CN106952334B - The creation method and three-dimensional fitting system of the net model of human body - Google Patents
The creation method and three-dimensional fitting system of the net model of human body Download PDFInfo
- Publication number
- CN106952334B CN106952334B CN201710079459.9A CN201710079459A CN106952334B CN 106952334 B CN106952334 B CN 106952334B CN 201710079459 A CN201710079459 A CN 201710079459A CN 106952334 B CN106952334 B CN 106952334B
- Authority
- CN
- China
- Prior art keywords
- human body
- model
- net model
- net
- characteristic parameter
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0641—Shopping interfaces
- G06Q30/0643—Graphical representation of items or shoppers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2200/00—Indexing scheme for image data processing or generation, in general
- G06T2200/08—Indexing scheme for image data processing or generation, in general involving all processing steps from image acquisition to 3D model generation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2210/00—Indexing scheme for image generation or computer graphics
- G06T2210/16—Cloth
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Geometry (AREA)
- Computer Graphics (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Processing Or Creating Images (AREA)
Abstract
The present invention provides a kind of creation methods of net model of the human body based on depth camera, comprising the following steps: S1: obtaining at least two amplitude deepness images of the human body under at least two given poses using depth camera；S2: the characteristic parameter of the net model of human body is obtained from least two amplitude deepness images；S3: the net model of human body is created according to the characteristic parameter.The method that the present invention creates the net model of human body can be to avoid loose clothing bring manikin error.In addition, based on the creation method of the net model of above-mentioned human body, the present invention also proposes a kind of three-dimensional fitting method and three-dimensional fitting system, and no matter human body wears loose clothing or compact clothing, it, can in real time and 360 ° show the effect worn the clothes using three-dimensional fitting method of the invention or system.
Description
Technical field
The present invention relates to optical technology and field of computer technology, and in particular to a kind of net mould of human body based on depth camera
The creation method and three-dimensional fitting system of type.
Background technique
With shopping at network, the arriving in private customization epoch, tried on from traditional solid shop/brick and mortar store-purchasing model gradually can be to net
Network tries/individual's customization-purchasing model transition on.At present in prior art, network fitting remains in the two-dimensional fitting stage, although
There are many three-dimensional fitting applications, but the precision and real-time actually fitted are still undesirable.The creation of human 3d model is to realize
Network fitting and the premise of private customization accurately acquire the net model of human body and help to realize the accurate of suit length collocation
Property, while the human body hundreds of characteristic attributes for characterizing human appearance can be disposably obtained using the net model of human body, so as to quilt
For realizing that the private of clothing customizes service.
It is a kind of preferably selection that human 3d model is obtained currently with the depth camera of consumer level, on the one hand can be with
Save the cost, the millimetre-sized measurement accuracy of another aspect consumer level depth camera also meet the requirement of the net model of human body enough.To the greatest extent
Pipe is in this way, still face some problems currently based on the fitting of human 3d model: the accurate acquisition of the net model of human body needs human body
The clothes of compact is worn, however in most cases, the personnel of being measured often wear loose clothes, wear in human body wide
When loose clothing, current fitting technology still cannot accurately obtain the net model of human body.
Summary of the invention
The technical problem to be solved in the present invention is that: existing fitting scheme is difficult to when human body wears loose clothing accurately
Obtain the net model of human body.
In order to solve the above technical problems, the present invention proposes a kind of creation method of net model of human body, comprising the following steps:
S1: at least two amplitude deepness images of the human body under at least two given poses are obtained using depth camera；
S2: the characteristic parameter of the net model of human body is obtained from least two amplitude deepness images；
S3: the net model of human body is created according to the characteristic parameter.
Preferably, the given pose in the step S1 refers under the given pose, the clothing of human body at least one portion
Be in compact condition.
Preferably, the step S2 includes:
S21: the characteristic parameter at each compact position of human body in the depth image under each given pose is extracted；
S22: by recoverable force operation or the method for establishing empirical equation, by the spy at position compact under each given pose
Sign parameter person is converted into the actual characteristic parameter of human body；
S23: the characteristic parameter after the actual characteristic parameter of human body after all conversions is summarized as the net model of human body.
Preferably, recoverable force operation described in the step S22 specifically includes:
(1) point cloud data at compact position under given pose is obtained, and is had after denoising and cavity filling
The point cloud data of high quality, and save the three-dimensional Euclidean coordinate information of each point, according to point cloud data obtain grid model M=(V,
E), wherein V=(v1,v2,…,vn)TIndicate that the matrix being made of the three-dimensional coordinate on vertex each in model, E indicate in model
All sides；
(2) canonical matrix S, the calculation formula S=CV for relativeness between each cloud in descriptive model are calculated,
Wherein C=(I-D-1It B) is transformation matrix, I is unit matrix, and D is diagonal matrix, the element D on diagonal lineii=di, diFor with
Point viAdjacent vertex number, B matrix can be expressed from the next:
(3) multiple deformation obligatory points are selected, it, will by obligatory point Euclidean coordinate new after the available recoverable force of calculating
New obligatory point coordinate is as restrictive condition and is added to transformation matrix, becomes new transformation matrix C ', finally, using public
Formula V '=C '-1S solves deformed vertex Euclidean coordinate.
Preferably, the step S3 includes:
S31: standardized human body's model is established；
S32: standardized human body's model is modified to obtain the net model of human body according to the characteristic parameter of the net model of the human body.
Preferably, the depth camera in the step S1 is based on structure light trigonometry, time flight method or binocular vision
The one of which of principle.
Preferably, human depth's image in the step S1 is to be rotated one week by human body around single depth camera
What mode obtained, or by using the synchronous acquisition of the multiple depth cameras for being distributed in human peripheral with different angle.
Preferably, the method for modification standardized human body's model in the step S32 is the method based on shaft distortion principle, or
Person is the method by carrying out successive ignition to SCAPE model.
Based on the creation method of the net model of above-mentioned human body, the present invention also proposes a kind of system for establishing the net model of human body, packet
Memory is included, for storing program；Processor runs described program, for controlling the system for establishing the net model of human body
Execute the above-mentioned method for establishing the net model of human body.
The present invention also proposes that a kind of computer readable storage medium comprising computer program, the computer program can be grasped
Make to make computer execute the above-mentioned method for establishing the net model of human body.
Based on the creation method of the net model of above-mentioned human body, the present invention also proposes a kind of three-dimensional fitting method, including following step
It is rapid:
T1: according to the creation method of the net model of above-mentioned human body, the net model of human body is created；
T2: creation clothing model；
T3: clothes effect is shown after clothing model is synthesized with the net model of human body.
The present invention also proposes a kind of three-dimensional fitting system, including memory, for storing program；Processor, described in operation
Program, to execute above-mentioned 3D fitting method for controlling the 3D dressing system.
The present invention also proposes that a kind of computer readable storage medium comprising computer program, the computer program can be grasped
Make to make computer execute above-mentioned 3D fitting method.
The invention has the benefit that the present invention provides a kind of creation sides of net model of the human body based on depth camera
Method, compared with prior art, the present invention can accurately obtain characteristics of human body's parameter by multiple given poses, then according to people
Body characteristics parameter obtains the net model of human body, can be to avoid loose clothing bring manikin error using method of the invention.
In addition, the creation method based on the net model of above-mentioned human body, the present invention also proposes a kind of three-dimensional fitting method and three-dimensional
Dressing system, no matter human body wears loose clothing or compact clothing, using three-dimensional fitting method of the invention or system, equal energy
The effect that enough real-time and 360 ° of displayings are worn the clothes.
Detailed description of the invention
Fig. 1 is the general flow chart of the net model creation method of human body in the specific embodiment of the invention.
Fig. 2 is the sub-process figure of the step S2 of the net model creation method of human body in the specific embodiment of the invention.
Fig. 3 is the sub-process figure of the step S3 of the net model creation method of human body in the specific embodiment of the invention.
