CN114758073A - Oral cavity digital system based on RGBD input and flexible registration - Google Patents
Oral cavity digital system based on RGBD input and flexible registration Download PDFInfo
- Publication number
- CN114758073A CN114758073A CN202210406432.7A CN202210406432A CN114758073A CN 114758073 A CN114758073 A CN 114758073A CN 202210406432 A CN202210406432 A CN 202210406432A CN 114758073 A CN114758073 A CN 114758073A
- Authority
- CN
- China
- Prior art keywords
- point cloud
- input
- rgbd
- vertexes
- edges
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Computer Graphics (AREA)
- Geometry (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Dental Tools And Instruments Or Auxiliary Dental Instruments (AREA)
Abstract
The invention relates to an oral cavity digital system based on RGBD input and flexible registration, and belongs to the technical field of oral cavity medical treatment digitization. The RGBD image is used as input, and a non-rigid scene reconstruction algorithm based on a surface detail migration technology is used for reconstructing a rigid part, a non-rigid part and a semi-rigid part in the oral cavity of a person and reconstructing albedo textures of the surface. The technical scheme of the invention is further improved as follows: in the process, firstly, the texture information or the high-dimensional features extracted from the texture information are given to the point cloud, and then the point cloud is segmented to extract an interesting region; in the point cloud registration part, different smooth constraint coefficients are given to different parts by a solving algorithm according to actual physical attributes of the different parts. The method mainly makes full use of the input texture and depth information, improves the modeling precision and the texture truth, reduces the scanning reconstruction time, and improves the use comfort and convenience.
Description
Technical Field
The invention relates to an oral cavity digital system based on RGBD input and flexible registration, and belongs to the technical field of oral cavity medical treatment digitization.
Background
Taking a dental model is an important way for diagnosis and curative effect confirmation in oral medical treatment, the dental model is an important information medium of the oral environment of a patient, and particularly in tooth shape correction and false tooth manufacturing, a doctor usually needs to manufacture a plurality of pairs of dental models in different periods for diagnosis and planning or observing the treatment effect for the patient, so that the pertinence of a diagnosis and treatment scheme is improved.
Clinically, the three-dimensional data acquisition of the internal teeth of the oral cavity is divided into two modes of extraoral scanning and intraoral scanning, the extraoral scanning is to scan the plaster model of the dentition of a patient by adopting scanning equipment to acquire digital three-dimension, the intraoral scanning is to directly scan and measure the tooth body and related soft and hard tissues by extending the scanning equipment into the mouth of the patient, the digital three-dimension is acquired in real time, and compared with the extraoral scanning mode, the intraoral scanning is truly unmodeled and digitized, is convenient and efficient, and has obvious advantages.
The main problems of the conventional intraoral scanning are that the requirement on the precision of scanning equipment is high, a powder spraying mode on teeth before scanning is adopted, the scanning equipment is large in size, difficult to operate in a limited intraoral environment, uncomfortable experience is brought to patients, the equipment cost is high, and great pressure is brought to the popularization of equipment purchasers and users, so that the scanning equipment is insufficient in the experience of the scanning process or the competitiveness facing the market.
Disclosure of Invention
The invention aims to provide an oral cavity digital system based on RGBD input and flexible registration.
In order to achieve the purpose, the invention adopts the technical scheme that:
an oral cavity digital system based on RGBD input and flexible registration adopts RGBD images as input, and uses a non-rigid scene reconstruction algorithm based on a surface detail migration technology to reconstruct albedo textures of a surface while reconstructing a rigid part, a non-rigid part and a semi-rigid part in an oral cavity of a person.
The technical scheme of the invention is further improved as follows: in the process, firstly, the texture information or the high-dimensional features extracted from the texture information are given to the point cloud, and then the point cloud is segmented to extract an interesting region; in the point cloud registration part, different smooth constraint coefficients are given to different parts by a solving algorithm according to actual physical attributes of the different parts.
