CN114283245A - Rendering method based on three-dimensional model hierarchical implicit field - Google Patents
Rendering method based on three-dimensional model hierarchical implicit field Download PDFInfo
- Publication number
- CN114283245A CN114283245A CN202210207907.XA CN202210207907A CN114283245A CN 114283245 A CN114283245 A CN 114283245A CN 202210207907 A CN202210207907 A CN 202210207907A CN 114283245 A CN114283245 A CN 114283245A
- Authority
- CN
- China
- Prior art keywords
- dimensional object
- ray
- octree
- node
- dimensional
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Landscapes
- Image Generation (AREA)
Abstract
The invention relates to a rendering method based on a three-dimensional model hierarchical implicit field, which is characterized by comprising the following steps: s1, constructing an octree based on source data of the three-dimensional object, and calculating existence and geometric complexity of the surface of the three-dimensional object in a space represented by octree nodes; when the three-dimensional object surface exists in the space represented by the octree node and the geometric complexity of the surface is greater than a preset value, the node is divided into sub-nodes until the geometric complexity is less than or equal to the preset value; s2, processing the three-dimensional object by adopting a ray stepping algorithm, and taking the intersection of the ray in the ray stepping algorithm and the octree node on the surface of the three-dimensional object as an initial sampling point on the ray; s3, calculating the gradient of the intersection position of the light ray and the surface of the three-dimensional object in the light ray stepping algorithm to obtain the normal direction of the intersection position; and S4, adding the intersection position of the ray and the surface of the three-dimensional object and the normal direction of the intersection position into a classical illumination model to render the three-dimensional object. The method is suitable for the field of three-dimensional model rendering.
Description
Technical Field
The invention relates to a rendering method based on a three-dimensional model hierarchical implicit field. The method is suitable for the field of three-dimensional model rendering.
Background
The research of efficient three-dimensional model representation and rendering methods is an important task of computer graphics. The implicit function representation has the characteristics of flexibly representing different topologies and representing models without the limitation of resolution, and can be widely applied to the representation of various shapes. Although some methods for representing an implicit function by combining with a neural network can arbitrarily approximate any three-dimensional shape without being limited by a function base, the problem is not well solved by the existing method how to efficiently draw the corresponding picture under the given camera parameter from the represented three-dimensional shape.
Existing methods of rendering implicitly represented three-dimensional geometries can be divided into indirect rendering methods and direct rendering methods.
The indirect rendering method comprises the steps of firstly extracting an isosurface of a three-dimensional shape which is implicitly represented, namely the surface of a model, by a marching cube algorithm/dual contouring algorithm and other methods, obtaining an explicit three-dimensional grid of the model, and then rendering the extracted three-dimensional geometric shape surface by a three-dimensional grid rendering method. The method has the advantages of high parallelization execution, easy and full utilization of the existing hardware pipeline based on triangle drawing design, and low efficiency. If complex geometric details need to be accurately captured, the three-dimensional mesh extraction algorithm needs to be executed at a very high resolution, the finally extracted three-dimensional mesh is also composed of a large number of triangular patches, and the rendering calculation cost is greatly increased. Furthermore, introducing a three-dimensional mesh as an intermediary inevitably increases the error of the rendering result.
The direct rendering method can directly draw a final two-dimensional picture from an implicit representation without converting the implicit function into an explicit representation. Direct rendering is usually based on a ray tracing algorithm, a typical example being the sphere tracing algorithm. The method directly queries the SDF values of sampling points in space of rays emitted from the viewpoint and judges whether the rays continue to travel or reach the surface of the model or not based on the SDF values. The method can directly render the picture without the help of explicit representation, and the efficiency is improved. However, the light often passes through a large blank space during the traveling process, still generating many unnecessary SDF queries, and considering the SDF value at a certain point of each query usually requires global computation of the whole implicit function describing the three-dimensional shape, and thus the efficiency is still low.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: aiming at the existing problems, a rendering method based on a three-dimensional model hierarchical implicit field is provided.
