CN110335343A - Based on RGBD single-view image human body three-dimensional method for reconstructing and device - Google Patents
Based on RGBD single-view image human body three-dimensional method for reconstructing and device Download PDFInfo
- Publication number
- CN110335343A CN110335343A CN201910512083.5A CN201910512083A CN110335343A CN 110335343 A CN110335343 A CN 110335343A CN 201910512083 A CN201910512083 A CN 201910512083A CN 110335343 A CN110335343 A CN 110335343A
- Authority
- CN
- China
- Prior art keywords
- dimensional
- human body
- manikin
- picture
- rgbd
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 50
- 238000005457 optimization Methods 0.000 claims abstract description 48
- 230000011218 segmentation Effects 0.000 claims abstract description 31
- 230000037237 body shape Effects 0.000 claims abstract description 7
- 238000013135 deep learning Methods 0.000 claims description 27
- 238000013527 convolutional neural network Methods 0.000 claims description 13
- 238000009877 rendering Methods 0.000 claims description 13
- 230000008859 change Effects 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 9
- 235000013399 edible fruits Nutrition 0.000 claims 1
- 230000008901 benefit Effects 0.000 description 5
- 230000006870 function Effects 0.000 description 3
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005286 illumination Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000001537 neural effect Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000000452 restraining effect Effects 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
-
- 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
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10028—Range image; Depth image; 3D point clouds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30196—Human being; Person
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computer Graphics (AREA)
- Geometry (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses one kind to be based on RGBD single-view image human body three-dimensional method for reconstructing and device, wherein method includes: the RGBD picture that human body is acquired by depth camera, and picture includes single-view color image and depth picture；Three-dimensional (3 D) manikin parameter, human body segmentation's information and two-dimentional artis information are respectively obtained according to RGBD picture；Human body three-dimensional artis information is obtained according to human body segmentation's information, two-dimentional artis information and depth picture, to constrain according to artis and body shape of the human body three-dimensional artis information to three-dimensional (3 D) manikin, and optimize three-dimensional (3 D) manikin parameter and three-dimensional (3 D) manikin；Depth picture is rendered according to the three-dimensional (3 D) manikin after optimization, and is advanced optimized using front of the single-view color image to the three-dimensional (3 D) manikin after optimization, to obtain the three-dimensional reconstruction result of human body.This method can use the three-dimensional reconstruction that the collected single frames single-view RGBD pictorial information of depth camera carries out human body.
Description
Technical field
The present invention relates to the three-dimensional reconstruction fields in computer vision, in particular to a kind of to be based on RGBD single-view figure
As human body three-dimensional method for reconstructing and device.
Background technique
With the continuous development of the three-dimensional reconstruction in computer vision field, pass through lightweight using depth camera
Mode, which rebuilds 3 D human body, to seem ever more important.The new architecture iPhoneX of the apple on the 13rd of September in 2017 publication, wherein preposition original
Depth (True Depth) camera causes great public opinion concern.According to the introduction of apple official, iPhoneX is thrown by preposition dot matrix
Shadow device will be more than 30000 sightless spot projections of naked eyes to face, further according to the reflection light point that infrared lens receive, just
Face depth map can be calculated.Predictably, the relevant three-dimensional reconstruction of three-dimensional reconstruction, especially human body, in the case where connecing
Come in the several years that broader development prospect will be had.Various three-dimensional reconstruction applications, for example, virtually change one's clothes, VR communication etc. be likely to by
Step marches toward people's lives.
However, at this stage, it is still the optimization problem for owing fixed that 3 D human body is reconstructed from single frames RGBD picture, it can not
The geological information of the back side and side is directly obtained from picture.And next depth camera will in the continuous popularization of mobile platform
A kind of trend can be become, depth camera is also likely to that very big improvement can be obtained on hardware, but is limited to algorithm, power and sky
Between influence, directly collected depth information precision is poor and institute's Noise is more from commercialization depth camera at this stage,
Detailed information is less.The point cloud being directly translated into using such depth map is often not satisfactory.
Summary of the invention
The application is to be made based on inventor to the understanding of following problems and discovery:
Deep learning is the present invention provides a kind of mode from the study carried out in statistical significance in data, and the present invention is
Using the method for deep learning, information from data focus utilization statistical significance has been carried out invisible part relatively reasonable
Completion, while sufficient optimization has been carried out to the geometric detail of visibility region.
The present invention is directed to solve at least some of the technical problems in related technologies.
For this purpose, an object of the present invention is to provide one kind to be based on RGBD single-view image human body three-dimensional method for reconstructing,
This method rebuilds thinking based on the 3 D human body of RGBD, it is intended that it is excellent to carry out deep layer to reconstructed results using RGBD information and data set
Change.
It is a kind of based on RGBD single-view image human body three-dimensional reconstructing device it is another object of the present invention to propose.
In order to achieve the above objectives, one aspect of the present invention embodiment proposes a kind of based on RGBD single-view image human body three-dimensional
Method for reconstructing, comprising: the RGBD picture of human body is acquired by depth camera, wherein the RGBD picture includes single-view colour
Picture and depth picture；Three-dimensional (3 D) manikin parameter, human body segmentation's information and two dimension is respectively obtained according to the RGBD picture to close
Nodal information；Human body three-dimensional is obtained according to human body segmentation's information, the two-dimentional artis information and the depth picture to close
Nodal information, to be carried out about according to artis and body shape of the human body three-dimensional artis information to three-dimensional (3 D) manikin
Beam, and optimize the three-dimensional (3 D) manikin parameter and the three-dimensional (3 D) manikin；According to the three-dimensional (3 D) manikin rendering after optimization
The depth picture, and it is further excellent using front of the single-view color image to the three-dimensional (3 D) manikin after the optimization
Change, to obtain the three-dimensional reconstruction result of the human body.
