CN111105864B - Method for dynamically transmitting 3D model - Google Patents
Method for dynamically transmitting 3D model Download PDFInfo
- Publication number
- CN111105864B CN111105864B CN201911372573.6A CN201911372573A CN111105864B CN 111105864 B CN111105864 B CN 111105864B CN 201911372573 A CN201911372573 A CN 201911372573A CN 111105864 B CN111105864 B CN 111105864B
- Authority
- CN
- China
- Prior art keywords
- model
- vertex
- scale
- grid
- algorithm
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H30/00—ICT specially adapted for the handling or processing of medical images
- G16H30/40—ICT specially adapted for the handling or processing of medical images for processing medical images, e.g. editing
-
- 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
- G06T17/20—Finite element generation, e.g. wire-frame surface description, tesselation
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Epidemiology (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Geometry (AREA)
- Computer Graphics (AREA)
- Radiology & Medical Imaging (AREA)
- Software Systems (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Primary Health Care (AREA)
- Public Health (AREA)
- Processing Or Creating Images (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a method for dynamically transmitting a 3D model, which comprises the following steps: 1) The server loads the 3D model; 2) The method comprises the steps that a server detects characteristic points of a 3D model and transmits the characteristic points to a client; 3) The client receives the characteristic points and performs surface reconstruction after receiving; 4) The server performs grid segmentation on the 3D model and transmits vertex data according to the modules; 5) The client receives the vertex of the module and rebuilds the local area; 6) And after all the modules are transmitted and rebuilt, grid simplification is carried out by using a QLIm algorithm. The method only needs to transmit the vertexes of the model, which greatly reduces the amount of data transmitted. By combining the technologies of feature point extraction, surface reconstruction, grid segmentation and the like, a user can quickly acquire the whole topology information of the model, and the accuracy and the detail degree of the model are continuously improved along with the increase of the number of the received vertexes.
Description
Technical Field
The invention relates to the field of computer graphics, in particular to a method for dynamically transmitting a 3D model.
Background
The computer graphic technology is widely applied in real life, a 3D model can greatly help a user to know related objects, in addition, the 3D model is a graphic file and has the characteristic of stretching, the model can not be blurred after being amplified, and the user can observe the model from multiple angles to acquire different information.
An important source of 3D models is the three-dimensional reconstruction of images, and scanning of medical devices, such as CT scanning, acquires a large amount of image data, and a large amount of 3D models are obtained by combining image segmentation techniques and three-dimensional reconstruction techniques. These models tend to be large, with a relatively large number of vertices and patches. The department in which the scanning device is located and the department in which the outpatient service of each department is located are often different, and the model needs to be transmitted by using a network. The traditional 3D model is often transmitted by storing files in PLY format and the like, the files are compressed and then transmitted by using a network, equipment at the other end can be decompressed and rendered and displayed by using corresponding software after the files are completely acquired, and if the model is large, a large amount of bandwidth is consumed, and the real-time acquisition of the information of the model by clients such as doctors is not facilitated.
Disclosure of Invention
The invention aims to overcome the defects and shortcomings of the prior art, and provides a method for dynamically transmitting a 3D model, which solves the problem that a user can display the model only after waiting for the model to be transmitted, and reduces the data volume of the model to be transmitted; the method uses characteristic point extraction and surface reconstruction technology, a user can acquire the whole information of the model in advance, and along with continuous acquisition of vertex data, the detail information of the model can be represented by using point clouds formed by vertexes, so that the user obtains the model with continuous perfection from the whole to the detail.
The aim of the invention is achieved by the following technical scheme:
a method of dynamically transmitting a 3D model, comprising the steps of:
1) The server loads the 3D model;
2) The method comprises the steps that a server detects characteristic points of a 3D model and transmits the characteristic points to a client;
3) The client receives the characteristic points and performs surface reconstruction after receiving;
4) The server performs grid segmentation on the 3D model and transmits vertex data according to the modules;
5) The client receives the vertex of the module and rebuilds the local area;
6) And after all the modules are transmitted and rebuilt, grid simplification is carried out by using a QLIm algorithm.
In step 1), the 3D model is a triangle mesh, vertices of the triangle mesh are represented by numerical values of (X, Y, Z) coordinates, and triangle patches are represented by IDs (V1, V2, V3) of the three vertices.
The triangle mesh format comprises PLY and STL.
In step 2), the server detects the feature points of the 3D model, and calculates local feature points by using the DoG operator, including the following steps:
2.1 Scale of computational grid
The scale of the grid is calculated using a local filter with uniform weights; for vertices with scale sIs +.>Can be calculated by the following formula:
2.2 Calculating local density
2.3 Calculating a DoG operator
Given two consecutive scales of a mesh model, the DoG function on scale sCalculated from the following formula:
2.4 Computing local scale and feature points of a grid
The local scale of the grid is defined by the following formula:
wherein s is i Is a filter, C is a parameter defined in advance to be applied to all feature descriptors, and is generally taken as 3; vertex pointThe condition for feature points on scale s is that they are locally maximum on both scale and spatial location. The number of the detected characteristic points is generally 5% -15% of the total number of the vertexes of the model.
And 3) after receiving the characteristic points, rendering and displaying the characteristic points by the client, and after receiving all the characteristic points, reconstructing the surface of the characteristic points by the client, wherein the reconstruction algorithm uses the Screen Poisson surface reconstruction, and the algorithm comprises the following steps:
3.1 Calculating the normal of the undirected point;
3.2 Constructing scale function fitting point cloud data, and adding vertex positions and gradient constraints;
3.3 Discretizing the scale function by using a linear basis function cluster;
3.4 Performing independent commander selection of the scale and calculation of the vertex boundary;
3.5 Constructing an octree grid and performing surface reconstruction by using a moving cube algorithm;
said step 4) segmenting the 3D model using a salient oriented 3D mesh segmentation algorithm comprising the steps of:
4.1 Extraction of significant vertices;
4.2 A) grouping of salient vertices;
4.3 Acquiring all vertices except the salient portion using a core approximation algorithm;
4.4 Calculating a partition boundary for each salient point;
4.5 Dividing a mesh model according to partition boundaries
After the model is divided, a plurality of modules are arranged, the vertexes of the modules can be randomly transmitted one by one, and the vertex information of the modules can be transmitted according to the specific sequence and the requirements of the client.
In step 5), the local area reconstruction is characterized as follows:
5.1 The vertex of each module is received for reconstruction;
5.2 A reconstructed region is limited to the region of the transferred module;
5.3 Using the local region reconstructed region to replace the corresponding region reconstructed using the feature points.
The step 6) is specifically as follows: the server transmits the vertex number and the surface patch information of the original model to the client, and the client uses QSLim algorithm to simplify according to the information, so that the accuracy of the original model is simplified, and the vertex number and the surface patch number of the reconstructed surface are reduced. The number of vertices and patches of the surface reconstructed by the Screen Poisson algorithm is more than that of the original model, usually 1-1.5 times more, so that the surface reconstructed by the Screen Poisson algorithm can be simplified by using a Qnlim algorithm and restored to the original number of vertices and patches.
Compared with the prior art, the invention has the following advantages and beneficial effects:
1. the invention only transmits the vertex information of the model, thereby greatly reducing the transmitted data volume.
2. The user of the invention can acquire the information of the corresponding model in real time, including the point cloud and the grid surface.
3. The invention uses the techniques of feature point extraction, surface reconstruction, grid segmentation and the like, so that a user can quickly acquire the whole information of the model, the detail information of the model can be represented by the point cloud firstly, and the grid information of the original model can be quickly reconstructed according to the complete vertex information at the later stage.
4. The invention provides a method for dynamically transmitting a 3D model, which combines a point cloud technology and a surface reconstruction technology. The method only needs to transmit the vertexes of the model, which greatly reduces the amount of data transmitted. By combining the technologies of feature point extraction, surface reconstruction, grid segmentation and the like, a user can quickly acquire the whole topology information of the model, and the accuracy and the detail degree of the model are continuously improved along with the increase of the number of the received vertexes.
Drawings
Fig. 1 is a flowchart of a method for dynamically transferring a 3D model according to the present invention.
Fig. 2 is a schematic diagram of an application scenario of the method shown in fig. 1.
FIG. 3 is a schematic diagram of the original model and the segmented modules.
Fig. 4 is a model diagram of a model reconstructed from feature points and composed by adding vertices of the body trunk module.
Fig. 5 is a graph of the point cloud of feature points and the change of adding vertices of each segment module one by one.
FIG. 6 is a model of the feature point reconstruction and a model map reconstructed by adding vertices of each segmentation module one by one.
Fig. 7 is a comparison of the original model and the simplified model.