Fig. 4 is the flow chart of three-dimensional fitting method in the specific embodiment of the invention.
Specific embodiment
It is further described below with reference to drawings and the specific embodiments.
1, the creation method of the net model of human body
The creation method of the net model of human body, as shown in Figure 1, comprising the following steps: S1: obtaining human body using depth camera and exist
At least two amplitude deepness images under at least two given poses；S2: the characteristic parameter of the net model of human body is obtained；S3: building human body
Net model.Wherein, at least two width depth of the human body under at least two given poses are obtained using depth camera in step sl
Image, the depth camera are based on the trigon depth camera of structure light, and the given pose refers to human body at least one
The clothing of a part is in compact condition.Above-mentioned steps will be described in detail below.
S1: at least two amplitude deepness images of the human body under at least two given poses are obtained using depth camera
Camera currently used for obtaining depth image mainly has based on structure light trigonometry, time flight method or binocular vision
Feel the depth camera of principle.
Encoded normal structure is projected into space using laser-projector based on the trigon depth camera of structure light
Light pattern, the difference of target depth is modulated normal structure light pattern in space, is passed through image correlation scheduling algorithm and is obtained
The difference of modulated structure light image and normal structure light pattern establishes the difference and target depth according to structure light trigonometry
Between relationship can solve the depth image of entire object space.
Depth camera based on time flight method utilizes Laser emission instrument to objective emission laser pulse, by optical receiver apparatus
It obtains pulse and records and be emitted to the received smooth flight time, the depth image of target can be calculated according to the flight time.
Depth camera based on Binocular Vision Principle, substantially similar to structure light trigonometry principle, difference is structure
Light trigonometry is actively to measure, and binocular vision is then passive measurement.Difference of the image obtained using left and right camera on parallax
Not, and the depth value for further calculating target after the parallax using triangulation principle is obtained by vision algorithm.
Three kinds of each superiority and inferiority of depth acquisition methods, cost is relatively low for structure light trigonometry, and depth acquisition is high-efficient, but multiple phases
Machine can have interference when measuring simultaneously to an object space.And the depth camera higher cost of time flight method.Binocular
The depth camera algorithm of vision is complicated, and has certain requirement to environment locating for target.Thus, for specifically using ring
Border can suitably choose different depth cameras.
In this embodiment, using the depth map for obtaining human body based on the trigon depth camera of structure light
Picture.
In other specific embodiments, the depth camera based on time flight method or Binocular Vision Principle can be used
To obtain the depth image of human body.
Generally, manikin should include 360 ° of field angle.And the field angle of single depth camera is limited, also has thus
A variety of alternative human depth's image acquisitions schemes.
First is that tested human body allows depth camera to obtain comprising human body by way of rotation one week using single depth camera
The multi-amplitude deepness image of 360 ° of information.Whole the three of human body can be extracted according to multi-amplitude deepness image using image registration algorithms
Tie up point cloud data.This method is difficult to realize real-time human body entirety three-dimensional data and obtains, and advantage is that cost is relatively low.
Second is that using the multiple depth cameras for being distributed in human peripheral, synchronously by multiple depth cameras of different angle
The multi-amplitude deepness image for obtaining reflection all three-dimensional informations of human body, obtains human body after finally being registered using multi-amplitude deepness image
Whole three dimensional point cloud.This method be substantially solved using the quantitative advantage of depth camera can not real-time mention
The problem of taking, the cost paid are higher costs.
In this embodiment, the quantitative selection of depth camera does not limit, and single depth camera extracts human body
The speed of feature is slower than multiple depth cameras.
The net model that human body can be accurately acquired is the basis for carrying out next step fitting, however in most situations
Under, the personnel of being measured often wear loose clothes, and obtained three-dimensional (3 D) manikin not can accurately reflect the truth of human body, take off
Clothing or the measurement that is in tights are not the preferred plan for solving the problems, such as this.
In addition, the registration of depth image needs to expend more computing resource, data demand is more accurate, the real-time of system
It will be poorer.There is presently no the consumer level solutions for capableing of real-time acquisition human body three-dimensional overall model.
