CN109242951A - A kind of face's real-time three-dimensional method for reconstructing - Google Patents
A kind of face's real-time three-dimensional method for reconstructing Download PDFInfo
- Publication number
- CN109242951A CN109242951A CN201810888258.8A CN201810888258A CN109242951A CN 109242951 A CN109242951 A CN 109242951A CN 201810888258 A CN201810888258 A CN 201810888258A CN 109242951 A CN109242951 A CN 109242951A
- Authority
- CN
- China
- Prior art keywords
- point cloud
- face
- point
- cloud
- frame
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 51
- 230000008447 perception Effects 0.000 claims abstract description 6
- 239000011159 matrix material Substances 0.000 claims description 24
- 230000009466 transformation Effects 0.000 claims description 12
- 238000013519 translation Methods 0.000 claims description 6
- 238000005070 sampling Methods 0.000 claims description 3
- 230000004927 fusion Effects 0.000 abstract 1
- 238000007781 pre-processing Methods 0.000 abstract 1
- 230000001629 suppression Effects 0.000 abstract 1
- 238000004364 calculation method Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 239000013598 vector Substances 0.000 description 2
- 230000008901 benefit Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000005056 compaction Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/33—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30196—Human being; Person
- G06T2207/30201—Face
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 Processing (AREA)
- Image Analysis (AREA)
Abstract
The present invention relates to a kind of face's real-time three-dimensional method for reconstructing, comprising: obtains point cloud data frame by depth perception equipment；To point cloud data carry out background filter out, simplify and noise suppression preprocessing formed foreground point cloud；Face's point cloud is extracted from the cloud of foreground point, and face's point cloud and trunk point cloud are split；It is registrated using the ICP algorithm face point cloud pretreated with previous frame to this frame face point cloud, face's threedimensional model is constructed by quick gridding, and smooth to face's threedimensional model progress Laplce, generate the threedimensional model of present frame point cloud；Above-mentioned steps are repeated, generate face's threedimensional model in real time.The present invention can not quickly identify face location by RGB information, rotate face's point cloud under two width different coordinates by registration Algorithm, move to the same coordinate system, then by obtaining threedimensional model after the fusion of two amplitude point clouds, gridding and exporting in real time.
Description
Technical field
The invention belongs to image procossings and three-dimensional reconstruction field, and in particular to a kind of face's real-time three-dimensional reconstruction side
Method.
Background technique
Real-time three-dimensional reconstruction technique is an important research side of computer vision field and Computer Image Processing field
To human face three-dimensional model is even more a research hotspot.With the development of 3-D technology, threedimensional model is widely used in industrial automatic
In the scenes such as change, digital medical, automatic Pilot auxiliary system and robot application.Acquisition three-dimensional information relies primarily on structure light and dissipates
Spot coding techniques and technique of binocular stereoscopic vision, wherein structure light speckle encoding technology is that infrared projector is thrown to body surface
Specific coding pattern is penetrated, forms speckle point via diffusing reflection, encoded information is acquired by imaging sensor and is decoded, to restore object
The depth information of body；Technique of binocular stereoscopic vision is then calculated by characteristic matching by shooting multiple image from different perspectives
Method finds characteristic point, in conjunction with camera interior and exterior parameter and is subject to epipolar-line constraint, the 3 d space coordinate of object is obtained with this.Knot
Structure light speckle encoding technical speed is fast, precision is high, robustness is high, is not influenced by object surface shape and natural lighting condition,
Opposite technique of binocular stereoscopic vision has wider array of application range.
Relative to the three-dimensional rebuilding method of general object, face's three-dimensional reconstruction requires point cloud precision height, noise small, defeated in real time
Threedimensional model out.Existing three-dimensional rebuilding method is computationally intensive, operation time is long, precision is insufficient.The calculation of point cloud triangle gridding
Method has Delaunay triangulation network lattice Reconstruction Method, greedy triangle projection algorithm, Poisson to rebuild, MC the methods of is rebuild, but these sides
Method operand is big, can not generate threedimensional model in real time when cloud is larger.
