An edge line generation method, device, apparatus and product

Abstract

The application provides an edge line generation method, device, equipment and product. The method comprises the following steps: obtaining a local mesh model of a target tooth based on a first three-dimensional mesh model of a mouth, the mouth comprising a plurality of teeth; performing multi-angle sectioning on the local mesh model to obtain a plurality of two-dimensional section views, performing edge point identification on the plurality of two-dimensional section views to obtain a plurality of initial edge points; taking each initial edge point as a center, and intercepting a corresponding sub-region mesh from the local mesh model; reducing the dimension of each sub-region mesh to render a two-dimensional image, performing edge point identification on the two-dimensional image to obtain an edge point set and a region type of each sub-region mesh; determining the weight of each edge point set according to the region type, and obtaining an edge line of the target tooth based on the edge point set and the weight of the edge point set. The technical scheme solves the defects that the shoulder features are lost due to excessive smoothing or sawteeth are generated due to insufficient smoothing in the traditional identification and fitting method, and improves the accuracy of edge line generation.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of oral restoration technology, and in particular to a method, apparatus, device and product for generating edge lines. Background Technology

[0002] In the field of dental prosthodontics, the introduction of digital technology has brought about profound changes. Among them, dental CAD / CAM (Computer Aided Design and Computer Aided Manufacturing) prosthodontic technology has become a research hotspot due to its advantages such as fast prosthodontic fabrication speed and high prosthetic precision. In the dental CAD / CAM prosthodontic process, the extraction of the edge lines of the prepared tooth is a key step in achieving a tight seal at the dental margins.

[0003] Currently, edge lines are typically extracted by performing global curvature analysis or point cloud semantic segmentation on the 3D mesh data of the preparatory volume. Curvature-based methods identify candidate points by detecting abrupt changes in curvature and then perform a global uniform fitting to generate edge lines; point cloud segmentation-based methods use deep learning models to segment the preparatory volume region, extract the boundaries, and then also use a global uniform fitting to generate edge lines.

[0004] Existing technologies use a globally unified fitting strategy to extract edge lines, which can easily lead to the retention of scanning burrs and jagged edges in the gentle buccal region, and excessive smoothing of the large curvature transition area and narrow adjacent segments on the lingual side, resulting in the loss of subtle steps and chamfer details on the shoulder, thus distorting the shape of the finally extracted edge lines. Summary of the Invention

[0005] This invention provides an edge line generation method, apparatus, device, and product to address the shortcomings of traditional recognition and fitting methods that lose shoulder features due to over-smoothing or produce jagged edges due to under-smoothing, thereby improving the accuracy of edge line generation.

[0006] According to one aspect of the present invention, an edge line generation method is provided, comprising: Based on the first three-dimensional mesh model of the oral cavity, a local mesh model of the target tooth is obtained, wherein the oral cavity contains multiple teeth; The local mesh model is sectioned from multiple angles to obtain multiple two-dimensional cross-sectional images. Edge points are identified on the multiple two-dimensional cross-sectional images to obtain multiple initial edge points. Centered on each of the initial edge points, a corresponding sub-region mesh is extracted from the local mesh model; Each of the sub-region grids is rendered into a two-dimensional image by dimensionality reduction, and edge point recognition is performed on the two-dimensional image to obtain the edge point set and region type of each sub-region grid. The weights of each edge point set are determined according to the region type, and the edge line of the target tooth is obtained based on the edge point set and its weight.

[0007] According to another aspect of the present invention, an edge line generating apparatus is provided, the apparatus comprising: The local mesh model acquisition module is used to acquire a local mesh model of a target tooth based on a first three-dimensional mesh model of the oral cavity, wherein the oral cavity contains multiple teeth; The initial edge point acquisition module is used to perform multi-angle sectioning on the local mesh model to obtain multiple two-dimensional cross-sectional images, and to identify edge points on the multiple two-dimensional cross-sectional images to obtain multiple initial edge points; The sub-region mesh extraction module is used to extract the corresponding sub-region mesh from the local mesh model, centered on each of the initial edge points; The edge point set acquisition module is used to render each of the sub-region grids into a two-dimensional image by reducing its dimension, and to perform edge point recognition on the two-dimensional image to obtain the edge point set and region type of each of the sub-region grids. An edge line acquisition module is used to determine the weight of each edge point set according to the region type, and obtain the edge line of the target tooth based on the edge point set and the weight of the edge point set.

[0008] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the edge line generation method according to any embodiment of the present invention.

[0009] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the edge line generation method described in any embodiment of the present invention.

[0010] According to another aspect of the present invention, a computer program product is provided, which, when executed by a processor, implements the edge line generation method as described in any of the embodiments of the present invention.