The technical scheme of the invention is further improved as follows: in order to achieve the overall accuracy of the nearest neighbor tracking algorithm of adjacent frames, all vertexes are adopted as optimization targets, all vertexes are directly optimized, and a new computing idea of two mesh nearest neighbor points is used.
The technical scheme of the invention is further improved as follows: the specific processing steps are as follows,
1) RGB graph processing
Extracting texture information or high-dimensional information corresponding to the point cloud by using the input RGB image, and meanwhile, performing illumination removal treatment on the RGB image, removing the influence of a light source and extracting the texture of the oral cavity area;
2) point cloud segmentation
Combining the RGB extracted information and the point cloud three-dimensional coordinates, and segmenting on the point cloud to extract an interested region;
3) point cloud preprocessing
The point cloud processing part is combined with the result of the previous step, a depth neural network is used for segmenting and smoothing the depth map, flying spots and less noise are removed, and meanwhile details are kept as much as possible; the network adopts a reverse depth map as one of input data, wherein the reverse depth map is the reciprocal of the depth of the point, which is actually stored by the numerical value in the map; the network structure is an Image To Image network comprising a structure not limited To a Hourglass network, and the input channel can also contain aligned RGB color pictures so as To utilize the prior knowledge in the RGB color pictures;
4) unified non-rigid registration of point clouds
The non-rigid registration method based on dense point cloud is adopted, and the method is implemented as follows,
firstly, supposing that a part of reconstructed scene point cloud is Vtgt, and a newly added frame of point cloud is marked as Vsrc, then the Vsrc can be converted to a degree similar to the Vtgt through non-rigid transformation, a triangle is quickly established by each adjacent three points in the point cloud, and a normal vector of a certain point of the triangle and the directions from the point to other two points form a local space system;
After the initial close point is determined through the nearest neighbor, the two mesh can be drawn, wherein all optimization variables are all vertexes, and a key control point does not need to be extracted from the vertexes to carry out sparse DQB (double quaternion interpolation) control, and interpolation operation is not needed.
5) Surface reconstruction
Restoring a surface from the point cloud using methods including, but not limited to, cedar reconstruction and marching cube;
6) geometrical post-processing
The method comprises the steps of removing repeated flying spots, filling holes, removing non-manifold surfaces and processing flying spot surfaces.
5. The RGBD-based input and flexible registration oral digitizer system according to claim 4, wherein: the completion hole is processed as follows,
firstly, searching the most marginal area of a geometric body according to a half-edge structure, finding N edges of the edges, sequencing according to the number of vertexes of the edges, triangularizing the area if the area needs to be increased uniformly by considering other smaller edges as holes except the largest edge of each mesh module, then carrying out the triangularization process including but not limited to a Delou method, and smoothing after the triangularization is finished;
the de-duplication flying spot process is as follows,
deleting unreferenced vertexes and repeated vertexes, deleting repeated triangular faces, separating grids according to communication areas to obtain a plurality of different blocks, then sequencing according to the surface areas or the number of the vertexes of the blocks to find an upper maximum communication area and a lower maximum communication area, and completely deleting other small-area communication areas;
The non-manifold surface removal process is as follows,
since the surface geometry of the object that is actually present should not have a non-manifold, the geometry needs to be processed in a non-manifold manner: firstly, according to the repetition degree of the edges, if the repetition degree is greater than 2, the edges with the repetition degree greater than 2 are considered to be non-manifold areas, all the edges are deleted from the middle of the triangular mesh after being selected, then all the communication areas are recalculated, the smaller areas are deleted according to the communication areas, the deleted edges are reconnected, and thus the non-manifold processing is completed.