The embodiment of the present invention acquires people by depth camera based on RGBD single-view image human body three-dimensional method for reconstructing
The single-view RBGD picture of body, after being rebuild by trained convolutional neural networks model and other algorithm flows
Three-dimensional (3 D) manikin, finally obtained three-dimensional (3 D) manikin can preferably show geometry letter of the human body in camera visibility region
Breath, while passing through the geometry estimation of invisible area relatively reasonable in the method for data-driven acquisition statistical significance.
In addition, according to the above embodiment of the present invention can also be had based on RGBD single-view image human body three-dimensional method for reconstructing
There is following additional technical characteristic:
Further, in one embodiment of the invention, described that 3 D human body is respectively obtained according to the RGBD picture
Model parameter, human body segmentation's information and two-dimentional artis information further comprise: deep using open source from the RGBD picture
Degree study and work HMR estimates to obtain the three-dimensional (3 D) manikin parameter；From the RGBD picture, open source deep learning work is utilized
Make Look into Person and obtains human body segmentation's information；From the RGBD picture, worked using open source deep learning
Open Pose estimates to obtain the two-dimentional artis information.
Further, in one embodiment of the invention, optimize the three-dimensional (3 D) manikin parameter and the three-dimensional people
Body Model further comprises: optimizing the three-dimensional (3 D) manikin parameter using gauss-newton method optimization, to obtain and the RGBD
The three-dimensional (3 D) manikin of picture fitting；Three-dimensional space is initialized using the three-dimensional (3 D) manikin with RGBD picture fitting
Between, it is subject to the human body three-dimensional artis information and the RGBD picture as information gain, and utilize the convolution of U-Net structure
Neural Network Optimization three-dimensional (3 D) manikin.
Further, in one embodiment of the invention, the described and utilization single-view color image is to described excellent
The front of three-dimensional (3 D) manikin after change advanced optimizes, and to obtain the three-dimensional reconstruction result of the human body, further comprises: benefit
With the single-view color image and from the method for 3D shape is restored in rendering to the three-dimensional (3 D) manikin after the optimization
Front advanced optimizes, and carries out trigonometric ratio reconstruction, to obtain the three-dimensional (3 D) manikin using tri patch as basic structure.
Further, in one embodiment of the invention, to the three-dimensional (3 D) manikin after the optimization front into
During one-step optimization, further includes: carry out constraint numerically using the depth picture.
In order to achieve the above objectives, another aspect of the present invention embodiment proposes a kind of based on RGBD single-view image human body three
Tie up reconstructing device, comprising: acquisition module, for acquiring the RGBD picture of human body by depth camera, wherein the RGBD picture
Including single-view color image and depth picture；Processing module, for respectively obtaining 3 D human body mould according to the RGBD picture
Shape parameter, human body segmentation's information and two-dimentional artis information；Optimization module, for according to human body segmentation's information, described two
Dimension artis information and the depth picture obtain human body three-dimensional artis information, according to the human body three-dimensional artis information
The artis and body shape of three-dimensional (3 D) manikin are constrained, and optimize the three-dimensional (3 D) manikin parameter and the three-dimensional
Manikin；Module is rebuild, for rendering the depth picture according to the three-dimensional (3 D) manikin after optimization, and utilizes the haplopia
Angle color image advanced optimizes the front of the three-dimensional (3 D) manikin after the optimization, to obtain the three-dimensional reconstruction of the human body
As a result.
The embodiment of the present invention acquires people by depth camera based on RGBD single-view image human body three-dimensional reconstructing device
The single-view RBGD picture of body, after being rebuild by trained convolutional neural networks model and other algorithm flows
Three-dimensional (3 D) manikin, finally obtained three-dimensional (3 D) manikin can preferably show geometry letter of the human body in camera visibility region
Breath, while passing through the geometry estimation of invisible area relatively reasonable in the method for data-driven acquisition statistical significance.
In addition, according to the above embodiment of the present invention can also be had based on RGBD single-view image human body three-dimensional reconstructing device
There is following additional technical characteristic:
Further, in one embodiment of the invention, the processing module is further used for from the RGBD picture
In, estimate to obtain the three-dimensional (3 D) manikin parameter using open source deep learning work HMR；From the RGBD picture, utilize
Open source deep learning work Look into Person obtains human body segmentation's information；From the RGBD picture, using opening
Depth study and work Open Pose estimates to obtain the two-dimentional artis information.
Further, in one embodiment of the invention, the optimization module is further used for utilizing gauss-newton method
Optimization optimizes the three-dimensional (3 D) manikin parameter, to obtain the three-dimensional (3 D) manikin being fitted with the RGBD picture, and utilizes institute
State with the RGBD picture fitting three-dimensional (3 D) manikin initialize three-dimensional space, be subject to the human body three-dimensional artis information and
The RGBD picture optimizes three-dimensional (3 D) manikin as information gain, and using the convolutional neural networks of U-Net structure.
Further, in one embodiment of the invention, described and rebuild module and be further used for using the haplopia
Angle color image and from rendering restore 3D shape method to the three-dimensional (3 D) manikin after the optimization front further
Optimization, and trigonometric ratio reconstruction is carried out, to obtain the three-dimensional (3 D) manikin using tri patch as basic structure.
Further, in one embodiment of the invention, further includes: constraints module, for after to the optimization
During the front of three-dimensional (3 D) manikin advanced optimizes, constraint numerically is carried out using the depth picture.
The additional aspect of the present invention and advantage will be set forth in part in the description, and will partially become from the following description
Obviously, or practice through the invention is recognized.
Detailed description of the invention
Above-mentioned and/or additional aspect and advantage of the invention will become from the following description of the accompanying drawings of embodiments
Obviously and it is readily appreciated that, in which:
Fig. 1 is the process based on RGBD single-view image human body three-dimensional method for reconstructing according to one embodiment of the invention
Figure；
Fig. 2 is the stream based on RGBD single-view image human body three-dimensional method for reconstructing according to a specific embodiment of the invention
Cheng Tu；
Fig. 3 is to be shown according to the structure based on RGBD single-view image human body three-dimensional reconstructing device of one embodiment of the invention
It is intended to.