Detailed Description
The present invention will be described in further detail with reference to examples and drawings, but embodiments of the present invention are not limited thereto.
As shown in fig. 1 and 2, the method for dynamically transmitting a 3D model according to the present embodiment includes the following steps:
1) The server loads a triangle mesh model, as shown in the left side of fig. 3, which is a armadillo model, and the model comprises 69191 vertexes and 138378 triangle patches;
2) The server detects and transmits the feature points, and the method comprises the following steps:
2.1 Scale of computational grid
The scale of the grid is calculated using a local filter with uniform weights. For vertices with scale sIs +.>Can be calculated by the following formula:
2.2 Calculating local density
2.3 Calculating a DoG operator
Given two consecutive scales of a mesh model, the DoG function on scale sCalculated from the following formula:
2.4 Computing local scale and feature points of a grid
The local scale of the grid is defined by the following formula:
wherein s is i Is a filter, C is a parameter defined in advance to be applied to all feature descriptors, and is generally taken as 3. Vertex pointThe condition for feature points on scale s is that they are locally maximum on both scale and spatial location. In this example, only feature points of scale 1 are measured, and 7800 is obtained altogetherThe number of vertices is 11.27% of the total number of points.
2.5 The server transmits the feature points to the client;
3) The client receives and reconstructs the characteristic points, and the method comprises the following steps:
after receiving the characteristic points, the client displays the characteristic points, and after receiving all the characteristic points, the client performs surface reconstruction on the characteristic points, wherein the reconstruction algorithm uses the Screen Poisson surface reconstruction, and the steps of the algorithm are as follows:
3.1 Calculating the normal of the undirected point;
3.2 Constructing scale function fitting point cloud data, and adding vertex positions and gradient constraints;
3.3 Discretizing the scale function by using a linear basis function cluster;
3.4 Performing independent commander selection of the scale and calculation of the vertex boundary;
3.5 Constructing an octree grid and performing surface reconstruction by using a moving cube algorithm;
the point cloud after receiving all the feature points is shown in the upper left corner diagram of fig. 5, the point cloud can display the topological shape of the 3D model in total, the model reconstructed by using the scenpoisson surface reconstruction algorithm is shown in the left side of fig. 4, and the topological shape of the whole model can be displayed well.
4) The server divides the model and transmits the vertex of each module;
the 3D model is segmented using a salient-oriented 3D mesh segmentation algorithm, which comprises the steps of:
4.1 Extraction of significant vertices;
4.2 A) grouping of salient vertices;
4.3 Using a core approximation algorithm to obtain all vertices except for the protrusion;
4.4 Calculating a partition boundary for each salient point;
4.5 Dividing the mesh model according to the partition boundary.
After armadillo model segmentation, 7 modules are arranged in total: left and right hands, left and right feet, head, body trunk and head, as shown on the right side of fig. 3. After the segmentation is finished, the vertexes of each module are transmitted one by one, or the boundaries of each module can be transmitted to a user first, and the user can send information to the server according to preference to arrange the transmission sequence of each module.
5) The client receives the module vertex and performs local reconstruction;
the local region reconstruction is characterized as follows:
5.1 The vertex of each module is received for reconstruction;
5.2 A reconstructed region is limited to the region of the transferred module;
5.3 Using the local region reconstructed region to replace the corresponding region reconstructed by the feature points;
the vertex of the module is received by the client, and then the vertex is displayed in the form of point cloud, and the local reconstruction is performed after the vertex is received, and the right side of fig. 4 is a graph of the armadiform model after the characteristic points are utilized to perform table reconstruction after the vertex of the body trunk module is received. And taking the top partially reconstructed region according to the boundary information of the segmentation algorithm, and only reconstructing the region of the currently received module, wherein the subimage with the subscript of 2 in fig. 6 shows a graph after the armadillo model receives the vertex of the body trunk module and performs partial reconstruction. The overall diagram of fig. 6 demonstrates the process of local surface reconstruction after receiving each module vertex, with each part of the model being progressively more accurate and refined.
6) Simplification of the model
After the vertex transfer of all the modules is completed, the number of the vertexes and the number of the patches reconstructed by using the Screen poisson are more than those of the original models, and the number of the vertexes and the number of the patches are reduced to the number of the original models by using a Qnlim algorithm according to the data of the original models. In the example, the number of vertexes reconstructed by the armadical model is 146089, which is 2.11 times of that of the original model; the number of patches was 292174, which was 2.11 times that of the original model. The model was reduced to a vertex number 73045, a patch number 146086, using the QSlim algorithm, which was very close to the original model. Comparison of the simplified model and the original model fig. 7 shows the original model on the left and the reconstructed model on the right. The simplified model can lose a certain precision, and whether the model is simplified or not can be determined by self according to the requirements.
In summary, after the scheme is adopted, the invention provides a new method for transmitting the 3D model, and an effective means is provided for real-time transmission and rendering of the 3D model by using the combination of feature point detection, grid segmentation and reconstruction algorithms, so that the method not only reduces the transmitted data quantity, but also provides an effective way for real-time rendering display, and has practical popularization and application values.
The above examples are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above examples, and any other changes, modifications, substitutions, combinations, and simplifications that do not depart from the spirit and principle of the present invention should be made in the equivalent manner, and the embodiments are included in the protection scope of the present invention.
Claims (6)
1. A method for dynamically transmitting a 3D model, comprising the steps of:
1) The server loads the 3D model;
2) The method comprises the steps that a server detects characteristic points of a 3D model and transmits the characteristic points to a client; the server detects the characteristic points of the 3D model, calculates local characteristic points by using a DoG operator, and comprises the following steps:
2.1 Scale of computational grid
The scale of the grid is calculated using a local filter with uniform weights; for vertices with scale sIs +.>Can be calculated by the following formula:
2.2 Calculating local density
2.3 Calculating a DoG operator
Given two consecutive scales of a mesh model, the DoG function on scale sCalculated from the following formula:
2.4 Computing local scale and feature points of a grid
The local scale of the grid is defined by the following formula:
wherein s is i Is a filter, C is a parameter defined in advance to be applied to all feature descriptors; vertex pointThe condition of feature points on the scale s is that they are locally maximum on both scale and spatial location;
3) The client receives the characteristic points and performs surface reconstruction after receiving;
4) The server performs grid segmentation on the 3D model and transmits vertex data according to the modules; the grid segmentation of the 3D model adopts a 3D grid segmentation algorithm facing to prominence, and the algorithm comprises the following steps:
4.1 Extraction of significant vertices;
4.2 A) grouping of salient vertices;
4.3 Acquiring all vertices except the salient portion using a core approximation algorithm;
4.4 Calculating a partition boundary for each salient point;
4.5 Dividing the grid model according to the partition boundary;
after the model is divided, a plurality of modules are arranged, the vertexes of the modules are randomly transmitted one by one, or the vertex information of the modules is transmitted according to the specific sequence and the requirements of a client;
5) The client receives the vertex of the module and rebuilds the local area;
6) And after all the modules are transmitted and rebuilt, grid simplification is carried out by using a QLIm algorithm.
2. The method of dynamically transferring a 3D model according to claim 1, wherein in step 1), the 3D model is a triangular mesh whose vertices are represented by the values of (X, Y, Z) coordinates and the triangular patches are represented by the IDs (V1, V2, V3) of the three vertices.
3. The method of dynamically transferring a 3D model according to claim 2, wherein the format of the triangle mesh comprises PLY, STL.
4. The method for dynamically transmitting 3D models according to claim 1, wherein step 3) after receiving the feature points, the client renders and displays the feature points, and after receiving all the feature points, the client reconstructs the surface of the feature points, and the reconstruction algorithm uses the screen poisson surface reconstruction, and the algorithm comprises the following steps:
3.1 Calculating the normal of the undirected point;
3.2 Constructing scale function fitting point cloud data, and adding vertex positions and gradient constraints;
3.3 Discretizing the scale function by using a linear basis function cluster;
3.4 Performing independent commander selection of the scale and calculation of the vertex boundary;
3.5 Constructing an octree grid and performing surface reconstruction using a moving cube algorithm.
5. The method for dynamically transferring 3D models according to claim 1, characterized in that in step 5) the local area reconstruction is characterized as follows:
5.1 The vertex of each module is received for reconstruction;
5.2 A reconstructed region is limited to the region of the transferred module;
5.3 Using the local region reconstructed region to replace the corresponding region reconstructed using the feature points.
6. Method for dynamically transferring 3D models according to claim 1, characterized in that said step 6) is in particular: the server transmits the vertex number and the surface patch information of the original model to the client, and the client uses QSLim algorithm to simplify according to the information, so that the accuracy of the original model is simplified, and the vertex number and the surface patch number of the reconstructed surface are reduced.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911372573.6A CN111105864B (en) | 2019-12-27 | 2019-12-27 | Method for dynamically transmitting 3D model |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911372573.6A CN111105864B (en) | 2019-12-27 | 2019-12-27 | Method for dynamically transmitting 3D model |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111105864A CN111105864A (en) | 2020-05-05 |
CN111105864B true CN111105864B (en) | 2023-06-16 |
Family
ID=70424843
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911372573.6A Active CN111105864B (en) | 2019-12-27 | 2019-12-27 | Method for dynamically transmitting 3D model |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111105864B (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103632371A (en) * | 2013-12-06 | 2014-03-12 | 河海大学常州校区 | Compatibility mesh segmentation based skeleton parameter computation method |
CN106997605A (en) * | 2017-03-22 | 2017-08-01 | 浙江大学 | It is a kind of that the method that foot type video and sensing data obtain three-dimensional foot type is gathered by smart mobile phone |
CN107818578A (en) * | 2017-10-11 | 2018-03-20 | 上海北冕信息科技有限公司 | A kind of fast human-face model algorithm for reconstructing and system based on register method |
CN109472802A (en) * | 2018-11-26 | 2019-03-15 | 东南大学 | A kind of surface grid model construction method constrained certainly based on edge feature |
-
2019
- 2019-12-27 CN CN201911372573.6A patent/CN111105864B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103632371A (en) * | 2013-12-06 | 2014-03-12 | 河海大学常州校区 | Compatibility mesh segmentation based skeleton parameter computation method |
CN106997605A (en) * | 2017-03-22 | 2017-08-01 | 浙江大学 | It is a kind of that the method that foot type video and sensing data obtain three-dimensional foot type is gathered by smart mobile phone |
CN107818578A (en) * | 2017-10-11 | 2018-03-20 | 上海北冕信息科技有限公司 | A kind of fast human-face model algorithm for reconstructing and system based on register method |
CN109472802A (en) * | 2018-11-26 | 2019-03-15 | 东南大学 | A kind of surface grid model construction method constrained certainly based on edge feature |
Also Published As
Publication number | Publication date |
---|---|
CN111105864A (en) | 2020-05-05 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2020192354A1 (en) | Blended urban design scene simulation method and system | |
CN101149842B (en) | Depth detail level adjustment of multi-dimensional image data with a client/server-based image rendering system | |
CN103021017B (en) | Three-dimensional scene rebuilding method based on GPU acceleration | |
CN106600683B (en) | A kind of adaptive method for reconstructing of grid model towards Bone CT sequence image | |
EP3278252B1 (en) | Virtual three-dimensional model generation based on virtual hexahedron models | |
CN106780736A (en) | Map data processing method and device, three-dimensional map generation method and device | |
CN114329747B (en) | Virtual-real entity coordinate mapping method and system for building digital twins | |
CN110503718B (en) | Three-dimensional engineering model lightweight display method | |
CN111862292A (en) | Data rendering method and device for power transmission line corridor and computer equipment | |
CN104376596A (en) | Method for modeling and registering three-dimensional scene structures on basis of single image | |
CN107886564A (en) | The method shown for realizing three-dimensional scenic | |
US8149237B2 (en) | Information processing apparatus and program | |
CN103077559B (en) | Based on the fruit ear three-dimensional rebuilding method of sequence image | |
CN111028335B (en) | Point cloud data block surface patch reconstruction method based on deep learning | |
CN105303616A (en) | Embossment modeling method based on single photograph | |
US20240212863A1 (en) | Unified anisotropic volume and surface mesh storage | |
CN111105864B (en) | Method for dynamically transmitting 3D model | |
CN109509259A (en) | A kind of reconstruction of medical images contour surface grid-search method method | |
Scholz et al. | Real‐time isosurface extraction with view‐dependent level of detail and applications | |
CN112687007A (en) | LOD technology-based stereo grid map generation method | |
CN111898276A (en) | Overhead line model processing method and device, computer equipment and medium | |
CN114332411A (en) | Method for generating three-dimensional graph real-time grid | |
CN109191556B (en) | Method for extracting rasterized digital elevation model from LOD paging surface texture model | |
CN113888633B (en) | Three-dimensional reconstruction method and device based on real-time positioning and mapping algorithm | |
Guo et al. | Multi-thread block terrain dynamic scheduling based on three-dimensional array and Sudoku |
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 |