In order to solve the problems, such as real-time and measurement accuracy, the scheme that present embodiment uses is, from depth camera
The characteristic information that can accurately reflect human body is extracted in human depth's image of acquisition, is created that virtually according to characteristic information
Can reflect the net model of the human body of human body three-dimensional feature.Once establishing after the net model of human body, the subsequent reality in fitting process
When display then need not be calculated frame by frame, only need to extract human skeleton frame by frame, by the posture of the net model of skeleton driven human body to reach
To the requirement of real-time.
S2: the characteristic parameter of the net model of human body is obtained
The situation (the non-thick clothes of loose clothes) when human body wears loose clothes is considered herein, it is similarly suitable to wear compact clothing
Situation when object.In addition, characteristics of human body here refers to characteristics of human body of the human body under normal stand posture.Human body wears loose clothing
When taking, many of body position can be covered by clothes, and covered position is divided to two kinds, and one kind is personal covering, Yi Zhongwei
Non- personal covering.Defaulting the personal model for covering (compact) position herein is the net model of actual human body, rather than personal covering
The model at (loose) position then differs greatly with the net model of human body.
In the step of obtaining the characteristic parameter of the net model of human body, as shown in Fig. 2, again including following three sub-steps:
S21: the characteristic parameter at each compact position of human body in the depth image under each given pose is extracted；S22: extensive by deforming
The characteristic parameter person at position compact under each given pose is converted human body reality by multiple operation or the method for establishing empirical equation
Characteristic parameter；S23: join after the actual characteristic parameter of human body after all conversions is summarized as the feature of the net model of human body
Number.Above-mentioned steps will be described in detail below.
S21: the characteristic parameter at each compact position of human body in the depth image under each given pose is extracted
Non- personal covering can be allowed to become personal covering by different gesture actions, for example with arms akimbo can then allow waist
Portion becomes personal and covers, lifts shank calf can be allowed to become personal and cover, lift a hand and bend over to side, can allow
Chest side becomes personal covering etc..
Thus, the posture for being able to reflect all characteristics of human body is previously set, tested human body is allowed to do posture one by one
Obtain the physical characteristic data that the net model of all human bodies needs.
Following table illustrates part posture and corresponding characteristics of human body.
| Serial number | Posture | Personal position |
| 1 | Uprightly | Shoulder |
| 2 | With arms akimbo | Waist lateral dimension |
| 3 | Lift thigh | Thigh |
| 4 | Lift shank | Shank |
| 5 | It raises one's hand, bend over to side | Arm, chest and waist side |
| 6 | Forwardly bend over | Back, buttocks |
| 7 | It squares one's shoulders | Chest, abdomen |
| … | … |
The characteristic of human body generally comprises: height, shoulder height, shoulder breadth, waistline, bust, thigh and calf profile, arm geometry etc.
Deng.
Additionally include the characteristic point at each position, i.e. artis.The extraction of artis can be based on similar kinect
The algorithm that SDK middle skeleton is extracted, that is, first pass through positioning head and human body trunk, then utilizes deep learning (K-Tree) algorithm
Orient other skeletal joint point positions.
S22: by recoverable force operation or the method for establishing empirical equation, by the spy at position compact under each given pose
Sign parameter person is converted into the actual characteristic parameter of human body
The characteristic of human body obtains aspect, directly acquired under specific posture be characterized in be used directly to as
Manikin feature.Such as when human body is forwardly bent over, the back of acquisition and the feature (width, cross section etc.) of buttocks
It is all through deformed data.It is special deformed data can be obtained into the human body under normal attitude by recoverable force operation
Sign.Alternatively, it is also possible to the methods of by machine learning, establish a kind of simple empirical equation, under Lai Fanying human body given pose
The relationship of characteristics of human body and actual characteristic.Both methods is illustrated separately below, leg muscle portion when being squatted down with human body
For the deformation of position, indirect gain is needed to be characterized in that thigh includes the multiple of both ends (thigh top and with shank junction)
The perimeter of cross section, it should be noted that not being here is to squat down completely, production when having contact in order to avoid the thigh back side and shank
The case where raw back data can not obtain.