Specific embodiment
The embodiment of the present invention is described below in detail, examples of the embodiments are shown in the accompanying drawings, wherein from beginning to end
Same or similar label indicates same or similar element or element with the same or similar functions.Below with reference to attached
The embodiment of figure description is exemplary, it is intended to is used to explain the present invention, and is not considered as limiting the invention.
Describe to propose according to embodiments of the present invention with reference to the accompanying drawings is rebuild based on RGBD single-view image human body three-dimensional
Method and device, describe to propose according to embodiments of the present invention first with reference to the accompanying drawings based on RGBD single-view image human body three-dimensional
Method for reconstructing.
Fig. 1 is the flow chart based on RGBD single-view image human body three-dimensional method for reconstructing of one embodiment of the invention.
As shown in Figure 1, should based on RGBD single-view image human body three-dimensional method for reconstructing the following steps are included:
In step s101, the RGBD picture of human body is acquired by depth camera, wherein RGBD picture includes single-view coloured silk
Chromatic graph piece and depth picture.
It is understood that the embodiment of the present invention can use colour (RGB) picture and depth of depth camera acquisition human body
Spend (Depth) picture (hereinafter referred to as RGBD picture.Wherein, depth camera can be the Microsoft Kinect first generation and second generation phase
Machine, Asus Xtion, light 3D sensing camera etc. in ratio difficult to understand, those skilled in the art can select specifically according to the actual situation
Depth camera is not specifically limited herein.
In step s 102, three-dimensional (3 D) manikin parameter, human body segmentation's information and two dimension are respectively obtained according to RGBD picture
Artis information.
Wherein, three-dimensional (3 D) manikin can be basic manikin (SMPL).It is understood that the embodiment of the present invention can
To estimate three-dimensional (3 D) manikin parameter, human body segmentation information and human body two dimension artis letter from collected RGB picture
Breath.
Further, in one embodiment of the invention, according to RGBD picture respectively obtain three-dimensional (3 D) manikin parameter,
Human body segmentation's information and two-dimentional artis information further comprise: from RGBD picture, utilizing open source deep learning work HMR
Estimation obtains three-dimensional (3 D) manikin parameter；From RGBD picture, obtained using open source deep learning work Look into Person
To human body segmentation's information；From RGBD picture, estimate to obtain two-dimentional artis letter using open source deep learning work Open Pose
Breath.
It is understood that the embodiment of the present invention is firstly the need of described in sharp as above step to collected RGB picture
Using open source deep learning work HMR estimate three-dimensional (3 D) manikin parameter, utilize open source deep learning work Look into
Person carries out human body segmentation information, carries out human body two dimension artis using open source deep learning work Open Pose and believes
The estimation of breath.
Specifically, from collected RGB picture, joined using open source deep learning work HMR estimation three-dimensional (3 D) manikin
Number；From collected RGB picture, human body segmentation letter is carried out using open source deep learning work Look into Person
Breath；From collected RGB picture, human body two dimension artis information is carried out using open source deep learning work Open Pose
Estimation.
In step s 103, human body three-dimensional is obtained according to human body segmentation's information, two-dimentional artis information and depth picture to close
Nodal information, to be constrained according to artis and body shape of the human body three-dimensional artis information to three-dimensional (3 D) manikin, and
Optimize three-dimensional (3 D) manikin parameter and three-dimensional (3 D) manikin.
It is understood that the embodiment of the present invention can use human body two dimension artis information, human body segmentation's information and depth
Picture is spent, estimates human body three-dimensional artis；According to human body three-dimensional artis information, optimize basic human mould using gauss-newton method
Shape parameter simultaneously obtains and the better three-dimensional (3 D) manikin of picture fitting effect；Three-dimensional space is initialized using three-dimensional (3 D) manikin,
It is subject to human body three-dimensional artis and RGBD picture as information gain, using the convolutional neural networks of U-Net structure to three-dimensional mould
Type optimizes.
Specifically, utilizing convolutional Neural net using human body two dimension artis information, human body segmentation's information and depth picture
Network estimates human body three-dimensional artis, further comprises: during being optimized using convolutional Neural net, being closed with two-dimension human body
Node carries out three-dimensional space and projects to the coordinates restriction in two-dimensional image plane, is carried out with depth map in three-dimensional space and as plane is hung down
The upward coordinates restriction of histogram, while constraining projection coordinate cannot be beyond the respective range after human body segmentation.
In addition, the method for being restored 3D shape using RGB picture and from rendering carries out into one the front of threedimensional model
Step optimization further comprises: (1) as similar as possible using the depth map value after the value constraint threedimensional model rendering of depth map；
(2) spheric harmonic function illumination decomposition method is utilized, multiplying for the illumination and intrinsic picture solved using threedimensional model normal vector is constrained
Product is as similar as possible to RGB picture, to enhance threedimensional model normal vector to the descriptive power of object detail.
In step S104, depth picture is rendered according to the three-dimensional (3 D) manikin after optimization, and utilize single-view cromogram
Piece advanced optimizes the front of the three-dimensional (3 D) manikin after optimization, to obtain the three-dimensional reconstruction result of human body.
Wherein, and using front of the single-view color image to the three-dimensional (3 D) manikin after optimization it advanced optimizes, with
To the three-dimensional reconstruction result of human body, further comprise: restoring the side of 3D shape using single-view color image and from rendering
Method advanced optimizes the front of the three-dimensional (3 D) manikin after optimization, and carries out trigonometric ratio reconstruction, to obtain being with tri patch
The three-dimensional (3 D) manikin of basic structure.