Summary of the invention
The problem big for operand existing in the prior art, precision is insufficient, the present invention provide a kind of face real-time three
The essence of three-dimensional reconstruction can be effectively improved by constructing good cloud topological structure in point cloud pretreatment by tieing up method for reconstructing
Degree and real-time, obtain accurate face's threedimensional model.
A kind of face's real-time three-dimensional method for reconstructing, includes the following steps:
S100: point cloud data is obtained by depth perception equipment；
S200: pretreatment is carried out to point cloud data and forms foreground point cloud, the pretreatment includes that point cloud background is filtered out, simplified
And denoising；
S300: face's point cloud is extracted from the cloud of foreground point, and face's point cloud and trunk point cloud are split；
S400: being registrated using ICP algorithm face's point cloud pretreated with previous frame to this frame face point cloud, will not
It rotated with face's point cloud under coordinate system, move to the same coordinate system, smooth by quick gridding and Laplce, reconstruction is worked as
The threedimensional model of previous frame point cloud；
S500: repeating step S100- step S400, generates face's threedimensional model in real time.
Optionally, in step s 200, described cloud background, which filters out, refers to by choosing the depth value i.e. range of point cloud z coordinate
To remove the redundancy background information in point cloud.
Optionally, in step s 200, the point cloud compressing using bounding box method, uniform grid method, Triangular meshes method and
Any one in curvature sampling method.
Optionally, the bounding box method includes:
1) cloud region is divided into several cubes according to actual physics section is equidistant；
2) retain a data point in each cube, using the data point as in all the points cloud in the cube
The heart.
Optionally, in step s 200, described cloud denoising, which refers to, is grouped into the click and sweep in cube with similar quality together
In one set.
Optionally, the range of described cloud z coordinate is 200mm-600mm.
Optionally, the cubic volume is set as 1mm × 1mm × 1mm or 2mm × 2mm × 2mm.
Optionally, the face point cloud pretreated with previous frame to this frame face point cloud using ICP algorithm is matched
Standard specifically includes::
1) internal reference of depth camera is demarcated by Zhang Zhengyou calibration method；
2) shooting previous frame depth map and this frame depth map extract feature point set p={ p1, p2... pnAnd q={ q1,
q2... qn, piAnd qiFor the corresponding characteristic point of two frame depth maps；
3) eigenmatrix E is solved according to matched characteristic point, obtains spin matrix R and translation matrix T by decomposing；
4) using spin matrix R and translation matrix T as the initial transformation matrix of ICP algorithm.
Optionally, the threedimensional model for rebuilding present frame point cloud refers to after obtaining the initial transformation matrix, uses
Orderly the mode of point cloud line is rebuild, and is specifically included:
1) initial point P (i, j)=P (1,1) is set, i=j=1；
2) point centered on initial point P (i, j), sequentially generates tri patch；
3) 2) i=i+2, j=j+2 are returned, until i is less than initial transformation matrix less than the line number and j of initial transformation matrix
Columns.
Compared with prior art, the invention has the benefit that
1, compared with conventional three-dimensional method for reconstructing, real-time, high-precision face three is may be implemented using depth camera in the present invention
Dimension rebuilds process flow, and solve traditional can not generate three-dimensional mould because arithmetic accuracy is low or calculation amount is excessively huge in real time
The problem of type；
2, the present invention can not quickly identify that face location, fast speed generate a frame face three-dimensional mould by RGB information
The time of type is within 0.07 second.
Detailed description of the invention
Fig. 1 is a kind of face's real-time three-dimensional method for reconstructing flow chart provided in an embodiment of the present invention；
Fig. 2 is that the detection of the embodiment of the present invention positions face's schematic diagram；
Fig. 3 is the quick gridding schematic diagram of point cloud of the embodiment of the present invention.
Specific embodiment
For a further understanding of the present invention, technical solution of the present invention is retouched in detail with reference to the accompanying drawings and examples
It states.
In one embodiment, as shown in Figure 1, a kind of face's real-time three-dimensional method for reconstructing, comprising:
Step S100: point cloud data frame is obtained by depth perception equipment；
Step S200: carrying out pretreatment to point cloud data and form foreground point cloud, the pretreatment include a point cloud background filter out,
It simplifies and denoises；
Step S300: face's point cloud is extracted from the cloud of foreground point, and face's point cloud and trunk point cloud are split；
Step S400: being registrated using ICP algorithm face's point cloud pretreated with previous frame to this frame face point cloud,
Face's point cloud under different coordinates is rotated, moves to the same coordinate system, weight smooth by quick gridding and Laplce
Build the threedimensional model of present frame point cloud；
Step S500: repeating step S100- step S400, generates face's threedimensional model in real time.
The present invention using depth camera can real-time, high-precision realization face three-dimensional reconstruction, overcome in conventional method
Because arithmetic accuracy is low or calculation amount huge the problem of cannot generating human face three-dimensional model in real time；The present invention by point cloud data into
Row pretreatment good cloud topological structure of building, then point cloud data is registrated by ICP algorithm, without by RGB
Information can quickly identify face location and construct face's threedimensional model.
In one embodiment, it in the specific embodiment of step S100, is obtained in space by depth perception equipment
Point cloud data, the point cloud data refer to the set of one group of vector in a three-dimensional coordinate system, these vectors usually with
The form of (X, Y, Z) three-dimensional coordinate indicates, and is generally mainly used to represent the external surface shape of an object.In addition, except (X,
Y, Z) except the geometric position information that represents, point cloud data also may indicate that the RGB color of a point, gray value, depth and point
Cut result.
In one embodiment, it in the specific embodiment of step S200, needs to locate the point cloud data of acquisition in advance
It manages, the pretreatment in the present embodiment includes that a point cloud background filters out, point cloud compressing and point cloud denoise.First, from depth perception equipment
It include a large amount of redundancy background information in the point cloud data stream directly acquired, it is therefore, it is necessary to carry out a cloud background to filter out, i.e., logical
It crosses and chooses depth value and put the range of cloud z coordinate namely to remove the redundancy background information in a cloud, in the present embodiment, put cloud z
Coordinate ranges preferably from 200mm-600mm；Second, bounding box method, uniform grid can be used when point cloud data is more dense
Method, Triangular meshes method and curvature sampling method carry out data compaction, and the present embodiment preferably uses bounding box method, i.e., will put cloud location
Domain is divided into several cubes according to actual physics section is equidistant, each cubic volume be set as 1mm × 1mm × 1mm or
2mm × 2mm × 2mm, and each cube only retains a data point, the data point retained, which is used as in the cube, to be owned
The center of point cloud.Meanwhile it putting there are much noise in cloud, the point that will there is similar quality in cube using algorithm of region growing
Incorporate into the same set, can effectively remove floated in a cloud at block noise and peel off spot noise.
In one embodiment, in the specific embodiment of step S300, face is extracted, generates face's point cloud,
It is specific as shown in Figure 2.Formation foreground point cloud after point cloud filters out by background, simplifies denoising, further, from the cloud of foreground point
Only retain face's point cloud except neck, body etc. point cloud；Further, face's point cloud and trunk point cloud are split.?
In the cloud of foreground point, its Laplace operator is sought on the direction y, that is, vertical direction, formula is as follows:
Wherein, z (n) indicates n-th point on vertical direction of depth value.In the depth of the place that face and chin have a common boundary point cloud
Angle value quickly changes, i.e. the value of Laplace operator is larger, to well separate face and trunk.
In one embodiment, in the specific embodiment of step S400, using traditional ICP algorithm to this frame face point
The cloud face point cloud pretreated with previous frame is registrated, by under different coordinates face's point cloud rotation, move to it is same
Coordinate system specifically includes:
1) internal reference of depth camera is demarcated by Zhang Zhengyou calibration method；
2) shooting previous frame depth map and this frame depth map extract feature point set p={ p1, p2... pnAnd q={ q1,
q2... qn, piAnd qiFor the corresponding characteristic point of two frame depth maps；
3) eigenmatrix E is solved according to matched characteristic point, obtains spin matrix R and translation matrix T by decomposing；
4) using spin matrix R and translation matrix T as the initial transformation matrix of ICP algorithm.
After obtaining initial transformation matrix, as shown in figure 3, carrying out face's threedimensional model by the way of orderly point cloud line
It rebuilds, specifically includes.
1) cloud is orderly put, if initial point P (i, j)=P (1,1), i=j=1；
2) point centered on P (i, j), sequentially generates tri patch: { P (i, j), P (i-1, j-1), P (i-1, j) }, { P
(i, j), P (i-1, j), P (i-1, j+1) }, { P (i, j), P (i-1, j+1), P (i, j+1) }, { P (i, j), P (i, j+1), P (i+
1, j+1) }, { P (i, j), P (i+1, j+1), P (i+1, j) }, { P (i, j), P (i+1, j), P (i+1, j-1) }, { P (i, j), P (i
+ 1, j-1), P (i, j-1) }, { P (i, j), P (i, j-1), P (i-1, j-1) }；
3) i=i+2, j=j+2, return step 2), until i is less than initial transformation less than the line number and j of initial transformation matrix
Matrix column number.
After the completion of face's reconstructing three-dimensional model, to face's threedimensional model, progress Laplce is smooth, specific embodiment
Are as follows: within a grid, if the collection for all the points that a point P is connected is combined into SP, take SPThe central point of middle all the points is updated as point P
Point afterwards can generate the threedimensional model of present frame point cloud according to being iterated in this way to all the points in cloud.
In one embodiment, in the specific embodiment of step S500, step S100- step S400, energy are repeated
Enough final generation face's threedimensional models in real time.
Real-time, high-precision face three-dimensional reconstruction process flow may be implemented using depth camera in the present invention, solves
Traditional arithmetic accuracy is low or calculation amount is excessively huge and the problem of threedimensional model can not be generated in real time.In addition, energy of the present invention
It is enough not identify face location quickly by RGB information, the time of a frame face threedimensional model is generated within 0.07 second, is had simultaneously
Standby good result.
The above is only a preferred embodiment of the present invention, protection scope of the present invention is not limited merely to above-mentioned implementation
Example, all technical solutions belonged under thinking of the present invention all belong to the scope of protection of the present invention.It should be pointed out that for the art
Those of ordinary skill for, several improvements and modifications without departing from the principles of the present invention also should be regarded as of the invention
Protection scope.
Claims (9)
1. a kind of face's real-time three-dimensional method for reconstructing, includes the following steps:
S100: point cloud data is obtained by depth perception equipment；
S200: pretreatment is carried out to point cloud data and forms foreground point cloud, the pretreatment includes that point cloud background filters out, simplifies and go
It makes an uproar；
S300: face's point cloud is extracted from the cloud of foreground point, and face's point cloud and trunk point cloud are split；
S400: it is registrated using the ICP algorithm face point cloud pretreated with previous frame to this frame face point cloud, by different seats
Face's point cloud under mark system rotates, moves to the same coordinate system, smooth by quick gridding and Laplce, rebuilds present frame
The threedimensional model of point cloud；
S500: repeating step S100- step S400, generates face's threedimensional model in real time.
2. in step s 200, described cloud background filters out the method according to claim 1, wherein preferred
Refer to that by choosing depth value be the range of point cloud z coordinate to remove the redundancy background information in a cloud.
3. according to right require 1 described in method, which is characterized in that in step s 200, the point cloud compressing is using surrounding
Any one in box method, uniform grid method, Triangular meshes method and curvature sampling method.
4. according to the method described in claim 3, it is characterized in that, the bounding box method includes:
1) cloud region is divided into several cubes according to actual physics section is equidistant；
2) retain a data point in each cube, using the data point as the center of all the points cloud in the cube.
5. the method according to claim 1, wherein in step s 200, described cloud denoising refers to cube
In be grouped into the same set with the click and sweep of similar quality.
6. according to the method described in claim 2, it is characterized in that, the range of described cloud z coordinate is 200mm-600mm.
7. according to the method described in claim 4, it is characterized in that, the cubic volume is set as 1mm × 1mm × 1mm or 2mm
×2mm×2mm。
8. the method according to claim 1, wherein described use ICP algorithm to this frame face point cloud and upper one
The pretreated face's point cloud of frame carries out registration and specifically includes:
1) internal reference of depth camera is demarcated by Zhang Zhengyou calibration method；
2) shooting previous frame depth map and this frame depth map extract feature point set p={ p1, p2... pnAnd q={ q1, q2...
qn, piAnd qiFor the corresponding characteristic point of two frame depth maps；
3) eigenmatrix E is solved according to matched characteristic point, obtains spin matrix R and translation matrix T by decomposing；
4) using spin matrix R and translation matrix T as the initial transformation matrix of ICP algorithm.
9. according to the method described in claim 8, it is characterized in that, it is described rebuild present frame point cloud threedimensional model refer to
To after the initial transformation matrix, is rebuild, is specifically included by the way of orderly point cloud line:
1) initial point P (i, j)=P (1,1) is set, i=j=1；
2) point centered on initial point P (i, j), sequentially generates tri patch；
3) i=i+2, j=j+2, return 2), until i be less than initial transformation matrix line number and j be less than initial transformation matrix column
Number.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810888258.8A CN109242951A (en) | 2018-08-06 | 2018-08-06 | A kind of face's real-time three-dimensional method for reconstructing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810888258.8A CN109242951A (en) | 2018-08-06 | 2018-08-06 | A kind of face's real-time three-dimensional method for reconstructing |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109242951A true CN109242951A (en) | 2019-01-18 |
Family
ID=65070236
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810888258.8A Pending CN109242951A (en) | 2018-08-06 | 2018-08-06 | A kind of face's real-time three-dimensional method for reconstructing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109242951A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110060336A (en) * | 2019-04-24 | 2019-07-26 | 北京华捷艾米科技有限公司 | Three-dimensional facial reconstruction method, device, medium and equipment |
| CN110415351A (en) * | 2019-06-21 | 2019-11-05 | 北京迈格威科技有限公司 | Methods, devices and systems based on single image building three-dimensional grid |
| CN110694921A (en) * | 2019-09-06 | 2020-01-17 | 北京农业智能装备技术研究中心 | Vegetable sorting system and control method thereof |
| WO2020199834A1 (en) * | 2019-04-03 | 2020-10-08 | 腾讯科技（深圳）有限公司 | Object detection method and apparatus, and network device and storage medium |
| CN112069923A (en) * | 2020-08-18 | 2020-12-11 | 东莞正扬电子机械有限公司 | 3D face point cloud reconstruction method and system |
| CN113393505A (en) * | 2021-06-25 | 2021-09-14 | 浙江商汤科技开发有限公司 | Image registration method, visual positioning method, related device and equipment |
| CN114565689A (en) * | 2022-02-28 | 2022-05-31 | 燕山大学 | Axial symmetry three-dimensional model data compression reconstruction method |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106803267A (en) * | 2017-01-10 | 2017-06-06 | 西安电子科技大学 | Indoor scene three-dimensional rebuilding method based on Kinect |
-
2018
- 2018-08-06 CN CN201810888258.8A patent/CN109242951A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106803267A (en) * | 2017-01-10 | 2017-06-06 | 西安电子科技大学 | Indoor scene three-dimensional rebuilding method based on Kinect |
Non-Patent Citations (2)
| Title |
|---|
| 庞浩等: "ICP算法的改进及两台Kinect对人体的重建", 《大连工业大学学报》 * |
| 杨海清等: "基于多Kinect的三维人脸重建研究", 《浙江工业大学学报》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020199834A1 (en) * | 2019-04-03 | 2020-10-08 | 腾讯科技（深圳）有限公司 | Object detection method and apparatus, and network device and storage medium |
| CN110060336A (en) * | 2019-04-24 | 2019-07-26 | 北京华捷艾米科技有限公司 | Three-dimensional facial reconstruction method, device, medium and equipment |
| CN110415351A (en) * | 2019-06-21 | 2019-11-05 | 北京迈格威科技有限公司 | Methods, devices and systems based on single image building three-dimensional grid |
| CN110415351B (en) * | 2019-06-21 | 2023-10-10 | 北京迈格威科技有限公司 | Method, device and system for constructing three-dimensional grid based on single image |
| CN110694921A (en) * | 2019-09-06 | 2020-01-17 | 北京农业智能装备技术研究中心 | Vegetable sorting system and control method thereof |
| CN112069923A (en) * | 2020-08-18 | 2020-12-11 | 东莞正扬电子机械有限公司 | 3D face point cloud reconstruction method and system |
| CN113393505A (en) * | 2021-06-25 | 2021-09-14 | 浙江商汤科技开发有限公司 | Image registration method, visual positioning method, related device and equipment |
| CN113393505B (en) * | 2021-06-25 | 2023-11-03 | 浙江商汤科技开发有限公司 | Image registration method, visual positioning method, related device and equipment |
| CN114565689A (en) * | 2022-02-28 | 2022-05-31 | 燕山大学 | Axial symmetry three-dimensional model data compression reconstruction method |
| CN114565689B (en) * | 2022-02-28 | 2024-02-02 | 燕山大学 | Axisymmetric three-dimensional model data compression reconstruction method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109242951A (en) | A kind of face's real-time three-dimensional method for reconstructing | |
| CN110264567B (en) | Real-time three-dimensional modeling method based on mark points | |
| CN103021017B (en) | Three-dimensional scene rebuilding method based on GPU acceleration | |
| Livny et al. | Automatic reconstruction of tree skeletal structures from point clouds | |
| CN111696210A (en) | Point cloud reconstruction method and system based on three-dimensional point cloud data characteristic lightweight | |
| CN106651900B (en) | A kind of overhead strawberry three-dimensional modeling method in situ based on contours segmentation | |
| CN113178009B (en) | Indoor three-dimensional reconstruction method utilizing point cloud segmentation and grid repair | |
| Chen et al. | Uniform offsetting of polygonal model based on layered depth-normal images | |
| CN111899328B (en) | Point cloud three-dimensional reconstruction method based on RGB data and generation countermeasure network | |
| US20120313937A1 (en) | Coupled reconstruction of hair and skin | |
| Zhang et al. | Data-driven synthetic modeling of trees | |
| CN111243093A (en) | Three-dimensional face grid generation method, device, equipment and storage medium | |
| CN102222357B (en) | Foot-shaped three-dimensional surface reconstruction method based on image segmentation and grid subdivision | |
| Long et al. | Neuraludf: Learning unsigned distance fields for multi-view reconstruction of surfaces with arbitrary topologies | |
| CN104331924B (en) | Three-dimensional rebuilding method based on single camera SFS algorithms | |
| CN109242954A (en) | Multi-view angle three-dimensional human body reconstruction method based on template deformation | |
| CN107657659A (en) | The Manhattan construction method for automatic modeling of scanning three-dimensional point cloud is fitted based on cuboid | |
| CN113012293A (en) | Stone carving model construction method, device, equipment and storage medium | |
| CN106023147B (en) | The method and device of DSM in a kind of rapidly extracting linear array remote sensing image based on GPU | |
| CN111060006A (en) | Viewpoint planning method based on three-dimensional model | |
| CN107633518A (en) | A kind of product design detection method based on Kinect | |
| Zhang et al. | Data-driven flower petal modeling with botany priors | |
| Osman Ulusoy et al. | Dynamic probabilistic volumetric models | |
| Yu et al. | Incremental reconstruction of manifold surface from sparse visual mapping | |
| CN116152417A (en) | Multi-viewpoint perspective space fitting and rendering method and device |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190118 |