[0011] This invention provides an embodiment of a method for obtaining a local mesh model of a target tooth based on a first three-dimensional mesh model of the oral cavity, which contains multiple teeth. Obtaining the local mesh allows subsequent sectioning and recognition to focus on the target region, effectively eliminating interference from irrelevant structures such as adjacent teeth. The local mesh model is sectioned from multiple angles to obtain multiple two-dimensional cross-sectional images. Edge point recognition is performed on these multiple two-dimensional cross-sectional images to obtain multiple initial edge points. The three-dimensional edge recognition problem is transformed into an edge point detection problem involving multiple two-dimensional cross-sectional images, reducing the problem complexity. Furthermore, locking the initial edge points provides reliable initial points for subsequent fine-grained localization. A corresponding sub-region mesh is extracted from the local mesh model, centered on each initial edge point. Extracting sub-region meshes centered on the initial edge points narrows the search range and filters out interference. Even if the initial edge points have deviations, the true edge lines still fall within the sub-region mesh, allowing for precise location through subsequent fine-grained recognition. Each sub-region mesh is then rendered into a two-dimensional image, and edge point recognition is performed on the two-dimensional image to obtain the edge point set and region type of each sub-region mesh. Edge point recognition is performed on the dimensionality-reduced 2D image, enabling edge point recognition to be performed on a 2D plane, significantly reducing computational complexity. The weights of each edge point set are determined based on the region type, and the edge line of the target tooth is obtained based on the edge point sets and their weights. Differential weighted fitting preserves the angular features of the shoulder region while ensuring smoothness in the gingival sulcus region, overcoming the shortcomings of global uniform fitting which cannot accommodate the needs of different anatomical regions. In summary, the technical solution of this embodiment obtains a 2D cross-sectional view by multi-angle sectioning of a first 3D mesh model, obtains initial edge points from the 2D cross-sectional view, obtains sub-region meshes based on the initial edge points, converts the sub-region meshes into a 2D image, performs edge point recognition on the 2D image to obtain the edge point sets and region types of the sub-region meshes, determines the weights of each edge point set based on the region type, and obtains the edge line of the target tooth based on the edge point sets and their weights. This solves the defects of traditional recognition and fitting methods, such as loss of shoulder features due to over-smoothing or jagged edges due to under-smoothing, and improves the accuracy of edge line generation.

[0012] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0013] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 This is a flowchart of an edge line generation method according to an embodiment of the present invention; Figure 2 This is a flowchart of an edge line generation process in an embodiment of the present invention; Figure 3 This is a schematic diagram of a first three-dimensional mesh model in an embodiment of the present invention; Figure 4 This is a schematic diagram of an initial edge point recognition embodiment of the present invention; Figure 5 This is a schematic diagram of an edge point set recognition embodiment of the present invention; Figure 6 This is a schematic diagram of the structure of an edge line generation device according to an embodiment of the present invention; Figure 7 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0015] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0016] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0017] It is understood that before using the technical solutions disclosed in the various embodiments of the present invention, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in the present invention and their authorization should be obtained in accordance with relevant laws and regulations through appropriate means.

[0018] Figure 1 This is a flowchart illustrating an edge line generation method provided in an embodiment of the present invention. This embodiment is applicable to extracting tooth edge lines based on a three-dimensional mesh model of the oral cavity. The method can be executed by the edge line generation device in this embodiment, which can be implemented in software and / or hardware. This device can be integrated into electronic devices such as computer equipment, servers, mobile terminals, or processors. Figure 1 As shown, the method specifically includes the following steps: S110. Based on the first three-dimensional mesh model of the oral cavity, obtain the local mesh model of the target tooth, wherein the oral cavity contains multiple teeth.

[0019] In this embodiment of the invention, the first three-dimensional mesh model can be specifically understood as a three-dimensional digital model representing the oral and maxillofacial region. The first three-dimensional mesh model contains surface morphological data of the complete dentition. The target tooth can be specifically understood as the tooth whose edge lines need to be extracted. Edge line extraction of the target tooth is typically used for single crown restoration. The target tooth has undergone tooth preparation. The target tooth can be pre-defined by the user. The local mesh model can be specifically understood as a local three-dimensional mesh model of the area containing the target tooth, extracted from the first three-dimensional mesh model. The size of the local mesh model is smaller than that of the first three-dimensional mesh model. The local mesh model only contains the target tooth and its surrounding local area.

[0020] Specifically, based on the first 3D mesh model of the oral cavity and the location of the target tooth, a local mesh model of the target tooth is extracted from the first 3D mesh model. The oral cavity contains multiple teeth. Extracting a local mesh model confines all subsequent operations, such as sectioning, rendering, and recognition, to the area near the target tooth, avoiding interference from irrelevant structures and significantly reducing the computational scale of subsequent steps.

[0021] Optionally, the process of acquiring the first three-dimensional mesh model includes: acquiring a second three-dimensional mesh model, which is constructed from the oral cavity acquisition image; acquiring segmentation information of the second three-dimensional mesh model, and acquiring the major axis direction of each tooth based on the segmentation information; the segmentation information is used to identify each tooth in the second three-dimensional mesh model; performing plane fitting according to the major axis direction of each tooth, and taking the normal direction of the fitted plane as the target direction; adjusting the orientation of the second three-dimensional mesh model according to the target direction to obtain the first three-dimensional mesh model, wherein the occlusal surface of the first three-dimensional mesh model faces the target direction.

[0022] In this embodiment of the invention, the second three-dimensional mesh model can be specifically understood as an original three-dimensional mesh model constructed based on images acquired from the oral cavity. The orientation of the second three-dimensional mesh model has not been standardized, meaning the orientation of its occlusal surface is random and may deviate from the preset target orientation. Segmentation information can be specifically understood as information used to identify each tooth in the second three-dimensional mesh model. Based on the segmentation information, different teeth can be distinguished from a second three-dimensional mesh model containing multiple teeth, thereby locating the target tooth and calculating the long axis direction of each tooth. The long axis direction of a tooth can be specifically understood as the direction of the main axis extending from the crown to the root of the tooth. In the second three-dimensional mesh model, the long axis direction of the tooth is calculated using data from each tooth in the segmentation information. The target orientation can be specifically understood as the orientation fitted based on the long axis direction of each tooth in the oral cavity. The target orientation is used to adjust the second three-dimensional mesh model to the first three-dimensional mesh model.

[0023] Specifically, a second 3D mesh model is obtained. This second 3D mesh model is constructed from images acquired in the oral cavity. Segmentation information of the second 3D mesh model is obtained. This segmentation information can be obtained through manual annotation or segmentation algorithms; this invention does not limit this. The major axis direction of each tooth is obtained based on the segmentation information. Plane fitting is performed based on the major axis direction of each tooth, and the normal direction of the fitted plane is used as the target direction. For example, least squares is used to perform plane fitting based on the major axis direction of each tooth, and the normal direction of the fitted plane is used as the target direction. The second 3D mesh model is oriented according to the target direction, with the occlusal surface of the second 3D mesh model facing the target direction, resulting in a first 3D mesh model. By fitting the target direction based on the tooth's major axis direction and obtaining the first 3D mesh model, the relative spatial relationship between each cut surface and the target tooth remains consistent during subsequent multi-angle sectioning, thereby ensuring the consistency of feature extraction and the reliability of model recognition.

[0024] It should be noted that after obtaining the second 3D mesh model, it is determined whether the interlocking surface of the second 3D mesh model is facing the target direction. If the interlocking surface of the second 3D mesh model is facing the target direction, the second 3D mesh model is used as the first 3D mesh model; if the interlocking surface of the second 3D mesh model is not facing the target direction, the orientation of the second 3D mesh model is adjusted to obtain the first 3D mesh model. By determining the orientation of the interlocking surface, adjustments are only made when the mesh is not properly aligned, thus avoiding the repeated acquisition of the first 3D mesh model.

[0025] Optionally, obtaining a local mesh model of the target tooth based on the first three-dimensional mesh model of the oral cavity includes: obtaining the target tooth based on the segmentation information; and extracting a local mesh model of the target tooth from the first three-dimensional mesh model with the center of the target tooth as the center of the sphere and a first preset length as the radius.

[0026] In this embodiment of the invention, the first preset length can be specifically understood as the length used when performing local mesh cutting with the center of the target tooth as the sphere center. The value of the first preset length should ensure that the cut local mesh model completely covers the entire edge line area of ​​the target tooth, while minimizing interference from adjacent teeth and gingival tissue. For example, the first preset length can be 5 mm.

[0027] Specifically, based on the segmentation information, the location of the target tooth is obtained. A local mesh model of the target tooth is obtained by cropping from the first 3D mesh model with the center of the target tooth as the center and a first preset length as the radius. Cropping the local mesh model with the center of the target tooth as the center limits subsequent operations to the area near the target tooth, effectively eliminating interference from irrelevant structures such as adjacent teeth and the gum line.

[0028] S120. Perform multi-angle sectioning on the local mesh model to obtain multiple two-dimensional cross-sectional views, and perform edge point identification on the multiple two-dimensional cross-sectional views to obtain multiple initial edge points.

[0029] In this embodiment of the invention, the two-dimensional cross-sectional view can be specifically understood as a two-dimensional image obtained by sectional projection of a local mesh model. The two-dimensional cross-sectional view shows the cross-sectional contour of the target tooth at a specific angle. The two-dimensional cross-sectional view is used for edge point identification to obtain initial edge point positions. The initial edge point can be specifically understood as the point obtained after edge point identification on the two-dimensional cross-sectional view. Since this identification process is performed on the global two-dimensional cross-sectional view, the edge point identification is relatively limited, and there is a certain deviation between the obtained edge point position and the actual edge line position, therefore it is used as the initial edge point.

[0030] Specifically, the local mesh model is sectioned from multiple angles to obtain multiple two-dimensional cross-sectional images. A deep learning model can be used to identify edge points on these multiple cross-sectional images, resulting in multiple initial edge points. For example, the deep learning model could be the YOLO model. By sectioning from multiple angles, the positions of the edge lines in all directions are sampled, providing initial reference points for subsequent fine-tuning and curve fitting.

[0031] Optionally, the step of performing multi-angle sectioning on the local mesh model to obtain multiple two-dimensional cross-sectional views includes: rotating the local mesh model to multiple different angles with a preset coordinate axis as the rotation axis and a preset angle step; and projecting each rotated local mesh model onto the preset plane as the cutting plane to obtain multiple two-dimensional cross-sectional views.

[0032] In this embodiment of the invention, the preset coordinate axis can be specifically understood as the rotation axis used when rotating the local mesh model. The preset coordinate axis is used to control the spatial orientation of each cutting plane during multi-angle sectioning. By rotating the local mesh model around the preset coordinate axis at preset angle steps, local mesh models at different angles can be obtained. For example, the preset coordinate axis is the Y-axis in a coordinate system established with the center of the target tooth as the origin. The preset angle step can be specifically understood as the fixed angle of each rotation when rotating the local mesh model. The preset angle step is used to control the angular interval between each cutting plane during multi-angle sectioning. The preset angle step is preset and can be adjusted according to actual application needs; this invention does not limit this. For example, the preset angle step can be 45 degrees.

[0033] Specifically, using a preset coordinate axis as the rotation axis, the local mesh model is rotated to multiple different angles with a preset angle step, resulting in multiple local mesh models at different angles. Using a preset plane as the cutting plane, the local mesh model is cut along the preset plane, and the cut portion on one side is projected onto the cutting plane to obtain two-dimensional cross-sectional views corresponding to each rotation angle. For example, the preset coordinate axis is the Y-axis in a coordinate system established with the center of the target tooth as the origin, and the preset angle step is 45 degrees. Using the Y-axis as the rotation axis, the model is rotated counterclockwise by 45 degrees each time from the initial position, resulting in 8 local mesh models. For each local mesh model, it is cut along the YOZ plane. The portion of the local mesh model located to the left of the X-axis (i.e., x < 0) is projected onto the YOZ plane, resulting in 8 two-dimensional cross-sectional views. The YOZ plane is a plane in a coordinate system established with the center of the target tooth as the origin. Multi-angle sectioning transforms the problem of edge line recognition in three-dimensional space into a problem on multiple two-dimensional cross-sectional views, reducing the recognition difficulty; at the same time, multiple angles ensure that the position of the edge line in each direction is sampled, providing initial edge points in the full circumference for subsequent edge line fitting.

[0034] S130. Using each of the initial edge points as the center, extract the corresponding sub-region mesh from the local mesh model.

[0035] In this embodiment of the invention, the sub-region mesh can be specifically understood as a mesh extracted from the local mesh model, centered on each initial edge point. Each sub-region mesh includes the initial edge point and its vicinity. The size of the sub-region mesh is smaller than that of the local mesh model. The number of sub-region meshes is the same as the number of initial edge points.

[0036] Specifically, multiple sub-region grids are extracted from the local grid model, centered on each initial edge point. By extracting sub-region grids centered on the initial edge points, the scope is reduced from the entire target tooth to sub-region grids, filtering out long-distance interference and providing a foundation for subsequent fine recognition.

[0037] Optionally, the step of extracting the corresponding sub-region mesh from the local mesh model with each initial edge point as the center includes: extracting from the local mesh model with each initial edge point as the center and a second preset length as the radius to obtain the sub-region mesh corresponding to each initial edge point.

[0038] In this embodiment of the invention, the second preset length can be specifically understood as the radius used when truncating a sub-region grid centered on each initial edge point. The second preset length is used to control the truncated range of the sub-region grid. The value of the second preset length should ensure that the sub-region grid can cover the initial edge point and its vicinity, while ensuring that the actual edge line falls within the range of the sub-region grid. For example, the second preset length is 2mm.

[0039] Specifically, using each initial edge point as the center and a second preset length as the radius, a sub-region mesh corresponding to each initial edge point is obtained by cropping from the local mesh model. For example, using each initial edge point as the center and a radius of 2mm, a sub-region mesh corresponding to each initial edge point is obtained by cropping from the local mesh model. Cropping the sub-region mesh with the initial edge point as the center and a second preset length as the radius reduces the scope from the entire target tooth to the sub-region mesh, filtering out long-distance interference and providing a foundation for subsequent fine recognition.

[0040] S140. Render each of the sub-region grids into a two-dimensional image by reducing its dimension, and perform edge point recognition on the two-dimensional image to obtain the edge point set and region type of each sub-region grid.

[0041] In this embodiment of the invention, the two-dimensional image can be specifically understood as an image obtained by dimensionality reduction rendering of a sub-region grid. In the two-dimensional image, different color regions correspond to different surface orientations, and the color change is most drastic at the edge lines, which facilitates edge point recognition and region type classification. An edge point set can be specifically understood as a set of edge points obtained after edge point recognition of the two-dimensional image after dimensionality reduction rendering of the sub-region grid. Each edge point set corresponds to one sub-region grid. A region type can be specifically understood as the type of the sub-region grid. The region type is output synchronously during edge point recognition of the two-dimensional image and is used to identify the anatomical region category to which the sub-region grid belongs.

[0042] Specifically, each sub-region grid is rendered as a 2D image by reducing its dimensionality. Reducing the 3D sub-region grid to a 2D image allows edge point recognition to be performed on a 2D plane, significantly reducing computational complexity. A deep learning model is then used to perform edge point recognition on each 2D image, obtaining the edge point set and region type of each sub-region grid, providing a classification basis for subsequent differential weighted fitting. For example, the deep learning model could be the YOLO model.

[0043] Optionally, the step of dimensionality reduction rendering of each sub-region mesh into a two-dimensional image includes: calculating the normal direction of each vertex in the sub-region mesh; determining the color value corresponding to each vertex based on the components of the normal direction of each vertex in each coordinate axis direction; unfolding the sub-region mesh to a two-dimensional plane to obtain the mapped coordinates of each vertex on the two-dimensional plane; and rendering the two-dimensional image corresponding to the sub-region mesh based on the mapped coordinates of each vertex and the corresponding color value.

[0044] In this embodiment of the invention, the mapped coordinates can be specifically understood as the position coordinates of each vertex in the sub-region mesh after being mapped from three-dimensional space to a two-dimensional plane. The mapped coordinates are used to determine the pixel position of each vertex in the two-dimensional image. The normal direction of a vertex can be specifically understood as the orientation of the surface at each vertex in the sub-region mesh. Here, the normal direction of a vertex is a unit vector in three-dimensional space, represented by three components: x, y, and z, corresponding to the orientation of the normal direction along the X-axis, Y-axis, and Z-axis, respectively.

[0045] Specifically, the normal direction of each vertex in the sub-region grid is calculated. The x, y, and z components of the normal direction of each vertex are then mapped to the red, green, and blue channel values ​​of the RGB (Red, Green, Blue) color space, respectively. Since the normal direction is a normalized unit vector, its components range from [-1, 1], while the image color channels range from [0, 1]. Before mapping, the components of the normal direction need to be normalized to map the components of each coordinate axis to the range [0, 1]. The normalization formula is as follows: Where V represents the component of the normal direction along a certain coordinate axis; T represents the color channel value corresponding to this component after mapping. Thus, the x, y, and z components of the vertex normal direction are mapped to red, green, and blue channel values, respectively, and each vertex obtains a corresponding RGB color value. Color encoding transforms the geometric orientation of the surface into visual features, making color abrupt changes at edges easier for the model to capture. The sub-region mesh is unfolded onto a two-dimensional plane to obtain the mapped coordinates of each vertex on the two-dimensional plane. For example, a harmonic unfolding algorithm can be used, i.e., parametric unfolding based on the cotangent-weighted discrete Laplace equation and fixed-boundary Dirichlet conditions to obtain the mapped coordinates of each vertex on the two-dimensional plane. Based on the mapped coordinates of each vertex and its corresponding color value, a two-dimensional image corresponding to the sub-region mesh is rendered. For example, the mapped coordinates of each vertex can be used as the pixel position, and the corresponding RGB color value as the pixel color at that pixel position. An interpolation algorithm is used to fill the pixel regions between vertices to generate a complete two-dimensional image. The rendered 2D image synchronously preserves the positional information of edge points and the orientation features of the surface, providing directly input visual data for subsequent edge point recognition and region type classification, thus realizing an effective conversion from geometric data to visual data.

[0046] Optionally, the region types include the acromion region, the gingival sulcus region, and the mixed region.

[0047] In this embodiment of the invention, the shoulder area can be specifically understood as a region on the target tooth surface that is relatively flat and has minimal curvature variation. The shoulder area requires high positional accuracy of its edge lines, and its angular features should be preserved during the fitting process. The gingival sulcus area can be specifically understood as a region on the target tooth close to the gingiva with significant curvature variation. The gingival sulcus area requires high smoothness of its edge lines, and its smooth transition should be ensured during the fitting process. The mixed area can be specifically understood as a transitional region between the shoulder area and the gingival sulcus area. The curvature characteristics of the mixed area are between those of the shoulder area and the gingival sulcus area, neither completely flat nor completely steep.

[0048] S150. Determine the weight of each edge point set according to the region type, and obtain the edge line of the target tooth based on the edge point set and the weight of the edge point set.

[0049] Specifically, the weights of each edge point set are determined based on the region type. In the case of the shoulder area, the weight of the edge point set is the first weight, with a value ranging from [0.8, 1]. In the case of the gingival sulcus area, the weight of the edge point set is the second weight, with a value ranging from [0.1, 0.5]. In the case of the mixed area, the weight of the edge point set is the third weight, with a value ranging from [0.7, 1]. A weighted fitting is then performed based on the edge point sets and their weights to obtain the edge line of the target tooth, as shown in the following formula: ; in, The weights of the edge point set of the i-th sub-region grid are: Let L be the set of edge points of the i-th sub-region grid, L be the position of the target edge line obtained after weighted fitting, and n be the number of sub-region grids. By performing differentiated weighted fitting on the edge points based on the region type identification results, the problem that traditional global uniform fitting methods cannot simultaneously preserve the shoulder edge angle and ensure the smoothness of the gingival sulcus is solved, making the extracted edge line have both positional accuracy and morphological rationality.

[0050] The technical solution of this embodiment obtains a local mesh model of the target tooth based on a first three-dimensional mesh model of the oral cavity, which contains multiple teeth. Obtaining the local mesh allows subsequent sectioning and recognition to focus on the target region, effectively eliminating interference from irrelevant structures such as adjacent teeth. The local mesh model is sectioned from multiple angles to obtain multiple two-dimensional cross-sectional images. Edge point recognition is performed on these multiple two-dimensional cross-sectional images to obtain multiple initial edge points. The three-dimensional edge recognition problem is transformed into an edge point detection problem involving multiple two-dimensional cross-sectional images, reducing the problem complexity. Furthermore, locking the initial edge points provides reliable initial points for subsequent fine-grained localization. A corresponding sub-region mesh is extracted from the local mesh model centered on each initial edge point. Extracting sub-region meshes centered on the initial edge points narrows the search range and filters out interference. Even if the initial edge points have deviations, the true edge lines still fall within the sub-region mesh, allowing for accurate positioning through subsequent fine-grained recognition. Each sub-region mesh is rendered into a two-dimensional image, and edge point recognition is performed on the two-dimensional image to obtain the edge point set and region type of each sub-region mesh. Edge point recognition is performed on the dimensionality-reduced 2D image, enabling edge point recognition to be performed on a 2D plane, significantly reducing computational complexity. The weights of each edge point set are determined based on the region type, and the edge line of the target tooth is obtained based on the edge point sets and their weights. Differential weighted fitting preserves the angular features of the shoulder region while ensuring smoothness in the gingival sulcus region, overcoming the shortcomings of global uniform fitting which cannot accommodate the needs of different anatomical regions. In summary, the technical solution of this embodiment obtains a 2D cross-sectional view by multi-angle sectioning of a first 3D mesh model, obtains initial edge points from the 2D cross-sectional view, obtains sub-region meshes based on the initial edge points, converts the sub-region meshes into a 2D image, performs edge point recognition on the 2D image to obtain the edge point sets and region types of the sub-region meshes, determines the weights of each edge point set based on the region type, and obtains the edge line of the target tooth based on the edge point sets and their weights. This solves the defects of traditional recognition and fitting methods, such as loss of shoulder features due to over-smoothing or jagged edges due to under-smoothing, and improves the accuracy of edge line generation.

[0051] Based on the above embodiments, an optional example is provided, which can be used to extract the edge lines of a target tooth. A flowchart of the edge line generation process is shown below. Figure 2 As shown in the figure, firstly, based on the first three-dimensional mesh model of the oral cavity, a local mesh model of the target tooth is obtained; based on the local mesh model, a two-dimensional cross-sectional view is obtained, and initial edge points are obtained based on the two-dimensional cross-sectional view; based on the initial edge points, a sub-region mesh is obtained, and a two-dimensional image is obtained based on the sub-region mesh; edge point recognition is performed on the two-dimensional image; multiple edge point sets are fitted to obtain edge lines; and the edge lines are projected onto the jaw.

[0052] Specifically, the second and third-dimensional mesh models and segmentation information of the oral cavity are obtained. Based on the segmentation information, the major axis direction of each tooth is obtained. A plane is fitted using the least squares method, and the normal direction of the fitted plane is taken as the target direction, which is the Z-axis direction. The first three-dimensional mesh model is then adjusted so that the occlusal surface faces the Z-axis direction. A schematic diagram of the first three-dimensional mesh model is shown below. Figure 3 As shown. The location of the target tooth is found on the first 3D mesh model based on the segmentation information, and a local mesh model with a radius of 5mm and the center of the target tooth is obtained. A coordinate system is established with the center of the target tooth. The local mesh model is cut along the YOZ plane in this coordinate system, and the part with x < 0 is projected onto the YOZ plane to obtain a 2D cross-sectional view. The local mesh model is rotated counterclockwise by 45 degrees each time, with the Y-axis as the rotation axis, and the cutting of the local mesh model along the YOZ plane is repeated each time, with the part of the local mesh model with x < 0 projected onto the YOZ plane, resulting in a total of 8 2D cross-sectional views. By inferring and recognizing the 8 2D cross-sectional views from the YOLO model, the edge lines can be obtained through 2 points in each cross-section, resulting in a total of 16 initial edge points. A schematic diagram of the initial edge point recognition is shown below. Figure 4 As shown, sub-region meshes are obtained with a radius of 2mm centered on each initial edge point, resulting in a total of 16 sub-region meshes. For each vertex of each sub-region mesh, the vertex normal direction is calculated. The x, y, and z components of the normal direction of each vertex are mapped to the red, green, and blue channel values ​​in the RGB color space, respectively, obtaining the mapping relationship between the vertex normal direction and the RGB color space. A harmonic unfolding algorithm is used, which unfolds the sub-region meshes to a two-dimensional plane based on the cotangent-weighted discrete Laplace equation and the fixed boundary Dirichlet condition, obtaining the mapped coordinates of each vertex in the two-dimensional plane. Based on the mapped coordinates of each vertex and the color value corresponding to the vertex normal direction, a two-dimensional image corresponding to the sub-region mesh is rendered. The 2D images of the 16 sub-region meshes are identified through YOLO model inference, obtaining the edge point sets and region types (e.g., acromion region, gingival sulcus region, mixed region) within the sub-region meshes. Each edge point set can have 20 points. A schematic diagram of edge point set identification is shown below. Figure 5 As shown. Weights are set according to region type: For the acromion region, the weight of the edge point set is the first weight, with a value ranging from [0.8, 1]. For the gingival sulcus region, the weight of the edge point set is the second weight, with a value ranging from [0.1, 0.5]. For the mixed region, the weight of the edge point set is the third weight, with a value ranging from [0.7, 1]. A weighted fitting is performed based on the edge point set and its weights to obtain the edge line.

[0053] Figure 6This is a schematic diagram of an edge line generation device provided in an embodiment of the present invention. This embodiment is applicable to extracting tooth edge lines based on a three-dimensional mesh model of the oral cavity. The device can be implemented using software and / or hardware, and can be integrated into any device that provides edge line generation functionality, such as… Figure 6 As shown, the device for generating edge lines specifically includes: a local mesh model acquisition module 210, an initial edge point acquisition module 220, a sub-region mesh extraction module 230, an edge point set acquisition module 240, and an edge line acquisition module 250.

[0054] The local mesh model acquisition module 210 is used to acquire a local mesh model of a target tooth based on a first three-dimensional mesh model of the oral cavity, wherein the oral cavity contains multiple teeth; The initial edge point acquisition module 220 is used to perform multi-angle sectioning on the local mesh model to obtain multiple two-dimensional cross-sectional views, and to perform edge point identification on the multiple two-dimensional cross-sectional views to obtain multiple initial edge points; The sub-region mesh extraction module 230 is used to extract the corresponding sub-region mesh from the local mesh model with each of the initial edge points as the center; The edge point set acquisition module 240 is used to render each of the sub-region grids into a two-dimensional image by dimensionality reduction, perform edge point recognition on the two-dimensional image, and obtain the edge point set and region type of each of the sub-region grids. The edge line acquisition module 250 is used to determine the weight of each edge point set according to the region type, and obtain the edge line of the target tooth based on the edge point set and the weight of the edge point set.

[0055] The technical solution of this embodiment obtains a local mesh model of the target tooth based on a first three-dimensional mesh model of the oral cavity, which contains multiple teeth. Obtaining the local mesh allows subsequent sectioning and recognition to focus on the target region, effectively eliminating interference from irrelevant structures such as adjacent teeth. The local mesh model is sectioned from multiple angles to obtain multiple two-dimensional cross-sectional images. Edge point recognition is performed on these multiple two-dimensional cross-sectional images to obtain multiple initial edge points. The three-dimensional edge recognition problem is transformed into an edge point detection problem involving multiple two-dimensional cross-sectional images, reducing the problem complexity. Furthermore, locking the initial edge points provides reliable initial points for subsequent fine-grained localization. A corresponding sub-region mesh is extracted from the local mesh model centered on each initial edge point. Extracting sub-region meshes centered on the initial edge points narrows the search range and filters out interference. Even if the initial edge points have deviations, the true edge lines still fall within the sub-region mesh, allowing for accurate positioning through subsequent fine-grained recognition. Each sub-region mesh is rendered into a two-dimensional image, and edge point recognition is performed on the two-dimensional image to obtain the edge point set and region type of each sub-region mesh. Edge point recognition is performed on the dimensionality-reduced 2D image, enabling edge point recognition to be performed on a 2D plane, significantly reducing computational complexity. The weights of each edge point set are determined based on the region type, and the edge line of the target tooth is obtained based on the edge point sets and their weights. Differential weighted fitting preserves the angular features of the shoulder region while ensuring smoothness in the gingival sulcus region, overcoming the shortcomings of global uniform fitting which cannot accommodate the needs of different anatomical regions. In summary, the technical solution of this embodiment obtains a 2D cross-sectional view by multi-angle sectioning of a first 3D mesh model, obtains initial edge points from the 2D cross-sectional view, obtains sub-region meshes based on the initial edge points, converts the sub-region meshes into a 2D image, performs edge point recognition on the 2D image to obtain the edge point sets and region types of the sub-region meshes, determines the weights of each edge point set based on the region type, and obtains the edge line of the target tooth based on the edge point sets and their weights. This solves the defects of traditional recognition and fitting methods, such as loss of shoulder features due to over-smoothing or jagged edges due to under-smoothing, and improves the accuracy of edge line generation.

[0056] Optionally, based on the above embodiments, the device further includes a first three-dimensional mesh model acquisition module, configured to: acquire a second three-dimensional mesh model, the second three-dimensional mesh model being constructed from the oral cavity acquisition image; acquire segmentation information of the second three-dimensional mesh model, and acquire the major axis direction of each tooth based on the segmentation information; the segmentation information is used to identify each tooth in the second three-dimensional mesh model; perform plane fitting according to the major axis direction of each tooth, and take the normal direction of the fitted plane as the target direction; adjust the orientation of the second three-dimensional mesh model according to the target direction to obtain the first three-dimensional mesh model, wherein the occlusal surface of the first three-dimensional mesh model faces the target direction.

[0057] Based on the above embodiments, optionally, the local mesh model acquisition module 210 is used to: acquire the target tooth based on the segmentation information; and extract a local mesh model of the target tooth from the first three-dimensional mesh model with the center of the target tooth as the center of the sphere and a first preset length as the radius.

[0058] Based on the above embodiments, optionally, the initial edge point acquisition module 220 is used to: rotate the local mesh model to multiple different angles with a preset coordinate axis as the rotation axis and a preset angle step size; and project each rotated local mesh model onto the preset plane as the cutting plane to obtain multiple two-dimensional cross-sectional views.

[0059] Based on the above embodiments, optionally, the sub-region mesh extraction module 230 is used to: extract from the local mesh model with each initial edge point as the center and a second preset length as the radius to obtain the sub-region mesh corresponding to each initial edge point.

[0060] Based on the above embodiments, optionally, the edge point set acquisition module 240 is used to: calculate the normal direction of each vertex in the sub-region mesh; determine the color value corresponding to each vertex according to the components of the normal direction of each vertex in each coordinate axis direction; unfold the sub-region mesh to a two-dimensional plane to obtain the mapped coordinates of each vertex on the two-dimensional plane; and render and generate a two-dimensional image corresponding to the sub-region mesh according to the mapped coordinates of each vertex and the corresponding color value.

[0061] Based on the above embodiments, optionally, the region type includes the acromion region, the gingival sulcus region, and the mixed region.

[0062] The above-described products can perform the methods provided in any embodiment of the present invention, and have the corresponding functional modules and beneficial effects for performing the methods.

[0063] Figure 7 A schematic diagram of an electronic device 10, which can be used to implement embodiments of the present invention, is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0064] like Figure 7As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0065] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0066] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as edge line generation methods.

[0067] In some embodiments, the edge line generation method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or mounted on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the edge line generation method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the edge line generation method by any other suitable means (e.g., by means of firmware).

[0068] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0069] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0070] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0071] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0072] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0073] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0074] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0075] This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the edge line generation method according to any embodiment of the invention.

[0076] In implementing the computer program product, computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof. Programming languages ​​include object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0077] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for generating edge lines, characterized in that, include: Based on the first three-dimensional mesh model of the oral cavity, a local mesh model of the target tooth is obtained, wherein the oral cavity contains multiple teeth; The local mesh model is sectioned from multiple angles to obtain multiple two-dimensional cross-sectional images. Edge points are identified on the multiple two-dimensional cross-sectional images to obtain multiple initial edge points. Centered on each of the initial edge points, a corresponding sub-region mesh is extracted from the local mesh model; Each of the sub-region grids is rendered into a two-dimensional image by dimensionality reduction, and edge point recognition is performed on the two-dimensional image to obtain the edge point set and region type of each sub-region grid. The weights of each edge point set are determined according to the region type, and the edge line of the target tooth is obtained based on the edge point set and its weight.

2. The method according to claim 1, characterized in that, The process of obtaining the first three-dimensional mesh model includes: A second three-dimensional mesh model is obtained, which is constructed from the oral cavity acquisition images; Obtain segmentation information of the second three-dimensional mesh model, and obtain the major axis direction of each tooth based on the segmentation information; the segmentation information is used to identify each tooth in the second three-dimensional mesh model. Plane fitting is performed based on the long axis direction of each tooth, and the normal direction of the fitted plane is taken as the target direction. The second three-dimensional mesh model is oriented according to the target direction to obtain the first three-dimensional mesh model, wherein the interlocking surface of the first three-dimensional mesh model faces the target direction.

3. The method according to claim 2, characterized in that, The first three-dimensional mesh model based on the oral cavity is used to obtain a local mesh model of the target tooth, including: Based on the segmentation information, the target tooth is obtained; Using the center of the target tooth as the center of the sphere and a first preset length as the radius, a local mesh model of the target tooth is obtained by cutting out the first three-dimensional mesh model.

4. The method according to claim 1, characterized in that, The process of cutting the local mesh model from multiple angles to obtain multiple two-dimensional cross-sectional views includes: Using a preset coordinate axis as the rotation axis, the local mesh model is rotated to multiple different angles according to a preset angle step size; Using a preset plane as the sectional plane, each rotated local mesh model is projected onto the sectional plane to obtain multiple two-dimensional cross-sectional views.

5. The method according to claim 1, characterized in that, The step of extracting corresponding sub-region meshes from the local mesh model, centered on each initial edge point, includes: Using each initial edge point as the center and a second preset length as the radius, a sub-region mesh corresponding to each initial edge point is obtained by extracting from the local mesh model.

6. The method according to claim 1, characterized in that, The step of reducing the dimensionality of each sub-region grid to render a two-dimensional image includes: Calculate the normal direction of each vertex in the sub-region mesh; The color value corresponding to each vertex is determined based on the components of the normal direction of each vertex in each coordinate axis direction; The sub-region mesh is unfolded onto a two-dimensional plane to obtain the mapped coordinates of each vertex on the two-dimensional plane. Based on the mapped coordinates and corresponding color values ​​of each vertex, a two-dimensional image corresponding to the sub-region mesh is rendered.

7. The method according to claim 1, characterized in that, The region types include the acromion region, the gingival sulcus region, and the mixed region.

8. An edge line generating device, characterized in that, include: The local mesh model acquisition module is used to acquire a local mesh model of a target tooth based on a first three-dimensional mesh model of the oral cavity, wherein the oral cavity contains multiple teeth; The initial edge point acquisition module is used to perform multi-angle sectioning on the local mesh model to obtain multiple two-dimensional cross-sectional images, and to identify edge points on the multiple two-dimensional cross-sectional images to obtain multiple initial edge points; The sub-region mesh extraction module is used to extract the corresponding sub-region mesh from the local mesh model, centered on each of the initial edge points; The edge point set acquisition module is used to render each of the sub-region grids into a two-dimensional image by reducing its dimension, and to perform edge point recognition on the two-dimensional image to obtain the edge point set and region type of each of the sub-region grids. An edge line acquisition module is used to determine the weight of each edge point set according to the region type, and obtain the edge line of the target tooth based on the edge point set and the weight of the edge point set.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor to enable the at least one processor to perform the edge line generation method according to any one of claims 1-7.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the edge line generation method according to any one of claims 1-7.

Images