Specifically, rendering depth map according to the threedimensional model after optimization, restore three-dimensional using RGB picture and from rendering
The method of shape advanced optimizes the front of threedimensional model, and predominantly it is thin to provide more geometry for three-dimensional (3 D) manikin
Section needs to carry out constraint numerically using depth picture during this；Finally, utilizing the people optimized in resulting three-dimensional space
Body Model carries out trigonometric ratio reconstruction, finally obtains convenient for showing using tri patch as the three-dimensional (3 D) manikin of basic structure.
It will further be explained by specific embodiment based on RGBD single-view image human body three-dimensional method for reconstructing below
It states, as shown in Fig. 2, specific as follows:
Step S1, part of data acquisition.Using depth camera, such as Microsoft's Kinect first generation and second generation camera, Asus
Light 3D sensing camera etc. in Xtion, ratio difficult to understand acquires colored (RGB) picture of single-view and depth (Depth) picture of human body
(hereinafter referred to as RGBD picture).
Wherein, in step sl, the precision of depth map, depth camera pair used by the embodiment of the present invention needs are limited to
Human geometry in photographed scene has certain descriptive power.
Step S2, data processing section.To collected RGB picture, the embodiment of the present invention is firstly the need of benefit step as above
Described in estimate basic manikin (SMPL) parameter using open source deep learning work HMR, utilize open source deep learning
The Look into Person that works carries out human body segmentation information, carries out human body using open source deep learning work OpenPose
The estimation of two-dimentional artis information.
Wherein, in step s 2, the embodiment of the present invention needs three open source work that can obtain just picture collected
Normal processing result for extreme posture, such as is stood upside down, rolling, and treatment effect may be to be improved.
Step S3, optimization and reconstruction part.After obtaining information as above, preparatory trained convolutional neural networks and depth are utilized
Degree figure, carries out the estimation of 3 D human body artis.Trained solver is recycled to optimize basic three-dimensional (3 D) manikin, and will be as
Upper model and input of remaining prior information as three-dimensional optimized convolutional neural networks, most by trained parameter output in advance
The three-dimensional (3 D) manikin rebuild eventually.
Wherein, in step s3, the embodiment of the present invention is provided with reasonable damage to two convolutional neural networks and solver
Function, parameter and initialization value are lost, has trained the network weight parameter after restraining in the collected data set of institute in advance, it can
The model after this pre-training is directly applied to actual optimization and reconstruction process.
To sum up, the embodiment of the present invention is intended to carry out people using the collected single frames single-view RGBD pictorial information of depth camera
The three-dimensional reconstruction of body is utilized the RGBD human body picture for the single frames single-view that depth camera is shot as input information, adopts
Basic configuration, two-dimensional framework and the human body segmentation information of human body are extracted from RGB picture with the method based on deep learning,
After the three-dimensional framework for estimating human body in conjunction with depth map, in summary use of information convolutional neural networks encode-solve to it
Code, so that the three-dimensional (3 D) manikin after being optimized, it is being carried out further using the constraint of the numerical value of RGB picture and depth map
The optimization of geometric detail, the three-dimensional (3 D) manikin after finally obtaining optimized reconstruction.
It is proposed according to embodiments of the present invention based on RGBD single-view image human body three-dimensional method for reconstructing, pass through depth phase
Machine acquires the single-view RBGD picture of human body, can be obtained by trained convolutional neural networks model and other algorithm flows
Three-dimensional (3 D) manikin after to reconstruction, finally obtained three-dimensional (3 D) manikin can preferably show human body in camera visibility region
In geological information, while being estimated by the geometry that the method for data-driven obtains relatively reasonable invisible area in statistical significance
Meter.
It is rebuild referring next to what attached drawing description proposed according to embodiments of the present invention based on RGBD single-view image human body three-dimensional
Device.
Fig. 3 is the structural representation based on RGBD single-view image human body three-dimensional reconstructing device of one embodiment of the invention
Figure.
As shown in figure 3, should include: acquisition module 100, processing based on RGBD single-view image human body three-dimensional reconstructing device 10
Module 200, optimization module 300 and reconstruction module 400.
Wherein, acquisition module 100 is used for the RGBD picture by depth camera acquisition human body, wherein RGBD picture includes
Single-view color image and depth picture.Processing module 200 be used for according to RGBD picture respectively obtain three-dimensional (3 D) manikin parameter,
Human body segmentation's information and two-dimentional artis information.Optimization module 300 be used for according to human body segmentation's information, two-dimentional artis information and
Depth picture obtains human body three-dimensional artis information, with the artis according to human body three-dimensional artis information to three-dimensional (3 D) manikin
It is constrained with body shape, and optimizes three-dimensional (3 D) manikin parameter and three-dimensional (3 D) manikin.Module 400 is rebuild to be used for according to excellent
Three-dimensional (3 D) manikin after change renders depth picture, and using single-view color image to the three-dimensional (3 D) manikin after optimization just
Face advanced optimizes, to obtain the three-dimensional reconstruction result of human body.The device 10 of the embodiment of the present invention can use depth camera and adopt
The single frames single-view RGBD pictorial information collected carries out the three-dimensional reconstruction of human body, and obtained three-dimensional (3 D) manikin being capable of preferable table
Existing geological information of the human body in camera visibility region, while by relatively reasonable in the method for data-driven acquisition statistical significance
Invisible area geometry estimation.
Further, in one embodiment of the invention, processing module 200 is further used for from RGBD picture, benefit
Estimate to obtain three-dimensional (3 D) manikin parameter with open source deep learning work HMR；From RGBD picture, open source deep learning work is utilized
Make Look into Person and obtains human body segmentation's information；From RGBD picture, open source deep learning work Open Pose is utilized
Estimation obtains two-dimentional artis information.
Further, in one embodiment of the invention, optimization module 300 is further used for excellent using gauss-newton method
Change optimization three-dimensional (3 D) manikin parameter, to obtain the three-dimensional (3 D) manikin being fitted with RGBD picture, and intends using with RGBD picture
The three-dimensional (3 D) manikin of conjunction initializes three-dimensional space, is subject to human body three-dimensional artis information and RGBD picture as information gain,
And optimize three-dimensional (3 D) manikin using the convolutional neural networks of U-Net structure.
