CN103021017B - Three-dimensional scene rebuilding method based on GPU acceleration - Google Patents
Three-dimensional scene rebuilding method based on GPU acceleration Download PDFInfo
- Publication number
- CN103021017B CN103021017B CN201210513744.4A CN201210513744A CN103021017B CN 103021017 B CN103021017 B CN 103021017B CN 201210513744 A CN201210513744 A CN 201210513744A CN 103021017 B CN103021017 B CN 103021017B
- Authority
- CN
- China
- Prior art keywords
- point
- cloud
- dimensional
- gpu
- image
- 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.)
- Expired - Fee Related
Links
Landscapes
- Processing Or Creating Images (AREA)
- Image Generation (AREA)
Abstract
The invention provides a three-dimensional scene rebuilding system based on GPU (Graphics Processing Unit). The method comprises that: calculating feature points of two-dimensional image in a way of SIFT (scale-invariant feature transform) algorithm parallel to GPU speedup, matching the feature points, every image camera calibration, dense points cloud reconstruction, point cloud filtering, denoising, point cloud meshing and feature mapping through the GPU to get a three-dimensional scene. The three-dimensional scene rebuilding system based on GPU can output a final three-dimensional model by inputting a series of two dimensional images and the overall process is automatic with no need of manual intervention. The system has the advantages of high reconstruction speed, high accuracy and low error.
Description
Technical field
The present invention relates to computer vision and field of Computer Graphics, particularly, relate to a kind of based on GPU(Graphic Processing Unit, graphic process unit) method for reconstructing three-dimensional scene that accelerates.Background technology
At computer vision field, reconstruct the three-dimensional point cloud of object from two dimensional image, be one of the primary study problem in this field always.It not only combines numerous subject knowledges, and has boundless application prospect at numerous areas such as aeroplane mapping, scenario simulation, vision guided navigation, medical diagnosis, historical relic's protection, ecommerce, virtual realities.Three-dimensional rebuilding method based on image makes full use of the relevant knowledge of computer vision and computer graphics, recovers the three-dimensional model of object from the single width or multiple image of actual photographed, it can be thought the inverse process that camera is taken pictures.China's Chang'e I ring moon satellite has passed the three-dimensional image of moonscape back, and it has also declared publicly the huge applications prospect of three-dimensional reconstruction to common people.How more realistically, the three-dimensional model obtaining real world easily impels computer vision research person constantly propose new method and improve existing method.
The technological means of computing machine construction body three-dimensional models is divided into following three kinds usually: the method for geometric modeling, obtain the method for three-dimensional model and the three-dimensional rebuilding method based on image with three-dimensional scanning device.
The method of geometric modeling refer to by the geometric transformation such as translation, rotation, no-load voltage ratio of the geometric element such as point, line, surface, body and also, hand over, the set operation such as difference, produce object model that the is actual or imagination.Geometric modeling method originates from the seventies in last century, and oneself is through being widely used in the fields such as computer-aided design (CAD), cartoon making, ad production, production of film and TV now.At present, popular much outstanding modeling software on the market, as 3DMAX, Maya, AutoCAD etc.They are all the geometric elements utilizing some basic, are carried out the model of complex structure by a series of geometric operation.This method requires fully to grasp contextual data, simultaneously the operation more complicated of related software, and often need skilled operating personnel and possess abundant professional knowledge, production process is relatively complicated.
Utilize three-dimensional scanning device can obtain accurate three-dimensional model, these equipment comprise depth scan instrument, three-dimensional camera, laser instrument etc.The measuring head structure different according to scanner, can be divided into contact and contactless two kinds, wherein, contact measuring head can be divided into again hard gauge head and soft survey first two, contact measuring head needs directly to contact with testee, the three-dimensional information of acquisition object.Contactless gauge head make use of optics and Principles of Laser, is called three-dimensional laser scanner.Utilize three-dimensional scanning device to use simple to the method that object carries out three-dimensional reconstruction, precision is very high, and builds model and take time relatively less, is therefore widely used in the fields such as reverse-engineering, virtual reality, environmental simulation.But these high-precision equipment are all very expensive usually, therefore do not have good popularization.
Three-dimensional rebuilding method based on image makes full use of the relevant knowledge of computer vision and computer graphics, the three-dimensional model of object is recovered from the single width or multiple image of actual photographed, it can be thought the inverse process that camera is taken pictures, if publication number is the method for reconstructing three-dimensional scene of the single image disclosed in the Chinese invention patent of 101714262A.At present, by updating method for reconstructing, modeling process is robotization more and more, and hand labor intensity is more and more lighter, and modeling cost is more and more lower.And in equipment, only need a common digital camera, and cheap, be applicable to the reconstruct of any scene, and the three-dimensional model with " photo level " high realism can be obtained.
Method at present based on image reconstruction three-dimensional point cloud mainly contains two classes: SFM(Structure From Motion, from movable information, recover three-dimensional scenic) and SFS(Shape From Silhouette, from profile Information recovering three-dimensional scenic).The benefit of first kind method to obtain more details, but can produce noise; The benefit of Equations of The Second Kind method can carry out extraordinary process to boundary problem, but it is worse for the process of details.
Summary of the invention
For defect of the prior art, the object of this invention is to provide a kind of method for reconstructing three-dimensional scene accelerated based on GPU, three-dimensional scenic can be obtained fast automatically.
For achieving the above object, the technical solution used in the present invention is: carry out GPU to two dimensional image and to walk abreast SIFT(Scale-invariant feature transform, scale invariant feature is changed) unique point calculating, and carry out Feature Points Matching, then camera calibration is carried out to every width image, carry out dense point cloud reconstruction again, then a cloud filtering is carried out, denoising, then a cloud gridding, carry out texture finally by GPU, obtain three-dimensional scenic.
Particularly, a kind of method for reconstructing three-dimensional scene accelerated based on GPU, it is by following steps specific implementation:
Step one, use camera in different positions, different angles takes pictures to scene, obtains the two-dimensional image sequence of real scene.
The SIFT algorithm of step 2, use GPU parallel accelerate obtains the unique point of every width image, and carries out Feature Points Matching.
Step 3, use Bundle Adjustment(bundle adjustment) algorithm, automatically obtain the camera parameter (camera matrix) of all images, and obtain unique point position in three dimensions, generate initial sparse point cloud model.
Step 4, use PMVS(Patch-based Multi-view Stereo, block-based various visual angles stereo scene) algorithm, the spatial point reconstructed by unique point is before referred to as Seed Points, diffusion, from Seed Points, utilizes consecutive point to have similar normal direction and the characteristic of position, progressively spreads the spatial point reconstructed around it, after diffusion terminates, carry out filtration treatment, point more weak to gray consistency, Geometrical consistency is rejected, obtains dense point cloud model.
Step 5, Iamge Segmentation is carried out to every piece image, obtain prospect and background, then Dian Yun back projection is passed through in image, the point that mistake generates is removed, first time refining point cloud, then use low-pass filter to carry out filtering to three-dimensional model, filter some isolated points (high frequency noise), second time refining point cloud.
Step 6, KNN(k-Nearest Neighbor algorithm is used to the some cloud generated, the most neighbouring node algorithm of K) thought calculation level cloud in the normal vector of each point, then by the cell(grid of vector projection to a sphere) in, the principal direction of calculation level cloud, this direction and third direction, calculate rotation matrix again, by consistent with three-dimensional change in coordinate axis direction for the main normal direction of a cloud.
The storage organization of step 7, reorganization point cloud, some cloud is placed in the Octree of space, and then carry out A* search from the point of the top, the search triangle of increment is unilateral, will put cloud gridding.
Step 8, from two dimensional image, find gridding after texture corresponding to all triangular plate faces, then carry out texture by GPU, final rendering goes out whole scene.
Preferably, in step one, the camera that adopts obtains the method for the two-dimensional image sequence of destination object, need take in different angles and different positions, only can not stand in a fixing position shooting.
Preferably, in step 2, the GPU that adopts accelerates SIFT, and system needs to install CUDA(Compute UnifiedDevice Architecture, the computing platform that video card manufacturer NVidia releases) running environment, ensure that SIFT can walk abreast by GPU process.
Preferably, in step 4, adopt PMVS to carry out the acquisition of dense point cloud, need image grid division, the size of grid can be selected, and grid is less, and the some cloud of acquisition is denser.
Preferably, in step 5, first image is carried out to the segmentation of prospect and background, the mode of then adopted back projection, for by Dian Yun back projection in every width image, if dropped on a little outside prospect or image, then retain this point, if project in background, then delete this point.Re-use low-pass filter, filter isolated point.
Preferably, in step 7, based on the gridding of A* search, will ensure two conditions when generation triangle gridding: the first, triangle is full as much as possible, not too narrow, require that leg-of-mutton each angle is greater than 30 degree, the second, plane is smooth as far as possible, and two namely adjacent tri patchs will be similar to 180 degree.
Preferably, in step 8, texture can use OpenGL SL(Open Graphics Library ShadingLanguage, and language is played up in open graphic package storehouse) image is saved as texture, then binded texture, plays up.
Compared with prior art, the present invention has following beneficial effect:
The present invention, by carrying out multi-angle to a scene, multipointly to take pictures, and can obtain three-dimensional scenic fast automatically.The present invention mainly contains three remarkable advantages: 1. the present invention is by GPU parallel computation SIFT feature point, accelerates the speed of rebuilding, and solves the slow-footed problem of three-dimensional reconstruction; 2. the present invention is by using the way of SFM and SFS fusion, successfully solves the problem on three-dimensional reconstruction border; 3. the present invention does not need manual intervention, can realize rebuilding robotization completely.
Accompanying drawing explanation
By reading the detailed description done non-limiting example with reference to the following drawings, other features, objects and advantages of the present invention will become more obvious:
Fig. 1 is the process flow diagram that the present invention is based on the method for reconstructing three-dimensional scene that GPU accelerates.
Fig. 2 is the result of an Embodiment B undle Adjustment.
Fig. 3 is dense point cloud embodiment illustrated in fig. 2.
Fig. 4 is net result embodiment illustrated in fig. 2.
Embodiment
Below in conjunction with specific embodiment, the present invention is described in detail.Following examples will contribute to those skilled in the art and understand the present invention further, but not limit the present invention in any form.It should be pointed out that to those skilled in the art, without departing from the inventive concept of the premise, some distortion and improvement can also be made.These all belong to protection scope of the present invention.
As shown in Figure 1, use the method for reconstructing three-dimensional scene accelerated based on GPU, obtained the Three-dimension Reconstruction Model of real scene by following steps:
Step one, use camera in different positions, different angles takes pictures to scene, obtains the two-dimensional image sequence of real scene.
The SIFT algorithm of step 2, use GPU parallel accelerate obtains the unique point of every width image.And carry out Feature Points Matching.The GPU that adopted accelerates SIFT, and system installs CUDA(Compute Unified DeviceArchitecture, the computing platform that video card manufacturer NVidia releases) running environment, ensure that SIFT can walk abreast by GPU process.
Step 3, use Bundle Adjustment algorithm, automatically obtain the camera parameter (camera matrix) of all images, and obtain unique point position in three dimensions, generate initial sparse point cloud model, as Fig. 2.
Step 4, use PMVS algorithm, the spatial point reconstructed by unique point is before referred to as Seed Points, diffusion is from Seed Points, consecutive point are utilized to have similar normal direction and the characteristic of position, progressively spread the spatial point reconstructed around it, after diffusion terminates, carry out filtration treatment, point more weak to gray consistency, Geometrical consistency is rejected. obtain dense point cloud model.As Fig. 3.Adopt PMVS to carry out the acquisition of dense point cloud, need image grid division, the size of grid can be selected, and grid is less, and the some cloud of acquisition is denser.
Step 5, Iamge Segmentation is carried out to every piece image, obtain prospect and background, then Dian Yun back projection is passed through in image, the point that mistake generates is removed, first time refining point cloud, then use low-pass filter to carry out filtering to three-dimensional model, filter some isolated points (high frequency noise), second time refining point cloud.In this step, Dian Yun back projection in every width image, if dropped on a little outside prospect or image, has then been retained this point, if project in background, has then deleted this point, re-use low-pass filter, filter isolated point by the mode of employing back projection.
Step 6, the some cloud generated is used to KNN thought calculation level cloud in the normal vector of each point, then by vector projection in the cell of a sphere, the principal direction of calculation level cloud, secondary direction and third direction, calculate rotation matrix again, by consistent with three-dimensional change in coordinate axis direction for the main normal direction of a cloud.
The storage organization of step 7, reorganization point cloud, some cloud is placed in the Octree of space, and then carry out A* search from the point of the top, the search triangle of increment is unilateral, will put cloud gridding.In this step, based on the gridding of A* search, two conditions to be ensured: first when generation triangle gridding, triangle is full as much as possible, not too narrow, require that leg-of-mutton each angle is greater than 30 degree, second, plane is smooth as far as possible, and two namely adjacent tri patchs will be similar to 180 degree.
Step 8, from two dimensional image, find gridding after texture corresponding to all triangular plate faces, then carry out texture by GPU, final rendering goes out whole scene.Texture can use OpenGL SL(Open GraphicsLibrary Shading Language, and language is played up in open graphic package storehouse) image is saved as texture, then binded texture, plays up.Final result as shown in Figure 4.
The present embodiment, by carrying out multi-angle to a scene, multipointly to be taken pictures, and can obtain three-dimensional scenic fast automatically, by GPU parallel computation SIFT feature point, solves the slow-footed problem of three-dimensional reconstruction, solves the problem on three-dimensional reconstruction border simultaneously.
Above specific embodiments of the invention are described.It is to be appreciated that the present invention is not limited to above-mentioned particular implementation, those skilled in the art can make various distortion or amendment within the scope of the claims, and this does not affect flesh and blood of the present invention.
Claims (7)
1., based on the method for reconstructing three-dimensional scene that GPU accelerates, it is characterized in that, it is made up of following steps:
Step one, use camera in different positions, different angles takes pictures to scene, obtains the two-dimensional image sequence of real scene;
The SIFT algorithm of step 2, use GPU parallel accelerate obtains the unique point of every width image, and carries out Feature Points Matching;
Step 3, use Bundle Adjustment algorithm, automatically obtain camera parameter and the camera matrix of all images, and obtain unique point position in three dimensions, generate initial sparse point cloud model;
Step 4, use PMVS algorithm, the spatial point reconstructed by unique point is before referred to as Seed Points, diffusion is from Seed Points, consecutive point are utilized to have similar normal direction and the characteristic of position, progressively spread the spatial point reconstructed around it, after diffusion terminates, carry out filtration treatment, point more weak to gray consistency, Geometrical consistency is rejected, obtains dense point cloud model;
Step 5, Iamge Segmentation is carried out to every piece image, obtain prospect and background, then Dian Yun back projection is passed through in image, the point that mistake generates is removed, first time refining point cloud, then use low-pass filter to carry out filtering to three-dimensional model, filter some high frequency noise isolated points, second time refining point cloud;
Step 6, the some cloud generated is used to KNN thought calculation level cloud in the normal vector of each point, then by vector projection in the cell of a sphere, the principal direction of calculation level cloud, secondary direction and third direction, calculate rotation matrix again, by consistent with three-dimensional change in coordinate axis direction for the main normal direction of a cloud; The main normal direction of described some cloud gives directions the principal direction of cloud, secondary direction and third direction;
The storage organization of step 7, reorganization point cloud, some cloud is placed in the Octree of space, and then carry out A* search from the point of the top, the search triangle of increment is unilateral, will put cloud gridding;
Step 8, from two dimensional image, find gridding after texture corresponding to all triangular plate faces, then carry out texture by GPU, final rendering goes out whole scene.
2. a kind of method for reconstructing three-dimensional scene accelerated based on GPU according to claim 1, it is characterized in that: the camera adopted in step one obtains the method for the two-dimensional image sequence of destination object, need take in different angles and different positions, only can not stand in a fixing position shooting.
3. a kind of method for reconstructing three-dimensional scene accelerated based on GPU according to claim 1, it is characterized in that: the GPU that adopts in step 2 accelerates SIFT, system installs the running environment of CUDA, ensure that SIFT can walk abreast by GPU process.
4. a kind of method for reconstructing three-dimensional scene accelerated based on GPU according to claim 1, it is characterized in that: in step 4, adopt PMVS to carry out the acquisition of dense point cloud, need to image grid division, the size of grid can be selected, grid is less, and the some cloud of acquisition is denser.
5. a kind of method for reconstructing three-dimensional scene accelerated based on GPU according to claim 1, it is characterized in that: the segmentation first image being carried out to prospect and background in step 5, the mode of then adopted back projection, by Dian Yun back projection in every width image, if dropped on a little outside prospect or image, then retain this point, if project in background, then delete this point, re-use low-pass filter, filter isolated point.
6. a kind of method for reconstructing three-dimensional scene accelerated based on GPU according to claim 1, it is characterized in that: based on the gridding of A* search in step 7, two conditions to be ensured: first when generating triangle gridding, triangle is full as much as possible, not too narrow, leg-of-mutton each angle is greater than 30 degree, second, plane is smooth as far as possible, and two namely adjacent tri patchs will be similar to 180 degree.
7. a kind of method for reconstructing three-dimensional scene accelerated based on GPU according to claim 1, is characterized in that: the texture in step 8, and use OpenGL SL to be saved as texture by image, then binded texture, plays up.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210513744.4A CN103021017B (en) | 2012-12-04 | 2012-12-04 | Three-dimensional scene rebuilding method based on GPU acceleration |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210513744.4A CN103021017B (en) | 2012-12-04 | 2012-12-04 | Three-dimensional scene rebuilding method based on GPU acceleration |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103021017A CN103021017A (en) | 2013-04-03 |
| CN103021017B true CN103021017B (en) | 2015-05-20 |
Family
ID=47969583
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210513744.4A Expired - Fee Related CN103021017B (en) | 2012-12-04 | 2012-12-04 | Three-dimensional scene rebuilding method based on GPU acceleration |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103021017B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106887037A (en) * | 2017-01-23 | 2017-06-23 | 杭州蓝芯科技有限公司 | A kind of indoor three-dimensional rebuilding method based on GPU and depth camera |
| CN107507274A (en) * | 2017-08-30 | 2017-12-22 | 北京图航科技有限公司 | A kind of quick restoring method of public security criminal-scene three-dimensional live based on cloud computing |
| CN108629835A (en) * | 2017-03-20 | 2018-10-09 | 哈尔滨工业大学 | Based on EO-1 hyperion, true coloured picture and the indoor method for reconstructing and system for putting cloud complementation |
Families Citing this family (45)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103247045B (en) * | 2013-04-18 | 2015-12-23 | 上海交通大学 | A kind of method obtaining artificial scene principal direction and image border from multi views |
| CN103247041B (en) * | 2013-05-16 | 2016-02-10 | 北京建筑工程学院 | A kind of dividing method of the cloud data of the many geometric properties based on local sampling |
| WO2014186970A1 (en) * | 2013-05-23 | 2014-11-27 | Google Inc. | Systems and methods for generating three-dimensional models using sensed position data |
| CN103336959B (en) * | 2013-07-19 | 2016-09-28 | 西安电子科技大学 | A kind of vehicle checking method accelerated based on GPU multi-core parallel concurrent |
| US9265434B2 (en) * | 2013-12-18 | 2016-02-23 | Biosense Webster (Israel) Ltd. | Dynamic feature rich anatomical reconstruction from a point cloud |
| CN104112294B (en) * | 2014-07-04 | 2017-04-05 | 中国科学院上海光学精密机械研究所 | Intensity correlation imaging high speed three-dimensional reconfiguration system and method based on sparse constraint |
| CN104063899A (en) * | 2014-07-10 | 2014-09-24 | 中南大学 | Rock core shape-preserving three-dimensional reconstruction method |
| CN104952075A (en) * | 2015-06-16 | 2015-09-30 | 浙江大学 | Laser scanning three-dimensional model-oriented multi-image automatic texture mapping method |
| US9665978B2 (en) * | 2015-07-20 | 2017-05-30 | Microsoft Technology Licensing, Llc | Consistent tessellation via topology-aware surface tracking |
| JP6632709B2 (en) * | 2016-03-24 | 2020-01-22 | 富士ゼロックス株式会社 | Image processing apparatus, image processing method, and image processing program |
| CN105959175B (en) * | 2016-04-21 | 2019-10-22 | 南开大学 | Net flow assorted method based on the GPU kNN algorithm accelerated |
| CN107341768B (en) * | 2016-04-29 | 2022-03-11 | 微软技术许可有限责任公司 | Grid noise reduction |
| CN106097436B (en) * | 2016-06-12 | 2019-06-25 | 广西大学 | A kind of three-dimensional rebuilding method of large scene object |
| CN106600686B (en) * | 2016-12-06 | 2020-05-22 | 西安电子科技大学 | Three-dimensional point cloud reconstruction method based on multiple uncalibrated images |
| CN106683173B (en) * | 2016-12-22 | 2019-09-13 | 西安电子科技大学 | A method of Three-dimensional Gravity is improved based on neighborhood Block- matching and is laid foundations the dense degree of cloud |
| CN106887043A (en) * | 2017-03-08 | 2017-06-23 | 景致三维（江苏）股份有限公司 | The method of the method, device and three-dimensional modeling of the removal of three-dimensional modeling exterior point |
| CN107194985A (en) * | 2017-04-11 | 2017-09-22 | 中国农业大学 | A kind of three-dimensional visualization method and device towards large scene |
| CN107247834B (en) * | 2017-05-31 | 2019-09-24 | 华中科技大学 | A kind of three dimensional environmental model reconstructing method, equipment and system based on image recognition |
| CN107452037B (en) * | 2017-08-02 | 2021-05-14 | 北京航空航天大学青岛研究院 | GPS auxiliary information acceleration-based structure recovery method from movement |
| EP3467789A1 (en) * | 2017-10-06 | 2019-04-10 | Thomson Licensing | A method and apparatus for reconstructing a point cloud representing a 3d object |
| CN109658365B (en) * | 2017-10-11 | 2022-12-06 | 阿里巴巴(深圳)技术有限公司 | Image processing method, device, system and storage medium |
| CN107845073B (en) * | 2017-10-19 | 2020-02-14 | 华中科技大学 | Local self-adaptive three-dimensional point cloud denoising method based on depth map |
| CN107862718B (en) * | 2017-11-02 | 2020-01-24 | 深圳市自由视像科技有限公司 | 4D holographic video capture method |
| CN107767454A (en) * | 2017-11-10 | 2018-03-06 | 泰瑞数创科技(北京)有限公司 | A kind of three-dimensional mobile fast modeling method of outdoor scene, apparatus and system |
| CN109961516B (en) * | 2017-12-26 | 2023-09-26 | 宏达国际电子股份有限公司 | Surface acquisition method, device and non-transitory computer readable recording medium |
| CN108335352B (en) * | 2018-01-22 | 2020-05-19 | 华中科技大学 | Texture mapping method for multi-view large-scale three-dimensional reconstruction scene |
| CN108062758B (en) * | 2018-02-27 | 2019-06-28 | 山东师范大学 | A kind of crowd's generation emulation mode and system based on image segmentation algorithm |
| CN108520055A (en) * | 2018-04-04 | 2018-09-11 | 天目爱视（北京）科技有限公司 | A kind of product testing identification method compared based on temmoku point cloud |
| CN108259858B (en) * | 2018-04-10 | 2020-10-30 | 华雁智能科技(集团)股份有限公司 | Method and device for monitoring scene and equipment of transformer substation |
| CN108961151B (en) * | 2018-05-08 | 2019-06-11 | 中德（珠海）人工智能研究院有限公司 | A method of the three-dimensional large scene that ball curtain camera obtains is changed into sectional view |
| CN108734773A (en) * | 2018-05-18 | 2018-11-02 | 中国科学院光电研究院 | A kind of three-dimensional rebuilding method and system for mixing picture |
| CN108986218B (en) * | 2018-06-06 | 2023-03-31 | 东南大学 | PMVS-based rapid building dense point cloud reconstruction method |
| CN109118576A (en) * | 2018-07-19 | 2019-01-01 | 桂林电子科技大学 | Large scene three-dimensional reconstruction system and method for reconstructing based on BDS location-based service |
| CN109345557B (en) * | 2018-09-19 | 2021-07-09 | 东南大学 | Foreground and background separation method based on three-dimensional reconstruction result |
| CN109872280B (en) * | 2018-12-26 | 2023-03-14 | 江苏名通信息科技有限公司 | Denoising and simplifying method, device and system for three-dimensional plant leaf point cloud |
| CN110059651B (en) * | 2019-04-24 | 2021-07-02 | 北京计算机技术及应用研究所 | Real-time tracking and registering method for camera |
| CN110148099B (en) * | 2019-05-29 | 2021-09-10 | 北京百度网讯科技有限公司 | Projection relation correction method and device, electronic equipment and computer readable medium |
| CN110648389A (en) * | 2019-08-22 | 2020-01-03 | 广东工业大学 | 3D reconstruction method and system for city street view based on cooperation of unmanned aerial vehicle and edge vehicle |
| CN111063021B (en) * | 2019-11-21 | 2021-08-27 | 西北工业大学 | Method and device for establishing three-dimensional reconstruction model of space moving target |
| CN111508063A (en) * | 2020-04-13 | 2020-08-07 | 南通理工学院 | Three-dimensional reconstruction method and system based on image |
| CN111401482B (en) * | 2020-04-29 | 2024-03-19 | Oppo广东移动通信有限公司 | Feature point matching method and device, equipment and storage medium |
| CN112465984A (en) * | 2020-11-12 | 2021-03-09 | 西北工业大学 | Monocular camera sequence image three-dimensional reconstruction method based on double-layer filtering |
| CN112964174A (en) * | 2021-01-28 | 2021-06-15 | 上海工程技术大学 | Large-scale component shape error detection platform |
| CN112819774A (en) * | 2021-01-28 | 2021-05-18 | 上海工程技术大学 | Large-scale component shape error detection method based on three-dimensional reconstruction technology and application thereof |
| CN116704139A (en) * | 2023-08-07 | 2023-09-05 | 航天宏图信息技术股份有限公司 | Satellite image three-dimensional reconstruction method and device, electronic equipment and readable storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101593345A (en) * | 2009-07-01 | 2009-12-02 | 电子科技大学 | Three-dimensional medical image display method based on the GPU acceleration |
| CN101714262A (en) * | 2009-12-10 | 2010-05-26 | 北京大学 | Method for reconstructing three-dimensional scene of single image |
| CN102695040A (en) * | 2012-05-03 | 2012-09-26 | 中兴智能交通（无锡）有限公司 | Parallel high definition video vehicle detection method based on GPU |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2114086B1 (en) * | 2008-04-30 | 2012-12-26 | Renault S.A.S. | Ironless and leakage free coil transducer motor assembly |
-
2012
- 2012-12-04 CN CN201210513744.4A patent/CN103021017B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101593345A (en) * | 2009-07-01 | 2009-12-02 | 电子科技大学 | Three-dimensional medical image display method based on the GPU acceleration |
| CN101714262A (en) * | 2009-12-10 | 2010-05-26 | 北京大学 | Method for reconstructing three-dimensional scene of single image |
| CN102695040A (en) * | 2012-05-03 | 2012-09-26 | 中兴智能交通（无锡）有限公司 | Parallel high definition video vehicle detection method based on GPU |
Non-Patent Citations (4)
| Title |
|---|
| A Point-Cloud-Based Multiview Stereo Algorithm for Free-Viewpoint Video;Yebin Liu 等;《Visualization and Computer Graphics, IEEE Transactions on》;20100311;第16卷(第3期);407-418 * |
| Fusing Multiview and Photometric Stereo for 3D Reconstruction under Uncalibrated Illumination;Chenglei Wu 等;《Visualization and Computer Graphics, IEEE Transactions on》;20110609;第17卷(第8期);1082-1095 * |
| 基于GPU硬件加速的医学图像分割研究;韦轶群;《万方学位论文数据库》;20091231;1-66 * |
| 基于图像处理器的EWA Splatting 体绘制加速算法;秦红星 等;《上海交通大学学报》;20081228;第42卷(第12期);1943-1946 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106887037A (en) * | 2017-01-23 | 2017-06-23 | 杭州蓝芯科技有限公司 | A kind of indoor three-dimensional rebuilding method based on GPU and depth camera |
| CN106887037B (en) * | 2017-01-23 | 2019-12-17 | 杭州蓝芯科技有限公司 | indoor three-dimensional reconstruction method based on GPU and depth camera |
| CN108629835A (en) * | 2017-03-20 | 2018-10-09 | 哈尔滨工业大学 | Based on EO-1 hyperion, true coloured picture and the indoor method for reconstructing and system for putting cloud complementation |
| CN108629835B (en) * | 2017-03-20 | 2021-10-01 | 哈尔滨工业大学 | Indoor reconstruction method and system based on hyperspectral, true color image and point cloud complementation |
| CN107507274A (en) * | 2017-08-30 | 2017-12-22 | 北京图航科技有限公司 | A kind of quick restoring method of public security criminal-scene three-dimensional live based on cloud computing |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103021017A (en) | 2013-04-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103021017B (en) | Three-dimensional scene rebuilding method based on GPU acceleration | |
| Gortler et al. | The lumigraph | |
| CN109118582B (en) | Commodity three-dimensional reconstruction system and reconstruction method | |
| CN106803267B (en) | Kinect-based indoor scene three-dimensional reconstruction method | |
| CN111243093B (en) | Three-dimensional face grid generation method, device, equipment and storage medium | |
| Ecker et al. | Polynomial shape from shading | |
| CN106504320B (en) | Real-time three-dimensional reconstruction method based on GPU and facing to depth image | |
| CN111127633A (en) | Three-dimensional reconstruction method, apparatus, and computer-readable medium | |
| JP2024522287A (en) | 3D human body reconstruction method, apparatus, device and storage medium | |
| CN108961410A (en) | A kind of three-dimensional wireframe modeling method and device based on image | |
| CN103530907A (en) | Complicated three-dimensional model drawing method based on images | |
| Condorelli et al. | A comparison between 3D reconstruction using nerf neural networks and mvs algorithms on cultural heritage images | |
| Chen et al. | Research on 3D reconstruction based on multiple views | |
| Meerits et al. | Real-time scene reconstruction and triangle mesh generation using multiple RGB-D cameras | |
| JP2021026759A (en) | System and method for performing 3d imaging of objects | |
| CN108010122A (en) | A kind of human 3d model rebuilds the method and system with measurement | |
| Luan et al. | Research and development of 3D modeling | |
| CN109064550A (en) | A kind of body scans modeling method | |
| PP et al. | Efficient 3D visual hull reconstruction based on marching cube algorithm | |
| Ren et al. | A SKETCH-BASED 3D MODELING METHOD FOR 3D CRIME SCENE PRESENTATION | |
| CN112002019B (en) | Method for simulating character shadow based on MR mixed reality | |
| Zhu et al. | Plant Modeling Based on 3D Reconstruction and Its Application in Digital Museum. | |
| Han et al. | 3D reconstruction of dense model based on the sparse frames using RGBD camera | |
| Lin et al. | A novel tree-structured point cloud dataset for skeletonization algorithm evaluation | |
| Hao et al. | Image completion with perspective constraint based on a single image |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20150520 Termination date: 20171204 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |