system is:

X＝O'+sS+tT+uU

$s=\frac{T\&times;U\&CenterDot;(X-O^{\&prime;})}{T\&times;U\&CenterDot;S}$

$t=\frac{S\&times;U\&CenterDot;(X-O^{\&prime;})}{S\&times;U\&CenterDot;T}$

$u=\frac{S\&times;T\&CenterDot;(X-O^{\&prime;})}{S\&times;T\&CenterDot;U}$

In formula:

O' is the true origin in O '-STU space;

The axial vector that S, T, U are local coordinate system, and for s, t, u arbitrarily in control framework, all meet 0<s<1,0<t<1,0<u<1;

(2) build and control network P _{i, j, k}, respectively along S, T, U direction of principal axis is divided into l, m, n is interval, and the position of each summit in local coordinate system is defined as

$P_{i,j,k}=O^{\&prime;}+\frac{i}{l}S+\frac{j}{l}T+\frac{k}{n}U$

In formula, i=0,1 ..., l; J=0,1 ..., m; K=0,1 ..., n;

(3), according to the relation between described distortion framework and object to be deformed, change P _{i, j, k}position cause distortion, the local coordinate that has an O' is (s, u, t), X _ffdfor any point in local O '-STU coordinate system, it is as follows that local O '-STU coordinate is tied to the conversion regime of O-XYZ coordinate system:

$X_{\mathrm{ffd}}=\underset{i=0}{\overset{l}{\mathrm{\&Sigma;}}}\underset{j=0}{\overset{m}{\mathrm{\&Sigma;}}}\underset{k=0}{\overset{n}{\mathrm{\&Sigma;}}}{P}_{i,j,k}{B}_{\mathrm{il}}\left(s\right)B_{\mathrm{jm}}\left(t\right)B_{\mathrm{kn}}\left(u\right)$

In formula, B _il(s), B _jm(t), B _kn(u) be respectively l, m, m Bezier basis function.

5. three-dimensional simulation fitting method according to claim 1, wherein, the facial characteristics that described person model relates to is that user is by taking photograph, directly uploading local picture or utilize instrument and equipment scanning face to generate, step is as follows: user's browser of entering the Web page, take pictures, upload local photo or carry out face scanning by on-the-spot scanning device with camera, obtain a face face picture.

6. three-dimensional simulation fitting method according to claim 1, wherein, the modeling procedure of described model of place is as follows:

(1) automatically generate or import landform, data texturing corresponding to landform is set;

(2) by vector quantization data importing ground material object or by importing third-party three-dimensional modeling data, complete the generation of part ground object model in scene;

(3) by increasing particIe system, in scene, realize playing up of special-effect.