Further, in one embodiment of the invention, and rebuild module 400 be further used for it is colored using single-view
Picture and the front of the three-dimensional (3 D) manikin after optimization is advanced optimized from the method for restoring 3D shape in rendering, and is carried out
Trigonometric ratio is rebuild, to obtain the three-dimensional (3 D) manikin using tri patch as basic structure.
Further, in one embodiment of the invention, the device 10 of the embodiment of the present invention further include: constraints module.
Wherein, constraints module is used for during the front to the three-dimensional (3 D) manikin after optimization advanced optimizes, and utilizes depth map
Piece carries out constraint numerically.
It should be noted that aforementioned to the explanation based on RGBD single-view image human body three-dimensional method for reconstructing embodiment
Be also applied for the embodiment based on RGBD single-view image human body three-dimensional reconstructing device, details are not described herein again.
It is proposed according to embodiments of the present invention based on RGBD single-view image human body three-dimensional reconstructing device, pass through depth phase
Machine acquires the single-view RBGD picture of human body, can be obtained by trained convolutional neural networks model and other algorithm flows
Three-dimensional (3 D) manikin after to reconstruction, finally obtained three-dimensional (3 D) manikin can preferably show human body in camera visibility region
In geological information, while being estimated by the geometry that the method for data-driven obtains relatively reasonable invisible area in statistical significance
Meter.
In addition, term " first ", " second " are used for descriptive purposes only and cannot be understood as indicating or suggesting relative importance
Or implicitly indicate the quantity of indicated technical characteristic.Define " first " as a result, the feature of " second " can be expressed or
Implicitly include at least one this feature.In the description of the present invention, the meaning of " plurality " is at least two, such as two, three
It is a etc., unless otherwise specifically defined.
In the present invention unless specifically defined or limited otherwise, fisrt feature in the second feature " on " or " down " can be with
It is that the first and second features directly contact or the first and second features pass through intermediary mediate contact.Moreover, fisrt feature exists
Second feature " on ", " top " and " above " but fisrt feature be directly above or diagonally above the second feature, or be merely representative of
First feature horizontal height is higher than second feature.Fisrt feature can be under the second feature " below ", " below " and " below "
One feature is directly under or diagonally below the second feature, or is merely representative of first feature horizontal height less than second feature.
In the description of this specification, reference term " one embodiment ", " some embodiments ", " example ", " specifically show
The description of example " or " some examples " etc. means specific features, structure, material or spy described in conjunction with this embodiment or example
Point is included at least one embodiment or example of the invention.In the present specification, schematic expression of the above terms are not
It must be directed to identical embodiment or example.Moreover, particular features, structures, materials, or characteristics described can be in office
It can be combined in any suitable manner in one or more embodiment or examples.In addition, without conflicting with each other, the skill of this field
Art personnel can tie the feature of different embodiments or examples described in this specification and different embodiments or examples
It closes and combines.
Although the embodiments of the present invention has been shown and described above, it is to be understood that above-described embodiment is example
Property, it is not considered as limiting the invention, those skilled in the art within the scope of the invention can be to above-mentioned
Embodiment is changed, modifies, replacement and variant.
Claims (10)
1. one kind is based on RGBD single-view image human body three-dimensional method for reconstructing characterized by comprising
The RGBD picture of human body is acquired by depth camera, wherein the RGBD picture includes single-view color image and depth
Picture；
Three-dimensional (3 D) manikin parameter, human body segmentation's information and two-dimentional artis information are respectively obtained according to the RGBD picture；
Human body three-dimensional artis letter is obtained according to human body segmentation's information, the two-dimentional artis information and the depth picture
Breath, to be constrained according to artis and body shape of the human body three-dimensional artis information to three-dimensional (3 D) manikin, and it is excellent
Change the three-dimensional (3 D) manikin parameter and the three-dimensional (3 D) manikin；And
The depth picture is rendered according to the three-dimensional (3 D) manikin after optimization, and using the single-view color image to described excellent
The front of three-dimensional (3 D) manikin after change advanced optimizes, to obtain the three-dimensional reconstruction result of the human body.
2. the method according to claim 1, wherein described respectively obtain 3 D human body according to the RGBD picture
Model parameter, human body segmentation's information and two-dimentional artis information further comprise:
From the RGBD picture, estimate to obtain the three-dimensional (3 D) manikin parameter using open source deep learning work HMR；
From the RGBD picture, the human body segmentation is obtained using open source deep learning work Look into Person and is believed
Breath；
From the RGBD picture, estimate to obtain the two-dimentional artis information using open source deep learning work Open Pose.
3. the method according to claim 1, wherein optimizing the three-dimensional (3 D) manikin parameter and the three-dimensional people
Body Model further comprises:
Optimize the three-dimensional (3 D) manikin parameter using gauss-newton method optimization, to obtain the three-dimensional being fitted with the RGBD picture
Manikin；
Three-dimensional space is initialized using the three-dimensional (3 D) manikin with RGBD picture fitting, is subject to the human body three-dimensional and closes
Nodal information and the RGBD picture optimize 3 D human body as information gain, and using the convolutional neural networks of U-Net structure
Model.
4. the method according to claim 1, wherein the described and utilization single-view color image is to described excellent
The front of three-dimensional (3 D) manikin after change advanced optimizes, and to obtain the three-dimensional reconstruction result of the human body, further comprises:
Using the single-view color image and from the method for 3D shape is restored in rendering to the 3 D human body after the optimization
The front of model advanced optimizes, and carries out trigonometric ratio reconstruction, to obtain the 3 D human body mould using tri patch as basic structure
Type.