The technical scheme of the invention is further improved as follows: in step 4, three important penalty terms required in the iterative optimization process are as follows,
1. smoothness penalty for affine transformation of source triangle proximity points
This penalty ensures that the transformation of neighboring points is relatively smooth:
ES(v1...vn)＝sum(||Ti-Tj||).
v1 … vn is a vertex position in the source point cloud and is a variable, Ti and Tj are affine transformation relative to an original coordinate system after a local coordinate system formed by triangles in the source point cloud is modified, and ij is the number of an adjacent triangle set;
2. affine change minimization penalty for source triangles
To ensure that the source triangle changes to the B state with minimal changes, we take the following approach:
EI(v1...vn)＝sum(||Ti-I||).
Where I stands for a 3 x 3 identity matrix, it means that the penalty term constrains all affine transformations to be as equal as possible to the identity matrix, i.e. no transformation.
3. Minimizing penalty for source vertex position and target vertex position
This penalty term is denoted as:
Ec(v1...vn)＝sum(||vi-ci||).
where vi represents the vertex position in the source point cloud and ci represents the position of the point in the target point cloud that is closest to the vi spatial position and whose normal vector is fairly consistent.
Due to the adoption of the technical scheme, the invention has the following technical effects:
the method mainly makes full use of the input texture and depth information, improves the modeling precision and the texture truth, reduces the scanning reconstruction time, and improves the use comfort and convenience.
The invention adopts all vertexes as optimization targets, directly optimizes all vertexes, and provides a new computing idea of two mesh nearest points, thereby greatly improving the computing stability and the reconstruction accuracy of the algorithm.
The invention can improve the stability and speed of oral reconstruction, improve the accuracy of oral reconstruction and improve the physical rationality of oral reconstruction.
Drawings
FIG. 1 is a schematic block diagram of the overall process of the present invention;
FIG. 2 is a diagram of the nearest neighbor finding principle;
FIG. 3 is the cypress pine reconstruction principle;
FIG. 4 is a schematic diagram of curvature-based smoothing;
FIG. 5 is a non-manifold mesh body;
FIG. 6 is a partial spatial system.
Detailed Description
In order to make the technical means, the creation characteristics, the achievement purposes and the effects of the invention easy to understand, the invention is further described with the specific embodiments.
The method adopts an RGBWRGBD image as input, and reconstructs albedo texture of the surface while reconstructing a rigid part, a non-rigid part and a semi-rigid part in the oral cavity of a person by using a non-rigid scene reconstruction algorithm based on a surface detail migration technology.
According to the method, after the aligned RGBDRGBD image is acquired, point cloud is given with information provided by RGBR, and then the point cloud is segmented to extract semantic information. And performing non-rigid registration on the point clouds, and fusing multi-frame input point clouds. And finally reconstructing an exit cavity mesh from the fused point cloud. And acquiring albedo texture information by using the acquired RGBLGB information to finally obtain the color three-dimensional model of the oral cavity.
In the process, firstly, the texture information or the high-dimensional features extracted from the texture information are given to the point cloud, and then the point cloud is segmented to extract an interesting region, for example, the point cloud is divided into tongue/tooth/gum and other parts; in the point cloud registration part, different smooth constraint coefficients are given to different parts by a solving algorithm according to actual physical attributes of the different parts, and the following rules are usually followed: tongue > interface > gingiva > teeth. Under the framework, in order to improve the overall accuracy of the nearest tracking algorithm of adjacent frames, the idea of extracting sparse key points is abandoned, all vertexes are used as optimization targets, all vertexes are directly optimized, a new calculation idea of two mesh nearest points is provided, and the calculation stability and the reconstruction accuracy of the algorithm can be greatly improved.
As shown in detail in figure 1.
The following are specific examples
1. RGB graph processing
The texture information or high-dimensional information corresponding to the point cloud is extracted by using the input RGBRGB image (for example, the RGBRGB image is subjected to a semantic segmentation network, but the final result is not obtained, and the high-dimensional information after the backsbone is obtained and is given to the point cloud, so that the influence caused by the error of the semantic segmentation network can be reduced). Meanwhile, the RGBRGB image is subjected to illumination removing treatment, the influence of a light source is removed, and the texture of the oral cavity area is extracted.