The technical scheme adopted by the invention is as follows: a rendering method based on a three-dimensional model hierarchical implicit field is characterized by comprising the following steps:
s1, constructing an octree based on source data of the three-dimensional object, and calculating existence and geometric complexity of the surface of the three-dimensional object in a space represented by octree nodes;
when a three-dimensional object surface exists in a space represented by an octree node and the geometric complexity of the surface is greater than a preset value, the node is divided into sub-nodes until the geometric complexity of the three-dimensional object surface in each octree node with the three-dimensional object surface is less than or equal to the preset value;
s2, processing the three-dimensional object by adopting a ray stepping algorithm, and taking the intersection of the ray in the ray stepping algorithm and the octree node on the surface of the three-dimensional object as an initial sampling point on the ray;
s3, calculating the gradient of the intersection position of the light ray and the surface of the three-dimensional object in the light ray stepping algorithm to obtain the normal direction of the intersection position;
and S4, adding the intersection position of the ray and the surface of the three-dimensional object and the normal direction of the intersection position into a classical illumination model to render the three-dimensional object.
The geometric complexity of the three-dimensional object surface is based on the amount of variation of the three-dimensional object surface normal.
The step S2 includes:
s2-1, emitting a ray from a viewpoint to each pixel point of a screen to generate a group of rays, wherein the viewpoint position is X, and the emitted ray is X + tv;
s2-2, traversing octree nodes which are intersected with the ray and have structural characteristics representing the existence of the surface of the three-dimensional object from the root node in sequence according to the sequence of the ray passing through the octree nodes;
if the traversed node is a leaf node of an octree, calculating the intersection of the ray and the node (specifically, a linear equation of the ray is introduced into a plane of an Axis Alignment Bounding Box (AABB) forming the node, and the intersection of the ray and the node is obtained by solving), and setting the intersection as an initial sampling point;
if the octree node which is intersected with the ray and the structural feature of which indicates that the surface of the three-dimensional object exists is not traversed, the color of the pixel point corresponding to the ray is marked as the background, and the ray processing is finished;
s2-3, setting the current sampling point as XkQuerying to obtain the SDF valued=f(Xk) If the SDF value isdExceeds a threshold valueεThe sampling point continues to move forward along the ray by step tk= dIs mixing Xk+1=Xk+tk vSetting a new sampling point; if it is notdIf the size is small enough, namely sufficiently close to 0, the ray is considered to intersect with the surface of the shape, and the ray processing is finished;
s2-4, setting the current sampling point as XkIf at this time sample point XkHaving traversed the current octree node, X is detectedkWhether a three-dimensional object surface exists on a node where the current position is located or not;
if a surface is present, then XkSetting the node as the current octree node, and returning to the step S2-3; otherwise, return to step S2-2.
Step S3 includes:
normal to the surface of the three-dimensional object isThe opposite direction of the gradient direction of the SDF, the position where the ray intersects the surface of the three-dimensional object obtained in step S2 is X + tn vAt which position the gradient is calculatedy=∇f(X + tn v) To obtain the normal directionn=-y/||y||。
The ray stepping algorithm sets a maximum stepping length.
The source data of the three-dimensional object is a three-dimensional grid, a three-dimensional surface represented by a mathematical function or an equation, and a three-dimensional shape constructed based on a CSG tree.
The rendering method is suitable for representing and rendering a large scene.
A rendering device based on a three-dimensional model hierarchical implicit field is characterized by comprising:
the octree construction module is used for constructing an octree based on the source data of the three-dimensional object and calculating the existence and the geometric complexity of the surface of the three-dimensional object in the space represented by the octree nodes; when a three-dimensional object surface exists in a space represented by an octree node and the geometric complexity of the surface is greater than a preset value, the node is divided into sub-nodes until the geometric complexity of the three-dimensional object surface in each octree node with the three-dimensional object surface is less than or equal to the preset value;
the algorithm calculation module is used for processing the three-dimensional object by adopting a ray stepping algorithm, and taking the intersection of the ray in the ray stepping algorithm and the octree node on the surface of the three-dimensional object as an initial sampling point on the ray;
the normal direction calculation module is used for calculating the gradient of the intersection position of the light ray and the surface of the three-dimensional object in the light ray stepping algorithm to obtain the normal direction of the intersection position;
and the model rendering module is used for adding the intersection position of the light ray and the surface of the three-dimensional object and the normal direction of the intersection position into the classical illumination model to render the three-dimensional object.
A storage medium having stored thereon a computer program executable by a processor, the computer program comprising: the computer program, when executed, implements the steps of the three-dimensional model based hierarchical implicit field rendering method.
An electronic device having a memory and a processor, the memory having stored thereon a computer program executable by the processor, the electronic device comprising: the computer program, when executed, implements the steps of the three-dimensional model based hierarchical implicit field rendering method.
The invention has the beneficial effects that: the method obtains the structural characteristics of the octree by calculating the existence and the geometric complexity of the surface of the three-dimensional object in the space represented by the octree nodes, segments the nodes when the structural characteristics indicate that the surface exists and the geometric complexity does not meet the preset requirement until the geometric complexity of the inner surface of each octree node with the surface meets the preset requirement, and can well utilize the space sparsity of a complex large-scale scene by adaptively adjusting the octree nodes to realize efficient implicit expression rendering.
The value overhead of globally computing implicit functions is very high, so the SDF value is computed only in the local area of the query point (within the nodes where the surface exists) in the rendering process proposed by the present invention. And (3) based on the ray intersection detection of the geometric shape existence characteristics of the neural network codes in the sparse octree nodes, rays sent by the viewpoint skip the blank area without the geometric shape, and the ray intersection detection process of the hierarchical neural implicit function is accelerated.
In the invention, the local implicit function in the octree node performs sphere tracking to perform intersection detection, and the position of an intersection surface is quickly found by traversing points on a ray path in a large step.
According to the method, the maximum stepping length of the sphere tracking in the light stepping algorithm is set, so that the phenomenon that the light advancing step length is too large to penetrate through the surface to obtain an error rendering result due to the fitting error of the neural network at a position far away from the surface, namely a position with a large SDF value is prevented.
The method calculates the normal direction near the surface according to the gradient direction of the neural implicit function to obtain the normal direction of the surface of the model, and then adds the normal direction into the illumination model.
Drawings
FIG. 1 is a flow chart of an embodiment.
FIG. 2 is a flow diagram of single ray processing in an embodiment.
Fig. 3 is a schematic diagram of a rendering principle of the embodiment.
Detailed Description
As shown in fig. 1, the embodiment is a rendering method based on a three-dimensional model hierarchical implicit field, and is particularly suitable for representing and rendering a large scene, and specifically includes the following steps:
and S0, acquiring source data corresponding to the three-dimensional object. The source data of the three-dimensional object is data related to a three-dimensional object shape structure, and may be a three-dimensional mesh, a three-dimensional surface represented by a mathematical function or an equation, or a three-dimensional shape constructed based on a CSG tree. In computer graphics, a three-dimensional grid is a basic representation mode, and grid data can be obtained by gridding other three-dimensional data.
S1, constructing an octree based on source data of the three-dimensional object, calculating existence and geometric complexity (determined based on the variation of the surface normal of the three-dimensional object) of the surface of the three-dimensional object in the space represented by the current octree node after the octree root node corresponding to the three-dimensional object is segmented for multiple times, obtaining the structural characteristics of the node, and continuing to segment the node when the structural characteristics of the node do not meet preset requirements until the octree nodes meet the preset requirements.
And when the structural characteristics of the octree nodes indicate that the three-dimensional object surface exists in the space represented by the nodes and the geometric complexity of the surface is greater than a preset value, the nodes are cut into sub-nodes, and the structural characteristics of the sub-nodes are recalculated until the geometric complexity of the three-dimensional object surface in each octree node with the three-dimensional object surface is less than or equal to the preset value.
And S2, as shown in FIG. 2, processing the three-dimensional object by adopting a ray stepping algorithm, and taking the intersection of the ray in the ray stepping algorithm and the octree node on the surface of the three-dimensional object as an initial sampling point on the ray.
S2-1, emitting a ray from the viewpoint to each pixel point of the screen to generate a group of rays, wherein the viewpoint position is X, and the emitted ray is X + tn v；
S2-2, according to the ray X + tn vThe order of traversing octree nodes is in turn from the root nodeThe method comprises the following steps of (1) traversing octree nodes which are intersected with rays and have structural features representing the existence of the surface of a three-dimensional object;
if the traversed node is a leaf node of the octree, the plane of the AABB forming the node is substituted into a linear equation of the ray, the intersection of the ray and the node is obtained by solution, and the intersection is set as an initial sampling point
If the octree node which is intersected with the ray and the structural feature of which indicates that the surface of the three-dimensional object exists is not detected (no octree node exists in the front or no three-dimensional object surface exists in the octree node in the front), the color of the pixel point corresponding to the ray is marked as the background, and the ray processing is finished;
s2-3, setting the current sampling point as XkQuerying to obtain the SDF valued=f(Xk) If the SDF value isdExceeds a threshold valueε（d >ε) The sampling point continues to move forward along the ray by step tk= dIs mixing Xk+1=Xk+tk vSetting a new sampling point; if it is notdSufficiently small, i.e. sufficiently close to 0: (d <ε) If the shape surface is intersected with the ray, the ray processing is finished;
s2-4, setting the current sampling point as XkIf at this time sample point XkHaving traversed the current octree node (indicating that the ray does not intersect the three-dimensional corresponding surface within the current node), X is detectedkWhether a three-dimensional object surface exists on a node where the current position is located or not;
if a surface is present, then XkSetting the node as a new current octree node, and returning to the step S2-3; otherwise, return to step S2-2.
The above steps are repeated until the intersection with the surface or no octree nodes ahead. When intersecting the surface, note tn=t0+ t1+…+tn-1At this time, the sampling point position Xn= X + tnv, i.e. the position where the ray intersects the surface (see fig. 3, the dotted line segment does not set the sampling point, and the circle radius is SDF).
And S3, calculating the gradient of the intersection position of the ray and the surface of the three-dimensional object in the ray stepping algorithm, and obtaining the normal direction of the intersection position.
For a closed surface, the normal to the surface is the opposite direction to the gradient direction of the SDF. In step S2, the position X + t at which the ray intersects with the shape is obtainedn vAt which position the gradient is calculatedy=∇f(X + tn v) To obtain the normal directionn=-y/ ||y||。
And S4, adding the intersection position of the ray and the surface of the three-dimensional object and the normal direction of the intersection position into a classical illumination model to render the three-dimensional object.
In order to prevent the light from traveling too long through the surface due to the fitting error of the neural network at a position far from the surface, i.e., at a position with a large SDF value, and obtaining a wrong rendering result, the embodiment sets the maximum step length for the sphere tracking for the light stepping algorithm.
The embodiment also provides a rendering device based on a three-dimensional model hierarchical implicit field, which includes: the system comprises an octree construction module, an algorithm calculation module, a normal calculation module and a model rendering module.
In the embodiment, the octree construction module is used for constructing an octree based on source data of the three-dimensional object, and calculating the existence and the geometric complexity of the surface of the three-dimensional object in the space represented by the octree nodes; and when the three-dimensional object surface exists in the space represented by the octree node and the geometric complexity of the surface is greater than a preset value, the node is cut into sub-nodes until the geometric complexity of the three-dimensional object surface in each octree node with the three-dimensional object surface is less than or equal to the preset value.
The algorithm calculation module is used for processing the three-dimensional object by adopting a ray stepping algorithm, and the intersection of the ray in the ray stepping algorithm and the octree node on the surface of the three-dimensional object is used as an initial sampling point on the ray. And the normal direction calculation module is used for calculating the gradient of the intersection position of the ray and the surface of the three-dimensional object in the ray stepping algorithm to obtain the normal direction of the intersection position. And the model rendering module is used for adding the intersection position of the light ray and the surface of the three-dimensional object and the normal direction of the intersection position into the classical illumination model to render the three-dimensional object.
The present embodiment also provides a storage medium having stored thereon a computer program executable by a processor, the computer program, when executed, implementing the steps of the rendering method for layering an implicit field based on a three-dimensional model in this example.
The present embodiment also provides an electronic device having a memory and a processor, the memory storing thereon a computer program executable by the processor, the computer program, when executed, implementing the steps of the rendering method for layering an implicit field based on a three-dimensional model in this example.
Claims (10)
1. A rendering method based on a three-dimensional model hierarchical implicit field is characterized by comprising the following steps:
s1, constructing an octree based on source data of the three-dimensional object, and calculating existence and geometric complexity of the surface of the three-dimensional object in a space represented by octree nodes;
when a three-dimensional object surface exists in a space represented by an octree node and the geometric complexity of the surface is greater than a preset value, the node is divided into sub-nodes until the geometric complexity of the three-dimensional object surface in each octree node with the three-dimensional object surface is less than or equal to the preset value;
s2, processing the three-dimensional object by adopting a ray stepping algorithm, and taking the intersection of the ray in the ray stepping algorithm and the octree node on the surface of the three-dimensional object as an initial sampling point on the ray;
s3, calculating the gradient of the intersection position of the light ray and the surface of the three-dimensional object in the light ray stepping algorithm to obtain the normal direction of the intersection position;
and S4, adding the intersection position of the ray and the surface of the three-dimensional object and the normal direction of the intersection position into a classical illumination model to render the three-dimensional object.
2. The rendering method based on the three-dimensional model hierarchical implicit field according to claim 1, wherein: the geometric complexity of the three-dimensional object surface is based on the amount of variation of the three-dimensional object surface normal.
3. The rendering method for layering the implicit field based on the three-dimensional model according to claim 1, wherein the step S2 includes:
s2-1, emitting a ray from a viewpoint to each pixel point of a screen to generate a group of rays, wherein the viewpoint position is X, and the emitted ray is X + tv;
s2-2, traversing octree nodes which are intersected with the ray and have structural characteristics representing the existence of the surface of the three-dimensional object from the root node in sequence according to the sequence of the ray passing through the octree nodes;
if the currently traversed node is a leaf node of the octree, calculating the intersection of the ray and the node, and setting the intersection as an initial sampling point;
if the octree node which is intersected with the ray and the structural feature of which indicates that the surface of the three-dimensional object exists is not traversed, the color of the pixel point corresponding to the ray is marked as the background, and the ray processing is finished;
s2-3, setting the current sampling point as XkQuerying to obtain the SDF valued=f(Xk) If the SDF value isdExceeds a threshold valueεThe sampling point continues to move forward along the ray by step tk=d, mixing Xk+1=Xk+tk vSetting a new sampling point; if it is notdIf the size is small enough, namely sufficiently close to 0, the ray is considered to intersect with the surface of the shape, and the ray processing is finished;
s2-4, setting the current sampling point as XkIf at this time sample point XkHaving traversed the current octree node, X is detectedkWhether a three-dimensional object surface exists on a node where the current position is located or not;
if a surface is present, then XkSetting the node as the current octree node, and returning to the step S2-3; otherwise, return to step S2-2.
4. The rendering method for layering the implicit field based on the three-dimensional model according to claim 1, wherein the step S3 includes:
the normal direction of the three-dimensional object surface is opposite to the gradient direction of the SDF, and the intersection position of the ray and the three-dimensional object surface obtained in step S2 is X + tnv, calculating the gradient y = ∇ f (X + t) at this positionnv) to get the normal n = -y/| | y | |.
5. The rendering method based on the three-dimensional model hierarchical implicit field according to claim 1 or 3, characterized in that: the ray stepping algorithm sets a maximum stepping length.
6. The rendering method based on the three-dimensional model hierarchical implicit field according to claim 1, wherein: the source data of the three-dimensional object is a three-dimensional grid, a three-dimensional surface represented by a mathematical function or an equation, and a three-dimensional shape constructed based on a CSG tree.
7. The rendering method based on the three-dimensional model hierarchical implicit field according to claim 1, wherein: the rendering method is suitable for representing and rendering a large scene.
8. A rendering device based on a three-dimensional model hierarchical implicit field is characterized by comprising:
the octree construction module is used for constructing an octree based on the source data of the three-dimensional object and calculating the existence and the geometric complexity of the surface of the three-dimensional object in the space represented by the octree nodes; when a three-dimensional object surface exists in a space represented by an octree node and the geometric complexity of the surface is greater than a preset value, the node is divided into sub-nodes until the geometric complexity of the three-dimensional object surface in each octree node with the three-dimensional object surface is less than or equal to the preset value;
the algorithm calculation module is used for processing the three-dimensional object by adopting a ray stepping algorithm, and taking the intersection of the ray in the ray stepping algorithm and the octree node on the surface of the three-dimensional object as an initial sampling point on the ray;
the normal direction calculation module is used for calculating the gradient of the intersection position of the light ray and the surface of the three-dimensional object in the light ray stepping algorithm to obtain the normal direction of the intersection position;
and the model rendering module is used for adding the intersection position of the light ray and the surface of the three-dimensional object and the normal direction of the intersection position into the classical illumination model to render the three-dimensional object.
9. A storage medium having stored thereon a computer program executable by a processor, the computer program comprising: the computer program when executed implements the steps of the method for rendering based on a three-dimensional model hierarchical implicit field according to any one of claims 1 to 7.
10. An electronic device having a memory and a processor, the memory having stored thereon a computer program executable by the processor, the electronic device comprising: the computer program when executed implements the steps of the method for rendering based on a three-dimensional model hierarchical implicit field according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210207907.XA CN114283245B (en) | 2022-03-04 | 2022-03-04 | Rendering method based on three-dimensional model hierarchical implicit field |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210207907.XA CN114283245B (en) | 2022-03-04 | 2022-03-04 | Rendering method based on three-dimensional model hierarchical implicit field |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114283245A true CN114283245A (en) | 2022-04-05 |
| CN114283245B CN114283245B (en) | 2022-06-14 |
Family
ID=80882240
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210207907.XA Active CN114283245B (en) | 2022-03-04 | 2022-03-04 | Rendering method based on three-dimensional model hierarchical implicit field |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114283245B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050286748A1 (en) * | 2004-06-25 | 2005-12-29 | Lining Yang | System and method for fast generation of high-quality maximum/minimum intensity projections |
| US20050285858A1 (en) * | 2004-06-25 | 2005-12-29 | Siemens Medical Solutions Usa, Inc. | System and method for fast volume rendering |
| CN101458823A (en) * | 2008-12-19 | 2009-06-17 | 北京航空航天大学 | Real-time lighting drawing method under virtual stage environment |
| CN101826215A (en) * | 2010-04-19 | 2010-09-08 | 浙江大学 | Real-time secondary ray tracing concurrent rendering method |
| US20170323471A1 (en) * | 2016-05-06 | 2017-11-09 | National Taiwan University | 3D rendering method and 3D graphics processing device |
| CN107993279A (en) * | 2017-11-03 | 2018-05-04 | 中国人民解放军陆军工程大学 | A kind of scene accurate expression method |
| CN110476188A (en) * | 2017-03-30 | 2019-11-19 | 奇跃公司 | Centralization rendering |
| CN113643416A (en) * | 2021-08-13 | 2021-11-12 | 重庆中烟工业有限责任公司 | Three-dimensional image volume rendering method, three-dimensional image volume rendering device, and computer-readable storage medium |
-
2022
- 2022-03-04 CN CN202210207907.XA patent/CN114283245B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050286748A1 (en) * | 2004-06-25 | 2005-12-29 | Lining Yang | System and method for fast generation of high-quality maximum/minimum intensity projections |
| US20050285858A1 (en) * | 2004-06-25 | 2005-12-29 | Siemens Medical Solutions Usa, Inc. | System and method for fast volume rendering |
| CN101458823A (en) * | 2008-12-19 | 2009-06-17 | 北京航空航天大学 | Real-time lighting drawing method under virtual stage environment |
| CN101826215A (en) * | 2010-04-19 | 2010-09-08 | 浙江大学 | Real-time secondary ray tracing concurrent rendering method |
| US20170323471A1 (en) * | 2016-05-06 | 2017-11-09 | National Taiwan University | 3D rendering method and 3D graphics processing device |
| CN110476188A (en) * | 2017-03-30 | 2019-11-19 | 奇跃公司 | Centralization rendering |
| CN107993279A (en) * | 2017-11-03 | 2018-05-04 | 中国人民解放军陆军工程大学 | A kind of scene accurate expression method |
| CN113643416A (en) * | 2021-08-13 | 2021-11-12 | 重庆中烟工业有限责任公司 | Three-dimensional image volume rendering method, three-dimensional image volume rendering device, and computer-readable storage medium |
Non-Patent Citations (5)
| Title |
|---|
| ÖMER CENGIZ CELEBI,ET AL.: "Accelerating volume rendering by ray leaping with back steps", 《COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE》 * |
| SZYMON JABŁONSKI,ET AL.: "Real-time visualization of Dynamic Unlimited Objects Instancing", 《COMPUTER SCIENCE RESEARCH NOTES》 * |
| 张文胜 等: "基于八叉树邻域分析的光线跟踪加速算法", 《图学学报》 * |
| 张梓良: "肺结节医学图像体绘制技术研究", 《中国优秀博硕士学位论文全文数据库(硕士) 信息科技辑》 * |
| 赵烨梓 等: "基于机器学习的三维场景高度真实感绘制方法综述", 《软件学报》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114283245B (en) | 2022-06-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Ost et al. | Neural scene graphs for dynamic scenes | |
| Vespa et al. | Efficient octree-based volumetric SLAM supporting signed-distance and occupancy mapping | |
| US9721320B2 (en) | Fully parallel in-place construction of 3D acceleration structures and bounding volume hierarchies in a graphics processing unit | |
| US8014568B2 (en) | Method for computer-aided identification of the child octants of a parent octant, which are intersected by a beam, in an octree data structure by means of look-up tables | |
| US9396512B2 (en) | Fully parallel construction of k-d trees, octrees, and quadtrees in a graphics processing unit | |
| Zeng et al. | Pc-nbv: A point cloud based deep network for efficient next best view planning | |
| Kuo et al. | Patch2cad: Patchwise embedding learning for in-the-wild shape retrieval from a single image | |
| CN104050710A (en) | 3-d graphics rendering with implicit geometry | |
| Li et al. | kANN on the GPU with shifted sorting | |
| CN113052955A (en) | Point cloud completion method, system and application | |
| Dong et al. | ASH: A modern framework for parallel spatial hashing in 3D perception | |
| Sulzer et al. | Scalable Surface Reconstruction with Delaunay‐Graph Neural Networks | |
| Zarzar et al. | Segnerf: 3d part segmentation with neural radiance fields | |
| Hafiz et al. | Se-md: A single-encoder multiple-decoder deep network for point cloud generation from 2d images | |
| Kuhn et al. | Incremental division of very large point clouds for scalable 3d surface reconstruction | |
| CN114283245B (en) | Rendering method based on three-dimensional model hierarchical implicit field | |
| Shi et al. | Accurate implicit neural mapping with more compact representation in large-scale scenes using ranging data | |
| Lin et al. | A-SATMVSNet: An attention-aware multi-view stereo matching network based on satellite imagery | |
| Zhang et al. | Exploring semantic information extraction from different data forms in 3D point cloud semantic segmentation | |
| Cao et al. | Stable image matching for 3D reconstruction in outdoor | |
| Mateo et al. | Hierarchical, Dense and Dynamic 3D Reconstruction Based on VDB Data Structure for Robotic Manipulation Tasks | |
| CN106652032B (en) | A kind of parallel contour lines creation method of DEM based on Linux cluster platform | |
| Du et al. | Vipnet: A fast and accurate single-view volumetric reconstruction by learning sparse implicit point guidance | |
| Xiao et al. | Instance-aware monocular 3D semantic scene completion | |
| Zhang et al. | Dense Reconstruction from Visual SLAM with Probabilistic Multi-Sequence Merging |
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 |