5. according to the method described in claim 4, it is characterized in that, to the three-dimensional (3 D) manikin after the optimization front into
During one-step optimization, further includes:
Constraint numerically is carried out using the depth picture.
6. one kind is based on RGBD single-view image human body three-dimensional reconstructing device characterized by comprising
Acquisition module, for acquiring the RGBD picture of human body by depth camera, wherein the RGBD picture includes single-view coloured silk
Chromatic graph piece and depth picture；
Processing module, for respectively obtaining three-dimensional (3 D) manikin parameter, human body segmentation's information and two dimension according to the RGBD picture
Artis information；
Optimization module, for obtaining people according to human body segmentation's information, the two-dimentional artis information and the depth picture
Body three-dimensional artis information, with the artis and body shape according to the human body three-dimensional artis information to three-dimensional (3 D) manikin
It is constrained, and optimizes the three-dimensional (3 D) manikin parameter and the three-dimensional (3 D) manikin；And
Module is rebuild, for rendering the depth picture according to the three-dimensional (3 D) manikin after optimization, and it is color using the single-view
Chromatic graph piece advanced optimizes the front of the three-dimensional (3 D) manikin after the optimization, to obtain the three-dimensional reconstruction knot of the human body
Fruit.
7. device according to claim 6, which is characterized in that the processing module is further used for from the RGBD picture
In, estimate to obtain the three-dimensional (3 D) manikin parameter using open source deep learning work HMR；From the RGBD picture, utilize
Open source deep learning work Look into Person obtains human body segmentation's information；From the RGBD picture, using opening
Depth study and work Open Pose estimates to obtain the two-dimentional artis information.
8. device according to claim 6, which is characterized in that the optimization module is further used for utilizing gauss-newton method
Optimization optimizes the three-dimensional (3 D) manikin parameter, to obtain the three-dimensional (3 D) manikin being fitted with the RGBD picture, and utilizes institute
State with the RGBD picture fitting three-dimensional (3 D) manikin initialize three-dimensional space, be subject to the human body three-dimensional artis information and
The RGBD picture optimizes three-dimensional (3 D) manikin as information gain, and using the convolutional neural networks of U-Net structure.
9. device according to claim 6, which is characterized in that described and rebuild module and be further used for using the haplopia
Angle color image and from rendering restore 3D shape method to the three-dimensional (3 D) manikin after the optimization front further
Optimization, and trigonometric ratio reconstruction is carried out, to obtain the three-dimensional (3 D) manikin using tri patch as basic structure.
10. device according to claim 9, which is characterized in that further include:
Constraints module, for utilizing institute during the front to the three-dimensional (3 D) manikin after the optimization advanced optimizes
State the constraint of depth picture progress numerically.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910512083.5A CN110335343B (en) | 2019-06-13 | 2019-06-13 | Human body three-dimensional reconstruction method and device based on RGBD single-view-angle image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910512083.5A CN110335343B (en) | 2019-06-13 | 2019-06-13 | Human body three-dimensional reconstruction method and device based on RGBD single-view-angle image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110335343A true CN110335343A (en) | 2019-10-15 |
| CN110335343B CN110335343B (en) | 2021-04-06 |
Family
ID=68142018
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910512083.5A Expired - Fee Related CN110335343B (en) | 2019-06-13 | 2019-06-13 | Human body three-dimensional reconstruction method and device based on RGBD single-view-angle image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110335343B (en) |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111008966A (en) * | 2019-12-02 | 2020-04-14 | 深圳市繁维医疗科技有限公司 | RGBD-based single-view-angle human body measurement method and device and computer-readable storage medium |
| CN111031305A (en) * | 2019-11-21 | 2020-04-17 | 北京市商汤科技开发有限公司 | Image processing method and apparatus, image device, and storage medium |
| CN111127632A (en) * | 2019-12-20 | 2020-05-08 | 北京奇艺世纪科技有限公司 | Human body modeling model obtaining method and device, electronic equipment and storage medium |
| CN111476884A (en) * | 2020-03-30 | 2020-07-31 | 清华大学 | Real-time three-dimensional human body reconstruction method and system based on single-frame RGBD image |
| CN111739161A (en) * | 2020-07-23 | 2020-10-02 | 之江实验室 | Human body three-dimensional reconstruction method and device under shielding condition and electronic equipment |
| CN111862299A (en) * | 2020-06-15 | 2020-10-30 | 上海非夕机器人科技有限公司 | Human body three-dimensional model construction method and device, robot and storage medium |
| CN111968165A (en) * | 2020-08-19 | 2020-11-20 | 北京拙河科技有限公司 | Dynamic human body three-dimensional model completion method, device, equipment and medium |
| CN112330795A (en) * | 2020-10-10 | 2021-02-05 | 清华大学 | Human body three-dimensional reconstruction method and system based on single RGBD image |
| CN112819944A (en) * | 2021-01-21 | 2021-05-18 | 魔珐(上海)信息科技有限公司 | Three-dimensional human body model reconstruction method and device, electronic equipment and storage medium |
| CN112884638A (en) * | 2021-02-02 | 2021-06-01 | 北京东方国信科技股份有限公司 | Virtual fitting method and device |