(1) recoverable force
Firstly, obtain squat down after huckle point cloud data, and by denoising and cavity filling after obtain have it is higher
The point cloud data of quality, and the three-dimensional Euclidean coordinate information of each point is saved, grid model M=(V, E) is obtained according to point cloud data.
Wherein, V=(v1,v2,…,vn)TIndicate that the matrix being made of the three-dimensional coordinate on vertex each in model, E indicate own in model
Side.
Secondly, calculating the canonical matrix S for relativeness between each cloud in descriptive model, which is becoming
It does not change during shape.Calculation formula is S=CV, wherein C=(I-D-1It B) is transformation matrix, I is unit matrix, and D is pair
Angular moment battle array, the element D on diagonal lineii=di, diFor with point viAdjacent vertex number, B matrix can be expressed from the next:
Again, multiple deformation obligatory points are selected.Here according to priori knowledge, i.e. people's thigh under standing normal posture is transversal
Face is circle, and surface skin tension causes thigh transversal towards elliptical deformation when squatting down.It is available by ellipse by calculating
Circle becomes round rear new obligatory point Euclidean coordinate, and new obligatory point coordinate as restrictive condition and is added to transformation square
Battle array, becomes new transformation matrix C '.Finally, utilizing formula V '=C '-1S solves deformed vertex Euclidean coordinate.
(2) empirical equation
Here main purpose be set up squat down after under the multiple section girths of thigh and normal stand posture perimeter it
Between relationship.Since the leg profile of different crowd is different, deformation also can be different, thus to accurately obtain desired pass
System, it is also necessary to the crowd of different heights, weight, gender be sampled, the situation of change of record deformation front and back perimeter utilizes machine
Regression algorithm such as least square regression algorithm, logistic regression algorithm etc. in device study fits the relationship of deformation front and back, will
This relationship empirically formula.
S23: the characteristic parameter after the actual characteristic parameter of human body after all conversions is summarized as the net model of human body.
S3: the building net model of human body
The parametric modeling of human body can be regarded as by obtaining the net model of human body according to the characteristics of human body of extraction, as shown in figure 3,
The building net model of human body includes following two sub-step: S31 again: establishing standardized human body's model；S32: according to the net mould of the human body
Characteristic parameter modification standardized human body's model of type is to obtain the net model of human body.Under normal circumstances, it is necessary first to a standard people
Body Model, then according to obtained actual human body feature modification standardized human body model to obtain the human body that can reflect actual human body
Net model.
A kind of simple standardized human body's model can have the gridding of a certain proportion of regular shape according to priori knowledge
Manikin, modifying to standardized human body's model, there are mainly three types of situations for transformation: length deformation, width deformation and perimeter
Deformation.When the characteristics of human body of acquisition feature corresponding with master pattern is inconsistent, it is necessary to carry out length, width or perimeter and become
Shape.Specific deformation principle is shaft distortion principle, and so-called shaft distortion principle is by point and length to be deformed on model here
Degree, width or the corresponding axis of perimeter establish mapping relations one by one, and when axis variation, the coordinate of corresponding deformation point also becomes
Change, re-creates model after calculating new coordinate.
There are also manikins, such as SCAPE model etc. that some precision are more increased at present.Human body is obtained by SCAPE model
The process of net model can regard the iteration again and again to SCAPE model as, that is, establish the property for measuring characteristics of human body's gap
Energy function, by the way that SCAPE model is modified again and again iteration until performance function value reaches certain threshold range.
The net model algorithm of former human body is simple, and the real-time of system is higher, the disadvantage is that precision is poor；Latter algorithm compared with
Complexity, precision is high, the disadvantage is that real-time is poor, generally requires the parallel processing of GPU to solve the problems, such as real-time.
It is without restriction to the acquisition methods of the net model of human body in this embodiment.
2, the system of the net model of human body is established
In this embodiment, the system for establishing the net model of human body, including memory, for storing program；Processing
Device runs described program, to execute the above-mentioned side for establishing the net model of human body for controlling the system for establishing the net model of human body
Method.
In other specific embodiments, the system for establishing the net model of human body can also be a kind of comprising computer program
Computer readable storage medium, the computer program are operable to that computer is made to execute the above-mentioned side for establishing the net model of human body
Method.
3, application of the net model of human body in three-dimensional fitting
Three-dimensional fitting method, as shown in Figure 4, comprising the following steps: T1: the creation net model of human body；T2: creation clothing mould
Type；T3: clothes effect is shown after clothing model is synthesized with the net model of human body.
Generally, it fits first to the manikin of standard stance, this step can regard static fitting as；
Secondly the real-time fitting when actual human body postural change can regard dynamic as and fit.Wherein dynamic fitting is actually quiet
The extension of state fitting in time.Thus next mainly illustrate static fitting.
The clothing simulation model of comparative maturity is mass spring model at present, is needed after establishing clothing simulation model by clothing
Object is registered with the net model of human body.It is generally acknowledged that the highest point at the clothing back side and the center of neck after human body, it accordingly can be real
The preliminary registration of existing clothing and manikin；Then the local registration of various pieces is realized according to the current framework information of human body.
After Registration, calculating, laundry hits detection of the power of particle etc. can also be carried out, to simulate more true clothing
Display effect.
During the display of subsequent real-time, as long as the framework information by identifying human body, then according to the skeleton
Real-time 3D fitting can be thus achieved in the registration that information carries out part.
4, three-dimensional fitting system
In this embodiment, three-dimensional fitting system, including memory, for storing program；Processor runs institute
Program is stated, to execute above-mentioned 3D fitting method for controlling the 3D dressing system.
In other specific embodiments, three-dimensional fitting system can also be that a kind of computer comprising computer program can
Storage medium is read, the computer program is operable to that computer is made to execute above-mentioned 3D fitting method.
Present embodiment provides a kind of creation method of net model of the human body based on depth camera, with the prior art
It compares, present embodiment can accurately obtain characteristics of human body's parameter by multiple given poses, then according to human body spy
It levies parameter and obtains the net model of human body, it can be to avoid loose clothing bring manikin error using method of the invention.
In addition, the creation method based on the net model of above-mentioned human body, present embodiment also proposes a kind of three-dimensional fitting side
Method and three-dimensional fitting system, no matter human body wears loose clothing or compact clothing, is tried using the three-dimensional of present embodiment
Clothing method or system, can in real time and 360 ° show the effect worn the clothes.
The above content is specific embodiment is combined, further detailed description of the invention, and it cannot be said that this hair
Bright specific implementation is only limited to these instructions.For those skilled in the art to which the present invention belongs, this is not being departed from
Under the premise of inventive concept, several equivalent substitute or obvious modifications can also be made, and performance or use is identical, should all regarded
To belong to the scope of protection of the present invention.
Claims (9)
1. a kind of creation method of the net model of human body, comprising the following steps:
S1: at least two amplitude deepness images of the human body under at least two given poses are obtained using depth camera；The step S1
In given pose refer to that under the given pose, the clothing of human body at least one portion is in compact condition；
S2: the characteristic parameter of the net model of human body is obtained from least two amplitude deepness images；
S3: the net model of human body is created according to the characteristic parameter；
The step S2 includes:
S21: the characteristic parameter at each compact position of human body in the depth image under each given pose is extracted；
S22: by recoverable force operation or the method for establishing empirical equation, the feature at position compact under each given pose is joined
Number person is converted into the actual characteristic parameter of human body；
S23: the characteristic parameter after the actual characteristic parameter of human body after all conversions is summarized as the net model of human body；
Recoverable force operation described in the step S22 specifically includes:
(1) obtain given pose under compact position point cloud data, and by denoising and cavity filling after obtain having it is high-quality
The point cloud data of amount, and the three-dimensional Euclidean coordinate information of each point is saved, grid model M=(V, E) is obtained according to point cloud data,
In, V=(v1,v2,…,vn)TIndicate that the matrix being made of the three-dimensional coordinate on vertex each in model, E indicate all in model
Side；
(2) canonical matrix S, the calculation formula S=CV for relativeness between each cloud in descriptive model are calculated, wherein
C=(I-D-1It B) is transformation matrix, I is unit matrix, and D is diagonal matrix, the element D on diagonal lineii=di, diFor with point vi
Adjacent vertex number, B matrix can be expressed from the next:
(3) multiple deformation obligatory points are selected, it, will be new by obligatory point Euclidean coordinate new after the available recoverable force of calculating
Obligatory point coordinate is as restrictive condition and is added to transformation matrix, becomes new transformation matrix C ', finally, utilizing formula V '
=C '-1S solves deformed vertex Euclidean coordinate.
2. the creation method of the net model of human body according to claim 1, which is characterized in that the step S3 includes:
S31: standardized human body's model is established；
S32: standardized human body's model is modified to obtain the net model of human body according to the characteristic parameter of the net model of the human body.
3. the creation method of the net model of human body according to claim 1, which is characterized in that the human body in the step S1 is deep
Degree image is obtained in such a way that human body rotates one week around single depth camera, or by using with different angle
It is distributed in the synchronous acquisition of multiple depth cameras of human peripheral.
4. the creation method of the net model of human body according to claim 2, which is characterized in that the modification in the step S32
The method of standardized human body's model is the method based on shaft distortion principle, or by carrying out successive ignition to SCAPE model
Method.
5. a kind of system for establishing the net model of human body, which is characterized in that including memory, for storing program；Processor, operation
Described program, for controlling method of the system execution for establishing the net model of human body as described in claim 1-4 is any.
6. a kind of computer readable storage medium comprising computer program, the computer program are operable to hold computer
Method of the row as described in claim 1-4 is any.
7. a kind of three-dimensional fitting method, comprising the following steps:
T1: method according to claim 1 to 4 creates the net model of human body；
T2: creation clothing model；
T3: clothes effect is shown after clothing model is synthesized with the net model of human body.
8. a kind of three-dimensional fitting system, which is characterized in that including memory, for storing program；Processor runs the journey
Sequence, to execute three-dimensional fitting method as claimed in claim 7 for controlling the three-dimensional fitting system.
9. a kind of computer readable storage medium comprising computer program, the computer program are operable to hold computer
Row three-dimensional fitting method as claimed in claim 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710079459.9A CN106952334B (en) | 2017-02-14 | 2017-02-14 | The creation method and three-dimensional fitting system of the net model of human body |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710079459.9A CN106952334B (en) | 2017-02-14 | 2017-02-14 | The creation method and three-dimensional fitting system of the net model of human body |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106952334A CN106952334A (en) | 2017-07-14 |
| CN106952334B true CN106952334B (en) | 2019-08-30 |
Family
ID=59466418
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710079459.9A Active CN106952334B (en) | 2017-02-14 | 2017-02-14 | The creation method and three-dimensional fitting system of the net model of human body |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106952334B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107507269A (en) * | 2017-07-31 | 2017-12-22 | 广东欧珀移动通信有限公司 | Personalized three-dimensional model generating method, device and terminal device |
| CN107657657A (en) * | 2017-09-08 | 2018-02-02 | 明利 | A kind of three-dimensional human modeling method, device, system and storage medium |
| CN108053437B (en) * | 2017-11-29 | 2021-08-03 | 奥比中光科技集团股份有限公司 | Three-dimensional model obtaining method and device based on posture |
| CN109035393B (en) * | 2018-07-27 | 2023-01-03 | 东方梦幻文化产业投资有限公司 | Method and device for reconstructing three-dimensional animation |
| CN110322329A (en) * | 2019-07-09 | 2019-10-11 | 广州品唯软件有限公司 | A kind of method, apparatus and equipment that manikin is established |
| CN110926371A (en) * | 2019-11-19 | 2020-03-27 | 宁波舜宇仪器有限公司 | Three-dimensional surface detection method and device |
| CN112990997A (en) * | 2019-12-17 | 2021-06-18 | 青岛海尔洗衣机有限公司 | Virtual fitting method |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1266655C (en) * | 2004-12-09 | 2006-07-26 | 上海交通大学 | Human limb three-dimensional model building method based on image cutline |
| CN102855470B (en) * | 2012-07-31 | 2015-04-08 | 中国科学院自动化研究所 | Estimation method of human posture based on depth image |
| GB2515343B (en) * | 2013-06-21 | 2018-02-07 | Toshiba Res Europe Limited | Methods and systems for generating a three dimensional representation of a subject |
| CN104008571B (en) * | 2014-06-12 | 2017-01-18 | 深圳奥比中光科技有限公司 | Human body model obtaining method and network virtual fitting system based on depth camera |
| CN104992441B (en) * | 2015-07-08 | 2017-11-17 | 华中科技大学 | A kind of real human body three-dimensional modeling method towards individualized virtual fitting |
-
2017
- 2017-02-14 CN CN201710079459.9A patent/CN106952334B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN106952334A (en) | 2017-07-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106952334B (en) | The creation method and three-dimensional fitting system of the net model of human body | |
| CN106952335B (en) | Method and system for establishing human body model library | |
| CN104992441B (en) | A kind of real human body three-dimensional modeling method towards individualized virtual fitting | |
| US20210074005A1 (en) | Image processing method and apparatus, image device, and storage medium | |
| Chen et al. | Tensor-based human body modeling | |
| CN106952302B (en) | human body model automatic creation method and three-dimensional fitting system | |
| CN102999942B (en) | Three-dimensional face reconstruction method | |
| CN108416840A (en) | A kind of dense method for reconstructing of three-dimensional scenic based on monocular camera | |
| Zhu et al. | An efficient human model customization method based on orthogonal-view monocular photos | |
| CN102509349B (en) | Fitting method based on mobile terminal, fitting device based on mobile terminal and mobile terminal | |
| CN104346612B (en) | Information processing unit and display methods | |
| CN108875524A (en) | Gaze estimation method, device, system and storage medium | |
| CN108154550A (en) | Face real-time three-dimensional method for reconstructing based on RGBD cameras | |
| CN109086798A (en) | A kind of data mask method and annotation equipment | |
| CN102663820A (en) | Three-dimensional head model reconstruction method | |
| CN103606190A (en) | Method for automatically converting single face front photo into three-dimensional (3D) face model | |
| Cho et al. | An implementation of a garment-fitting simulation system using laser scanned 3D body data | |
| CN108564619B (en) | Realistic three-dimensional face reconstruction method based on two photos | |
| WO2021098143A1 (en) | Image processing method and device, image processing apparatus, and storage medium | |
| CN108664884A (en) | A kind of virtually examination cosmetic method and device | |
| CN109655011A (en) | A kind of method and system of Human Modeling dimension measurement | |
| CN108846892A (en) | The determination method and device of manikin | |
| Xu et al. | Modeling 3D human body with a smart vest | |
| CN107961531A (en) | A kind of virtual social system and its method of work based on motion capture dress ornament | |
| CN111768476A (en) | Expression animation redirection method and system based on grid deformation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CP01 | Change in the name or title of a patent holder | ||
| CP01 | Change in the name or title of a patent holder |
Address after: A808, Zhongdi building, industry university research base, China University of Geosciences, No.8, Yuexing Third Road, Nanshan District, Shenzhen, Guangdong 518000 Patentee after: Obi Zhongguang Technology Group Co., Ltd Address before: A808, Zhongdi building, industry university research base, China University of Geosciences, No.8, Yuexing Third Road, Nanshan District, Shenzhen, Guangdong 518000 Patentee before: SHENZHEN ORBBEC Co.,Ltd. |