2. Point cloud segmentation
Combining the information extracted by RGBRG and the three-dimensional coordinates of point cloud, segmenting on the point cloud to extract an interested area, and dividing the oral area into teeth, gums, tongues and the like.
3. Pre-processing of point clouds
And in the point cloud processing part, the depth neural network is used for segmenting and smoothing the depth map by combining the result of the last step, flying spots and less noise are removed, and details are kept as much as possible. The network adopts a reverse depth map as one of input data, the reverse depth map is an inverse number of the depth of a point, the numerical value in the map is actually stored, the network structure is an ImageToImage network with a Hourglass network structure, in addition, an input channel can also contain aligned RGB color pictures, and the priori knowledge in the RGB color map is utilized, for example, the places with stronger environment light masks at the teeth gaps and the flying spots of the mirror reflection points of teeth and gum are more, and the noise is larger.
4. Unified non-rigid registration of point clouds
In the traditional oral reconstruction, only gingiva and teeth can be well processed, because the two parts can easily meet rigid constraints, the front and rear frame point cloud input can be registered only by carrying out a plurality of iterations of a rigid nearest neighbor algorithm, a non-rigid registration method based on dense point cloud is adopted in the scheme, and the method is specifically implemented as follows:
firstly, a part of reconstructed scene point cloud is assumed to be VtgtAnd a newly added frame point cloud is marked as VsrcThen V issrcCan be converted to V by non-rigid transformationtgtComparing the similarity, we quickly establish a triangle with each adjacent three points in the point cloud (the method for constructing a triangle for two point clouds may adopt the classical Delaunay triangle method), and the normal vector of a certain point of the triangle and the directions from the point to other two points form a local spatial system, as shown in fig. 1:
where v4 is calculated as follows:
v4＝v1+(v2-v1)×(v3-v1)sqrt(|(v2-v1)×(v3-v1)|)
the representation of the local coordinate system can be written as:
V＝[v2-v1 v3-v1 v4-v1]
defining a spatial radial transformation Q and translation d, which is applied to the four vertices, results in:
Q*vi+d＝v’i
the transformed coordinate system is:
V’＝[v’2-v’1 v’3-v’1 v’4-v’1].
from the above, it can be known that the affine change between the local coordinate systems can be determined and solved by the coordinates of the three points of the triangle before and after the non-rigid change:
Q＝V’V-1
After determining the initial proximity point through the nearest neighbor, the algorithm may zoom in the two grids, where all optimization variables are all vertices, it is not necessary to extract key control points from the vertices to perform sparse DQB control, and it is not necessary to perform interpolation operation, as shown in fig. 2, the following are three important penalty terms required in the iterative optimization process:
one, smoothness penalty of affine transformation of source triangle proximity point
This penalty ensures that the transformation of neighboring points is relatively smooth:
ES(v1...vn)＝sum(||Ti-Tj||).
v1 … vn is the vertex position in the source point cloud and is a variable, Ti and Tj are affine transformation relative to the original coordinate system after the local coordinate system formed by the triangles in the source point cloud is modified, and ij is the number of the adjacent triangle set.
Second, affine change minimization punishment of source triangle
To ensure that the source triangle changes to the B state with minimal changes, we take the following approach:
EI(v1...vn)＝sum(||Ti-I||).
where I represents a 3 x 3 identity matrix, it means that the penalty term constrains all affine transformations to be as equal as possible to the identity matrix, i.e. no transformation.
Third, the minimum punishment of the source vertex position and the target vertex position
The penalty term is noted as:
Ec(v1...vn)＝sum(||vi-ci||).
where vi represents the vertex position in the source point cloud and ci represents the position of the point in the target point cloud that is closest to the vi spatial position and fairly consistent with the normal vector.
5. Surface reconstruction
The surface may be recovered from the point cloud using methods including, but not limited to, cedar reconstruction, marching cube, and the like. The process is a general flow, is not a main protection point, and only introduces a common method.
Poisson reconstruction (poisson reconstruction) is a very intuitive approach. The core idea is that the point cloud represents the position of the object surface, and the normal vector represents the inside and outside directions. By implicitly fitting an indicator function derived from the object, an estimate of a smooth object surface can be given.
Given an area MM and its boundaries
The indicating function χ M χ M is defined as
Thus, reconstructing
The problem of (2) translates to the problem of reconstructing χ M, as shown in fig. 3.
6. Geometric post-processing
The method comprises the steps of removing repeated flying spots, filling holes, removing non-manifold surfaces and processing flying spot surfaces.
Curvature based smoothing
The curvature-based smoothing is to move the vertex P along a straight line on the normal vector of the vertex P, and as can be seen from fig. 4, the general laplacian smoothing may say that the vertex P moves to a position similar to the center of gravity, while the moving position of the curvature-based smoothing is opposite to the normal vector, which is to better maintain the original approximate shape of the model.
By properly selecting the weights ω in the laplacian smoothing, we can make the curvature-based smoothing and the laplacian smoothing uniform algorithmically. It is derived that when ω is chosen to be a value equivalent to moving P as in fig. 2 above.
The position of the point P is still determined by the following formula, and only the weight calculation mode is modified:
filling holes:
firstly, the most marginal area of a geometric body is searched according to a half-edge structure, N edges are found, the edges are sorted according to the number of vertexes of the edges, except the largest edge, other smaller edges can be regarded as holes, the area needs to be triangulated, points are uniformly added according to the area, then a triangularization process including but not limited to a Delaunay method is carried out, and smoothing is carried out after the triangularization is finished.
And (3) removing repeated flying spots:
deleting unreferenced vertexes and repeated vertexes, deleting repeated triangular faces, separating the grids according to the communication areas to obtain a plurality of different blocks, then sequencing according to the surface areas or the number of vertexes of the blocks to find an upper maximum communication area and a lower maximum communication area, and completely deleting other small-area communication areas.
Non-manifold surface removal:
since the surface geometry of the object that is actually present should not have a non-manifold, the geometry needs to be processed in a non-manifold manner: firstly, according to the repetition degree of the edges, if the repetition degree is greater than 2, the edges with the repetition degree greater than 2 are considered to be non-manifold areas, all the edges are deleted from the middle of the triangular mesh after being selected, then all the communication areas are recalculated, the smaller areas are deleted according to the communication areas, the deleted edges are reconnected, and thus the non-manifold processing is completed. As shown in fig. 5.
The foregoing shows and describes the general principles and broad features of the present invention and advantages thereof. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, which are described in the specification and illustrated only to illustrate the principle of the present invention, but that various changes and modifications may be made therein without departing from the spirit and scope of the present invention, which fall within the scope of the invention as claimed. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (6)
1. An oral digitizer system based on RGBD input and flexible registration, comprising: the RGBD image is used as input, and a non-rigid scene reconstruction algorithm based on a surface detail migration technology is used for reconstructing a rigid part, a non-rigid part and a semi-rigid part in the oral cavity of a person and reconstructing albedo textures of the surface.
2. The RGBD-based input and flexible registration oral digitizer system of claim 1, wherein: in the process, firstly, the texture information or the high-dimensional features extracted from the texture information are given to the point cloud, and then the point cloud is segmented to extract an interesting region; in the point cloud registration part, different smooth constraint coefficients are given to different parts by a solving algorithm according to actual physical attributes of the different parts.
3. The RGBD-based input and flexible registration oral digitizer system of claim 2, wherein: in order to achieve the overall accuracy of the nearest neighbor tracking algorithm of adjacent frames, all vertexes are adopted as optimization targets, all vertexes are directly optimized, and a new computing idea of two mesh nearest neighbor points is used.
4. The RGBD-based input and flexible registration oral digitizer system according to claim 3, wherein: the specific processing steps are as follows,
1) RGB graph processing
Extracting texture information or high-dimensional information corresponding to the point cloud by using the input RGB image, and meanwhile, performing illumination removal treatment on the RGB image, removing the influence of a light source and extracting the texture of the oral cavity area;
2) point cloud segmentation
Combining RGB extracted information and point cloud three-dimensional coordinates, and segmenting on point clouds to extract an interested region;
3) Point cloud preprocessing
The point cloud processing part is combined with the result of the last step, a depth neural network is used for segmenting and smoothing the depth map, flying spots and less noise are removed, and details are kept as much as possible; the network adopts a reverse depth map as one of input data, wherein the reverse depth map is the reciprocal of the point depth actually stored in the value of the map; the network structure is an ImageToImage network comprising a structure not limited to a Hourglass network, and the input channels can also contain aligned RGB color pictures to utilize prior knowledge in RGB color pictures;
4) unified non-rigid registration of point clouds
The non-rigid registration method based on dense point cloud is adopted, and the method is implemented as follows,
firstly, supposing that a part of reconstructed scene point cloud is Vtgt, and a newly added frame of point cloud is marked as Vsrc, then the Vsrc can be converted to a degree similar to the Vtgt through non-rigid transformation, a triangle is quickly established by each adjacent three points in the point cloud, and a normal vector of a certain point of the triangle and the directions from the point to other two points form a local space system;
after the initial close point is determined through the nearest neighbor, the two meshes can be drawn, wherein all optimization variables are all vertexes, and the key control points do not need to be extracted from the vertexes to carry out sparse DQB control, and interpolation operation is not needed.
5) Surface reconstruction
Restoring a surface from the point cloud using methods including, but not limited to, cedar reconstruction and marching cube;
6) geometrical post-processing
The method comprises the steps of removing repeated flying spots, filling holes, removing non-manifold surfaces and processing flying spot surfaces.
5. The RGBD input and flexible registration based oral digitizer system according to claim 4, wherein: the completion hole is processed as follows,
firstly, searching the most marginal region of a geometric body according to a half-edge structure, finding N edges of the edges, sequencing according to the number of vertexes of the edges, triangularizing the region according to the requirement that points are uniformly added according to the area except the largest edge of each mesh module and other smaller edges can be regarded as holes, and then carrying out the triangularization process including but not limited to a Delaunay method, and smoothing after the triangularization is finished;
the de-duplication flying spot process is as follows,
deleting unreferenced vertexes and repeated vertexes, deleting repeated triangular faces, separating the grids according to the communication areas to obtain a plurality of different blocks, then sequencing according to the surface areas or the number of vertexes of the blocks to find an upper maximum communication area and a lower maximum communication area, and completely deleting other small-area communication areas;
The non-manifold surface removal process is as follows,
since the surface geometry of the object that exists in practice should not have a non-manifold, a non-manifold treatment of the geometry is required: firstly, according to the repetition degree of the edges, if the repetition degree is greater than 2, the edges with the repetition degree greater than 2 are considered to be non-manifold areas, all the edges are deleted from the middle of the triangular mesh after being selected, then all the communication areas are recalculated, the smaller areas are deleted according to the communication areas, the deleted edges are reconnected, and thus the non-manifold processing is completed.
6. The RGBD-based input and flexible registration oral digitizer system according to claim 4, wherein: in step 4, three important penalty terms required in the iterative optimization process are as follows,
1. smoothness penalty for affine transformation of source triangle proximity points
This penalty ensures that the transformation of neighboring points is relatively smooth:
ES(v1...vn)＝sum(||Ti-Tj||).
v1 … vn is a vertex position in the source point cloud and is a variable, Ti and Tj are affine transformation relative to an original coordinate system after a local coordinate system formed by triangles in the source point cloud is modified, and ij is the number of an adjacent triangle set;
2. affine change minimization penalty for source triangles
To ensure that the source triangle changes to the B state with minimal changes, we take the following approach:
EI(v1...vn)＝sum(||Ti-I||).
where I represents a 3 x 3 identity matrix, it means that the penalty term constrains all affine transformations to be as equal as possible to the identity matrix, i.e. no transformation.
3. Minimizing penalty for source vertex position and target vertex position
The penalty term is noted as:
Ec(v1...vn)＝sum(||vi-ci||).
where vi represents the vertex position in the source point cloud and ci represents the position of the point in the target point cloud that is closest to the vi spatial position and fairly consistent with the normal vector.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210406432.7A CN114758073A (en) | 2022-04-18 | 2022-04-18 | Oral cavity digital system based on RGBD input and flexible registration |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210406432.7A CN114758073A (en) | 2022-04-18 | 2022-04-18 | Oral cavity digital system based on RGBD input and flexible registration |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114758073A true CN114758073A (en) | 2022-07-15 |
Family
ID=82331974
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210406432.7A Pending CN114758073A (en) | 2022-04-18 | 2022-04-18 | Oral cavity digital system based on RGBD input and flexible registration |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114758073A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115375754A (en) * | 2022-10-21 | 2022-11-22 | 中信梧桐港供应链管理有限公司 | Storage yard volume detection method and device |
-
2022
- 2022-04-18 CN CN202210406432.7A patent/CN114758073A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115375754A (en) * | 2022-10-21 | 2022-11-22 | 中信梧桐港供应链管理有限公司 | Storage yard volume detection method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11735306B2 (en) | Method, system and computer readable storage media for creating three-dimensional dental restorations from two dimensional sketches | |
| Yeh et al. | Template-based 3d model fitting using dual-domain relaxation | |
| Tagliasacchi et al. | 3d skeletons: A state‐of‐the‐art report | |
| US9317970B2 (en) | Coupled reconstruction of hair and skin | |
| US7450749B2 (en) | Image processing method for interacting with a 3-D surface represented in a 3-D image | |
| US10346504B2 (en) | 3D modelling of bodies | |
| CN110610478B (en) | Medical image three-dimensional reconstruction method based on neighborhood topology | |
| CN115619773B (en) | Three-dimensional tooth multi-mode data registration method and system | |
| Liu et al. | High-quality textured 3D shape reconstruction with cascaded fully convolutional networks | |
| CN109118455B (en) | Ancient human skull craniofacial interactive restoration method based on modern soft tissue distribution | |
| Arakeri et al. | An effective and efficient approach to 3D reconstruction and quantification of brain tumor on magnetic resonance images | |
| CN114549540A (en) | Method for automatically fusing oral scanning tooth data and CBCT (Cone Beam computed tomography) data and application thereof | |
| Ben-Hamadou et al. | 3DTeethSeg'22: 3D Teeth Scan Segmentation and Labeling Challenge | |
| CN114758073A (en) | Oral cavity digital system based on RGBD input and flexible registration | |
| CN104899923A (en) | Method for constructing facial prosthesis optimized model based on smile expression geometrical characteristic modification | |
| Tian et al. | Efficient tooth gingival margin line reconstruction via adversarial learning | |
| Zhang et al. | An Implicit Parametric Morphable Dental Model | |
| CN111862176B (en) | Three-dimensional oral cavity point cloud orthodontic front and back accurate registration method based on palatine fold | |
| Goshtasby | Three-dimensional model construction from multiview range images: survey with new results | |
| CN115082640A (en) | Single image-based 3D face model texture reconstruction method and equipment | |
| Li et al. | ImTooth: Neural Implicit Tooth for Dental Augmented Reality | |
| Huang et al. | Three-dimensional reconstruction and 3D printing of kidney from computed tomography | |
| CN116385474B (en) | Tooth scanning model segmentation method and device based on deep learning and electronic equipment | |
| de Araujo Buck et al. | 3-D segmentation of medical structures by integration of raycasting with anatomic knowledge | |
| Khaleel et al. | A Review paper of 3D Surface Reconstruction of Coronary Arteries From Cardiovascular Angiography |
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 |