| CN113313828A (en) * | 2021-05-19 | 2021-08-27 | 华南理工大学 | Three-dimensional reconstruction method and system based on single-picture intrinsic image decomposition |
| CN113313818A (en) * | 2021-06-07 | 2021-08-27 | 聚好看科技股份有限公司 | Three-dimensional reconstruction method, device and system |
| CN113379904A (en) * | 2021-07-05 | 2021-09-10 | 东南大学 | Hidden space motion coding-based multi-person human body model reconstruction method |
| CN113468923A (en) * | 2020-03-31 | 2021-10-01 | 上海交通大学 | Human-object interaction behavior detection method based on fine-grained multi-modal common representation |
| CN113610889A (en) * | 2021-06-30 | 2021-11-05 | 奥比中光科技集团股份有限公司 | Human body three-dimensional model obtaining method and device, intelligent terminal and storage medium |
| CN113808256A (en) * | 2021-09-15 | 2021-12-17 | 天津大学 | High-precision holographic human body reconstruction method combined with identity recognition |
| CN114241160A (en) * | 2021-12-22 | 2022-03-25 | 重庆师范大学 | Single-view-angle blade three-dimensional reconstruction method based on deep learning |
| US11450068B2 (en) | 2019-11-21 | 2022-09-20 | Beijing Sensetime Technology Development Co., Ltd. | Method and device for processing image, and storage medium using 3D model, 2D coordinates, and morphing parameter |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101833788A (en) * | 2010-05-18 | 2010-09-15 | 南京大学 | Three-dimensional human modeling method by using cartographical sketching |
| CN105809681A (en) * | 2016-03-04 | 2016-07-27 | 清华大学 | Single camera based human body RGB-D data restoration and 3D reconstruction method |
| CN107507267A (en) * | 2017-07-28 | 2017-12-22 | 电子科技大学 | Human body back three-dimensional reconstruction method |
| CN108053469A (en) * | 2017-12-26 | 2018-05-18 | 清华大学 | Complicated dynamic scene human body three-dimensional method for reconstructing and device under various visual angles camera |
| CN108154551A (en) * | 2017-11-29 | 2018-06-12 | 深圳奥比中光科技有限公司 | The method and system of real-time dynamic reconstruction three-dimensional (3 D) manikin |
| CN108154550A (en) * | 2017-11-29 | 2018-06-12 | 深圳奥比中光科技有限公司 | Face real-time three-dimensional method for reconstructing based on RGBD cameras |
| EP3381017A1 (en) * | 2016-10-31 | 2018-10-03 | Google LLC | Face reconstruction from a learned embedding |
| CN109341707A (en) * | 2018-12-03 | 2019-02-15 | 南开大学 | Mobile robot three-dimensional map construction method under circumstances not known |
-
2019
- 2019-06-13 CN CN201910512083.5A patent/CN110335343B/en not_active Expired - Fee Related
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101833788A (en) * | 2010-05-18 | 2010-09-15 | 南京大学 | Three-dimensional human modeling method by using cartographical sketching |
| CN105809681A (en) * | 2016-03-04 | 2016-07-27 | 清华大学 | Single camera based human body RGB-D data restoration and 3D reconstruction method |
| EP3381017A1 (en) * | 2016-10-31 | 2018-10-03 | Google LLC | Face reconstruction from a learned embedding |
| CN107507267A (en) * | 2017-07-28 | 2017-12-22 | 电子科技大学 | Human body back three-dimensional reconstruction method |
| CN108154551A (en) * | 2017-11-29 | 2018-06-12 | 深圳奥比中光科技有限公司 | The method and system of real-time dynamic reconstruction three-dimensional (3 D) manikin |
| CN108154550A (en) * | 2017-11-29 | 2018-06-12 | 深圳奥比中光科技有限公司 | Face real-time three-dimensional method for reconstructing based on RGBD cameras |
| CN108053469A (en) * | 2017-12-26 | 2018-05-18 | 清华大学 | Complicated dynamic scene human body three-dimensional method for reconstructing and device under various visual angles camera |
| CN109341707A (en) * | 2018-12-03 | 2019-02-15 | 南开大学 | Mobile robot three-dimensional map construction method under circumstances not known |
Non-Patent Citations (2)
| Title |
|---|
| RAFAEL DINIZ.ETC: ""Real-Time 3D volumetric human body reconstruction from a single view RGB-D capture device"", 《ELECTRONIC IMAGING》 * |
| 赵艳: ""基于RGB彩色和深度信息的三维运动重建研究"", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (32)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11450068B2 (en) | 2019-11-21 | 2022-09-20 | Beijing Sensetime Technology Development Co., Ltd. | Method and device for processing image, and storage medium using 3D model, 2D coordinates, and morphing parameter |
| CN111031305A (en) * | 2019-11-21 | 2020-04-17 | 北京市商汤科技开发有限公司 | Image processing method and apparatus, image device, and storage medium |
| KR102406438B1 (en) * | 2019-11-21 | 2022-06-08 | 베이징 센스타임 테크놀로지 디벨롭먼트 컴퍼니 리미티드 | Image processing method and apparatus, image processing apparatus and storage medium |
| TWI750710B (en) * | 2019-11-21 | 2021-12-21 | 中國商北京市商湯科技開發有限公司 | Image processing method and apparatus, image processing device and storage medium |
| KR20210064113A (en) * | 2019-11-21 | 2021-06-02 | 베이징 센스타임 테크놀로지 디벨롭먼트 컴퍼니 리미티드 | Image processing method and apparatus, image processing apparatus and storage medium |
| CN111008966A (en) * | 2019-12-02 | 2020-04-14 | 深圳市繁维医疗科技有限公司 | RGBD-based single-view-angle human body measurement method and device and computer-readable storage medium |
| CN111127632A (en) * | 2019-12-20 | 2020-05-08 | 北京奇艺世纪科技有限公司 | Human body modeling model obtaining method and device, electronic equipment and storage medium |
| CN111127632B (en) * | 2019-12-20 | 2023-06-02 | 北京奇艺世纪科技有限公司 | Human modeling model acquisition method and device, electronic equipment and storage medium |
| CN111476884A (en) * | 2020-03-30 | 2020-07-31 | 清华大学 | Real-time three-dimensional human body reconstruction method and system based on single-frame RGBD image |
| CN111476884B (en) * | 2020-03-30 | 2022-10-25 | 清华大学 | Real-time three-dimensional human body reconstruction method and system based on single-frame RGBD image |
| CN113468923B (en) * | 2020-03-31 | 2022-09-06 | 上海交通大学 | Human-object interaction behavior detection method based on fine-grained multi-modal common representation |
| CN113468923A (en) * | 2020-03-31 | 2021-10-01 | 上海交通大学 | Human-object interaction behavior detection method based on fine-grained multi-modal common representation |
| CN111862299A (en) * | 2020-06-15 | 2020-10-30 | 上海非夕机器人科技有限公司 | Human body three-dimensional model construction method and device, robot and storage medium |
| CN111739161A (en) * | 2020-07-23 | 2020-10-02 | 之江实验室 | Human body three-dimensional reconstruction method and device under shielding condition and electronic equipment |
| CN111968165B (en) * | 2020-08-19 | 2024-01-23 | 北京拙河科技有限公司 | Dynamic human body three-dimensional model complement method, device, equipment and medium |
| CN111968165A (en) * | 2020-08-19 | 2020-11-20 | 北京拙河科技有限公司 | Dynamic human body three-dimensional model completion method, device, equipment and medium |
| CN112330795B (en) * | 2020-10-10 | 2022-10-28 | 清华大学 | Human body three-dimensional reconstruction method and system based on single RGBD image |
| CN112330795A (en) * | 2020-10-10 | 2021-02-05 | 清华大学 | Human body three-dimensional reconstruction method and system based on single RGBD image |
| CN112819944A (en) * | 2021-01-21 | 2021-05-18 | 魔珐(上海)信息科技有限公司 | Three-dimensional human body model reconstruction method and device, electronic equipment and storage medium |
| CN112884638B (en) * | 2021-02-02 | 2024-08-20 | 北京东方国信科技股份有限公司 | Virtual fitting method and device |
| CN112884638A (en) * | 2021-02-02 | 2021-06-01 | 北京东方国信科技股份有限公司 | Virtual fitting method and device |
| CN113313828B (en) * | 2021-05-19 | 2022-06-14 | 华南理工大学 | Three-dimensional reconstruction method and system based on single-picture intrinsic image decomposition |
| CN113313828A (en) * | 2021-05-19 | 2021-08-27 | 华南理工大学 | Three-dimensional reconstruction method and system based on single-picture intrinsic image decomposition |
| CN113313818A (en) * | 2021-06-07 | 2021-08-27 | 聚好看科技股份有限公司 | Three-dimensional reconstruction method, device and system |
| CN113610889B (en) * | 2021-06-30 | 2024-01-16 | 奥比中光科技集团股份有限公司 | Human body three-dimensional model acquisition method and device, intelligent terminal and storage medium |
| CN113610889A (en) * | 2021-06-30 | 2021-11-05 | 奥比中光科技集团股份有限公司 | Human body three-dimensional model obtaining method and device, intelligent terminal and storage medium |
| WO2023273093A1 (en) * | 2021-06-30 | 2023-01-05 | 奥比中光科技集团股份有限公司 | Human body three-dimensional model acquisition method and apparatus, intelligent terminal, and storage medium |
| CN113379904B (en) * | 2021-07-05 | 2022-02-15 | 东南大学 | Hidden space motion coding-based multi-person human body model reconstruction method |
| CN113379904A (en) * | 2021-07-05 | 2021-09-10 | 东南大学 | Hidden space motion coding-based multi-person human body model reconstruction method |
| CN113808256A (en) * | 2021-09-15 | 2021-12-17 | 天津大学 | High-precision holographic human body reconstruction method combined with identity recognition |
| CN113808256B (en) * | 2021-09-15 | 2023-06-09 | 天津大学 | High-precision holographic human body reconstruction method combined with identity recognition |
| CN114241160A (en) * | 2021-12-22 | 2022-03-25 | 重庆师范大学 | Single-view-angle blade three-dimensional reconstruction method based on deep learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110335343B (en) | 2021-04-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110335343A (en) | Based on RGBD single-view image human body three-dimensional method for reconstructing and device | |
| US11443480B2 (en) | Method and system for remote clothing selection | |
| CN109584353B (en) | Method for reconstructing three-dimensional facial expression model based on monocular video | |
| CN109377557B (en) | Real-time three-dimensional face reconstruction method based on single-frame face image | |
| CN107274493B (en) | Three-dimensional virtual trial type face reconstruction method based on mobile platform | |
| CN105354876B (en) | A kind of real-time volume fitting method based on mobile terminal | |
| EP3971841A1 (en) | Three-dimensional model generation method and apparatus, and computer device and storage medium | |
| US20230290040A1 (en) | Systems and methods for end to end scene reconstruction from multiview images | |
| CN107506714A (en) | A kind of method of face image relighting | |
| CN112784621B (en) | Image display method and device | |
| JP2019510297A (en) | Virtual try-on to the user's true human body model | |
| CN110223370A (en) | A method of complete human body's texture mapping is generated from single view picture | |
| Li et al. | In-home application (App) for 3D virtual garment fitting dressing room | |
| TWI750710B (en) | Image processing method and apparatus, image processing device and storage medium | |
| US11450068B2 (en) | Method and device for processing image, and storage medium using 3D model, 2D coordinates, and morphing parameter | |
| CN113628327A (en) | Head three-dimensional reconstruction method and equipment | |
| CN106127818A (en) | A kind of material appearance based on single image obtains system and method | |
| CN109871589A (en) | Intelligent clothing system and method based on Stereo face recognition | |
| Thalmann et al. | Modeling of populations | |
| WO2020104990A1 (en) | Virtually trying cloths & accessories on body model | |
| CN105913496A (en) | Method and system for fast conversion of real clothes to three-dimensional virtual clothes | |
| Wu et al. | [Retracted] 3D Film Animation Image Acquisition and Feature Processing Based on the Latest Virtual Reconstruction Technology | |
| CN108769644B (en) | Binocular animation stylized rendering method based on deep learning | |
| Gong | Application and Practice of Artificial Intelligence Technology in Interior Design | |
| CN115272628A (en) | Rendering method and device of three-dimensional model, computer equipment and storage medium |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20210406 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |