Generation method and system of initial grid model of gingiva

Abstract

The invention discloses a gingival initial grid model generation method and system, and relates to the technical field of gingival generation methods, and the method comprises the steps: collecting different boundary regions of a gingival; determining an upper surface region, an upper jaw region, a side surface region and a bottom surface region according to different boundary regions of the gingiva; an initial grid model of the gingiva is determined according to synthesis of the upper surface area, the palate area, the side face area and the bottom face area, integral synthesis of the upper surface area, the palate area, the side face area and the bottom face area is achieved, and multi-area integration of the upper surface area, the palate area, the side face area and the bottom face area is achieved. And the regional accuracy of the initial grid model is ensured, the accurate construction of the initial grid model is realized, and the initial grid model has no gingival aesthetic form.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of gingival generation method, and particularly relates to a gingival initial mesh model generation and system. BACKGROUND

[0002] With the development of technology, in the treatment scheme of complete denture, the gingiva is needed to bear and fix the denture and disperse the bite pressure, in the digital complete denture design solution, the corresponding initial mesh model is constructed for the gingiva, in the prior art, the existing initial mesh model does not fully consider the compatibility of different regions of the gingiva, which affects the regional accuracy of the initial mesh model, and the accurate construction of the initial mesh model cannot be realized. SUMMARY

[0003] The present application provides a gingival initial mesh model generation and system.

[0004] The present application provides a gingival initial mesh model generation, which comprises the following steps: Collecting different boundary regions of the gingiva; Determining the upper surface region, the palate region, the side region and the bottom region according to the different boundary regions of the gingiva; Determining the initial mesh model of the gingiva according to the synthesis of the upper surface region, the palate region, the side region and the bottom region.

[0005] The present application provides a gingival initial mesh model generation system, which is applied to the gingival initial mesh model generation. The collecting module is used for collecting different boundary regions of the gingiva; The region module is used for determining the upper surface region, the palate region, the side region and the bottom region according to the different boundary regions of the gingiva; The initial mesh model module is used for determining the initial mesh model of the gingiva according to the synthesis of the upper surface region, the palate region, the side region and the bottom region.

[0006] Compared with the prior art, the present application has the following advantages: In this embodiment of the invention, different boundary regions of the gingiva are collected using the method described in this embodiment; the upper surface region, maxillary region, lateral region, and bottom region are determined based on the different boundary regions of the gingiva; the initial mesh model of the gingiva is determined based on the synthesis of the upper surface region, maxillary region, lateral region, and bottom region, realizing the overall synthesis of the upper surface region, maxillary region, lateral region, and bottom region, and ensuring the regional accuracy of the initial mesh model, thus achieving the accurate construction of the initial mesh model. The initial mesh model does not have the aesthetic morphology of the gingiva. Attached Figure Description

[0007] Figure 1 This is a schematic diagram illustrating the process of generating the initial mesh model of the gingiva in an embodiment of the present invention; Figure 2 This is a flowchart illustrating step S11 in the generation of the initial mesh model of the gingiva in an embodiment of the present invention. Figure 3 This is a flowchart illustrating step S12 in the generation of the initial mesh model of the gingiva in an embodiment of the present invention. Figure 4 This is a flowchart illustrating step S13 in the generation of the initial mesh model of the gingiva in an embodiment of the present invention. Figure 5 This is a schematic diagram of the structural composition of the system for generating the initial mesh model of the gingiva in an embodiment of the present invention; Figure 6 This is a schematic diagram of the gingival boundary generated from the initial mesh model of the gingiva in an embodiment of the present invention; Figure 7 This is a schematic diagram of the initial mesh model of the gingiva in an embodiment of the present invention. Detailed Implementation

[0008] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0009] Please see Figures 1 to 7 A method for generating an initial mesh model of the gingiva, applied to scenarios involving the generation of an initial mesh model of the gingiva; the generation of the initial mesh model of the gingiva includes: Step S11: Collect samples from different boundary areas of the gingiva; Step S12: Determine the upper surface region, maxillary region, lateral region, and underside region based on the different boundary areas of the gingiva; Step S13: Determine the initial mesh model of the gingiva based on the synthesis of the upper surface region, the maxillary region, the lateral region, and the bottom region; refer to Figure 2 In step S11, the specific steps are as follows: S111: Collect multiple image data of the gingiva, determine the equidistant line of the gingiva boundary based on the recognition of multiple image data of the gingiva, and output the set of points of the equidistant line. S112: Determine the dentition region based on multiple image data of the gingiva, determine the dentition type features based on the detection of the dentition region, and mark the gingival width distribution of the dentition type features; S113: Determine key point segments based on the division of the upper equidistant line point set, and divide the buccal, lingual, and two-end boundary regions based on the key point segments and the gingival width distribution of the dental and jaw type characteristics.

[0010] In the embodiments of this application, a high-resolution oral endoscope or oral scanner is used to acquire images of the patient's gingiva from different angles (front, 45° side, 90° side); typically at least 5-8 images are required to ensure complete coverage; the image resolution is not less than 1920×1080, a standardized light source (approximately 5500K color temperature) is used to reduce color deviation, and a contrast agent (such as methylene blue solution) is used to enhance boundary clarity.

[0011] Image segmentation is performed using a deep learning network. The specific process is as follows: the CLAHE algorithm is used to enhance contrast, the gingival region is identified using a trained U-Net model, and morphological operations (opening and closing operations) are applied to remove noise. Optionally, the network input size is 512×512, the Dice coefficient is used as the loss function, and the IoU threshold is set to 0.85.

[0012] First, the gingival contour is extracted using Canny edge detection (threshold 100-200). Then, isometric lines are generated by offsetting the contour inward by 1.5mm (clinical standard gingival thickness) using a contour offset algorithm. The isometric lines are generated using B-spline curve fitting with a control point spacing of 0.2mm. The isometric lines are then discretized into a point set with a point spacing of 0.1mm. Each point contains three-dimensional coordinates (x, y, z) and normal vector information. The point set is output in PLY format, containing position and normal vector information.

[0013] Furthermore, the dentition region is determined based on multiple image data of the gingiva, and the dentition type features are determined based on the detection of the dentition region. The gingival width distribution of the dentition type features is marked, which is compatible with the overall consideration of the detection of the dentition region and ensures the accuracy of the dentition type features.

[0014] At this point, the acquired gingival images are standardized, including: adjusting all images to 1024×768 pixels, using a white balance algorithm (such as the gray-world algorithm) to eliminate color deviations under different light sources, applying a non-local mean denoising algorithm to preserve edge details while reducing image noise, and segmenting the dentition region using a deep learning semantic segmentation model (such as DeepLabv3+) to identify the dentition region. Input: preprocessed RGB image; Output: binary mask of the dentition region; Post-processing: applying morphological operations (dilation + erosion) to fill small holes and smooth boundaries.

[0015] A multi-class classifier (such as SVM or random forest) is used to identify dentofacial types. The following features are extracted from the segmented dentofacial regions: geometric features: arch curvature, dentition width, and anterior-posterior tooth ratio; texture features: contrast, energy, and entropy based on GLCM (gray-level co-occurrence matrix); shape features: Fourier descriptors and Hu moments; dentofacial type definitions: Class I: normal occlusion, regular arch shape; Class II: mandibular retrusion, narrow arch shape; Class III: mandibular protrusion, wide arch shape; special types: open bite, deep overbite, etc. Simultaneously, a measurement baseline was defined in the dentition region: from the gingival margin to the ligamentum symphysis, the gingival width was measured at equal intervals (0.5 mm) along the long axis of the tooth. Using an image ranging algorithm, based on the calibration relationship between pixels and actual dimensions, the dental arch was divided into 6 regions: central incisor region, lateral incisor region, canine region, premolar region, first molar region, and second molar region. For each region, the average gingival width, width standard deviation, and width variation trend (linear fitting slope) were calculated to generate a width distribution heatmap, and color coding was used to represent width changes.

[0016] Therefore, the key point segments are determined based on the division of the upper isometric line point set. The buccal, lingual, and end boundary regions are divided according to the key point segments and the gingival width distribution of the dental and jaw type characteristics. This approach takes into account the overall consideration of the division of the upper isometric line point set and ensures the accuracy of the key point segments.

[0017] At this point, the key point segmentation is determined based on the division of the upper equidistant line point set. Input: Upper equidistant line point set output by S111 (e.g., 1,423 points); Smoothing: Apply B-spline curve fitting to reduce noise impact; Parameter settings: Number of control points = 20, smoothing factor = 0.1; Key point detection algorithm: Local extremum detection: Find the local maximum curvature point; Threshold setting: Points with curvature > 0.15 are marked as candidate key points; Minimum distance constraint: Minimum distance between adjacent key points = 15mm; Key point classification: Anatomical key points: Midpoint of central incisor, vertex of canine, distal center of first molar, etc.; Geometric key points: Curvature extremum points, inflection points; Final key point set: Merge anatomical and geometric key points and remove redundancy.

[0018] Based on key points, the equidistant line is divided into 6-8 natural segments. Each segment is fitted with a polynomial to ensure a smooth transition. Segmental verification: The continuity of the first derivative at the connection points of adjacent segments is checked. Region division logic: Buccal region determination: Condition 1: Width value > 3.5mm (based on the width distribution of S112); Condition 2: Located on the outer side of the dental arch (judged by the dental arch centerline); Condition 3: Smooth curvature change (curvature standard deviation < 0.05); Lingual region determination: Condition 1: Width value < 2.8mm; Condition 2: Located on the inner side of the dental arch; Condition 3: Surface texture features conform to lingual features (based on texture analysis of S112); End boundary regions: Anterior boundary: The region from the central incisor to the canine; Posterior boundary: The region from the first molar to the posterior boundary; Transition region: The boundary transition is handled using a gradient function; Dental type adaptation: Class I (normal occlusion): Standard division parameters; Class II (distal occlusion): Buccal region expanded by 5-10%; Class III (mesial occlusion): Lingual region expanded by 5-10%.

[0019] In the real-time example of this application, the gingival boundary line is acquired, and the upper and lower equidistant lines of the gingival boundary line are marked. At the same time, the vertices on the nearest dental model to each point on the gingival boundary line are marked. The vertices and multiple faces within the 5-neighborhood of these vertices are found. A sub-model is obtained by synthesizing these vertices and faces. The above sub-model is smoothed by first-order Laplacian to obtain a model with uniform curvature variation.

[0020] Calculate the normal of each vertex on the model with uniform curvature change closest to each point on the gingival boundary line. Sequentially calculate the coordinates (tempUpPoint and tempDownPoint) of each point on the gingival boundary line after moving it along the normal direction of the vertex by the minimum gingival thickness distance and a distance equal to -minimum gingival thickness / 2. Calculate the distance between tempUpPoint and the curved gingival boundary line. If the distance is greater than the minimum gingival thickness distance * 0.9, add tempUpPoint to the point set of upBorderLine and tempDownPoint to the point set of downBorderLine. Optionally, tempUpPoint is a temporary coordinate point during the gingival boundary line processing. Specifically, it is the new position obtained by moving each point on the gingival boundary line along the normal direction of the nearest vertex by the minimum gingival thickness distance. This point represents the potential position after the gingival boundary expands outward, used to generate the upper gingival boundary line curve. If the distance between this point and the original gingival boundary line meets a specific condition (greater than 0.9 times the minimum gingival thickness), it will be added to the point set of the upper gingival boundary line.

[0021] `tempDownPoint` refers to another temporary coordinate point during the gingival boundary line processing. It is obtained by shifting each point on the gingival boundary line in the reverse direction along the normal to the nearest vertex by half the minimum gingival thickness (i.e., -minimum gingival thickness / 2). This point represents the potential location after the gingival boundary has contracted inward, used to generate the lower gingival boundary line curve. When the corresponding `tempUpPoint` satisfies the distance condition, this `tempDownPoint` is added to the point set of the lower gingival boundary line.

[0022] `upBorderLine` refers to the upper gingival boundary line, a curve generated through algorithmic processing. During the gingival model generation process, when the distance between `tempPoint` (i.e., `tempUpPoint`) and the original gingival boundary line is greater than 0.9 times the minimum gingival thickness, `tempPoint` is added to the point set of `upBorderLine`, thus forming the upper gingival boundary line. This curve represents the upper boundary of the gingival tissue in the vertical direction and is an important component in constructing a complete gingival model.

[0023] The `downBorderLine` refers to the lower gingival boundary line, a curve generated through algorithmic processing. During the gingival model generation process, when the distance between `tempDownPoint` and the original gingival boundary line is greater than 0.9 times the minimum gingival thickness, this `tempDownPoint` is added to the point set of `downBorderLine`, thus forming the lower gingival boundary line. This curve represents the lower boundary of the gingival tissue in the vertical direction, corresponding to the `upBorderLine` (upper gingival boundary line), together forming the boundary contour of the complete gingival model.

[0024] In general, the gingival boundary ranges of the maxillary and mandibular models are different. The maxillary model has an additional maxillary region compared to the mandibular model. In this case, the center points cP1 and cP2 of the first and last teeth are calculated respectively. Taking cP1 and cP2 as points on the plane and the buccal direction as a vector on the plane, two planes p1 and p2 are obtained. Calculate the intersection points of planes p1 and p2 at points cP1 and cP2 in the buccal direction with the upBorderLine (upper gingival boundary line), respectively, to obtain the center points sideBorderCenterP1 and sideBorderCenterP2 of the gingival boundaries (sideBorderCenterP1 and sideBorderCenterP2 refer to the center points of the lateral boundaries in the gingival model. These two points are usually used to define the position and direction of the lateral boundary lines of the gingival model). Divide the upBorderLine (upper gingival boundary line) into two sub-curves according to sideBorderCenterP1 and sideBorderCenterP2. The curve tempBuccleBorderLine is the part between sideBorderCenterP1 and sideBorderCenterP2, and the curve tempBongueBorderLine is the part between sideBorderCenterP2 and sideBorderCenterP1. Optionally, `tempBuccleBorderLine` refers to a temporary buccal boundary line. During gingival model generation, this term is commonly used to denote a temporary boundary line on the buccal side (i.e., the side closer to the cheek). This line may be an intermediate result generated during calculation or adjustment, used to define the boundary extent or contour of the gingival model on the buccal side. `tempBuccleBorderLine` may be used to assist in modeling or for further optimization before the final boundary line of the gingival model is determined.

[0025] For each tooth model, calculate the point tempPoint on the alveolar line closest to its center point, and construct a plane tempPlane with tempPoint and each tooth distally. Calculate the intersection of this plane with the two sub-curves to obtain buccleCollisionPoint (intersection with the buccal curve) and bongueCollisionPoint (intersection with the lingual curve).

[0026] Calculate the horizontal distance between tempPoint and buccleCollisionPoint (buccal collision point) respectively, and take the average value to obtain gingivaArgWidth (gingival parameter width). At the same time, calculate the horizontal distance between tempPoint and bongueCollisionPoint (lingual collision point) respectively, and calculate the standard deviation of the horizontal distance. If the standard deviation is greater than a certain threshold, the dental type is determined to be maxillary; otherwise, the dental type is determined to be mandibular.

[0027] Simultaneously, the upper equidistant line upBorderLine (upper gingival boundary line) is divided into four curve segments, corresponding to the buccal boundary upBuccleBorderLine (upper buccal boundary line), the lingual boundary upBongueBorderLine (upper lingual boundary line), and the two boundaries at both ends upSideBorderLine1 (upper side boundary line 1) and upSideBorderLine2 (upper side boundary line 2). Using the points sideBorderCenterP1 (side center point 1) and sideBorderCenterP2 (side center point 2) as centers, points are taken at certain distances along both the positive and negative directions on the upBorderLine (upper gingival boundary line) curve, resulting in four points: points segP1 (segment point 1) and segP2 (segment point 2) on both sides of sideBorderCenterP1, and points segP3 (segment point 3) and segP4 (segment point 4) on both sides of sideBorderCenterP2. These four points are used to divide the upBorderLine into four segments, as shown below. Figure 6 The sub-curve shown.

[0028] refer to Figure 3 In step S12, the specific steps are as follows: S121: Based on the division of the gingival boundary region, multiple curves are constructed for the buccal, central, and lingual positions of the teeth. The boundary curve of the alveolar bone morphology is determined based on the multiple curves related to the alveolar bone. The corresponding upper surface region is constructed based on the multiple curves for the buccal, central, and lingual positions of the teeth, the boundary curve of the alveolar bone morphology, and the buccal gingival boundary line. S122: Determine the tongue-side boundary curve segment based on the curve of the tongue-side position, and mark multiple sampling points. Construct a geometric framework for the maxillary region based on the multiple sampling points. Construct a UV spline surface based on the geometric framework of the maxillary region and the curve of the tongue-side position to output the corresponding maxillary region. S123: Construct the transition curve of the side profile based on the set of points on the upper equidistant line, and determine multiple side vertices based on the recognition of the transition curve of the side profile, and construct the side region based on the multiple side vertices. S124: Collect dental models, determine the gingival boundary based on the identification of dental models, mark multiple dental model vertices, and construct the bottom surface region of the gingiva based on the multiple dental model vertices.

[0029] In the embodiments of this application, the input data is: the set of points in the buccal, lingual, and end boundary regions divided in S113; the curve construction method is: buccal curve: fitted with a cubic B-spline curve, with a control point interval of 2 mm; the center curve: generated by the weighted average of the buccal and lingual curves, with a weight ratio of 0.6:0.4; the lingual curve: also fitted with a cubic B-spline curve, with a control point interval of 1.5 mm (because the lingual side is more complex); the curve smoothing process is: the Laplacian smoothing algorithm is applied, with a smoothing factor of 0.3; the curve continuity is guaranteed by G2 continuity (curvature continuity).

[0030] Acquire CT scan data or cone-beam CT (CBCT) data, extract alveolar bone using threshold segmentation and region growing algorithms, interpolate CT slices to ensure isotropic resolution, and apply the Canny edge detection operator with morphological closing operation (33 structuring elements). Employ non-uniform rational B-spline (NURBS) curves and adaptively distribute control points: dense in high curvature regions and sparse in low curvature regions.

[0031] Surface construction method: Coons surface interpolation method is used; Boundary conditions: Four boundary curves: buccal curve, lingual curve, front curve, and back curve; Internal constraints: Alveolar bone boundary curve is used as internal constraint condition; Surface parameterization: U direction: along the long axis of the tooth; V direction: around the tooth; Continuity guarantee: G1 continuity (tangential continuity); Mesh refinement: automatically increase mesh density in areas with high curvature.

[0032] It can construct anatomically accurate and geometrically continuous supragingival surface regions, providing a key geometric basis for subsequent gingival model construction; especially when dealing with different dentofacial types, this method can adaptively adjust curve and surface parameters to ensure that the generated model conforms to individualized anatomical features. In the real-time example of this application, the corresponding upper surface region is constructed as follows: (1) Calculation of the three curves L3, L4, and L5 related to tooth position a. Sample upSideBorderLine1 (upper boundary line 1) and upSideBorderLine2 (upper boundary line 2) evenly to obtain 7 points including the beginning and end points of the curve. In the direction from the cheek to the tongue, the 7 points on the upSideBorder1 curve are tempSide1P1 to tempSide1P7 (temporary point 1 to temporary point 7 of side 1), and the 7 points on the upSideBorder2 curve are tempSide2P1 to tempSide2P7 (temporary point 1 to temporary point 7 of side 2). b. Fit the center point of each tooth to tempSide1P4 and tempSide2P4 to form curve L4; c. Calculate the buccal center point tempBuccleP (buccal temporary point) of the cervical margin of each tooth. If the distance between tempBuccleP and the tooth center line is less than a threshold, then translate tempBuccleP in the buccal direction to a position that meets the threshold distance. Fit all tempBuccleP with tempSide1P3 and tempSide2P3 to form curve L3. d. Calculate the lingual center point tempBongueP (lingual temporary point) of the tooth using the same method as above, and fit all tempBongueP points with tempSide1P5 and tempSide2P5 to form curve L5; (2) Calculation of two alveolar bone-related curves L2 and L6 a. Sample the alveolar line (alveolar line drawn by the user) evenly to obtain n points, namely tempAPoint1 to tempAPointn (temporary alveolar point 1 to temporary alveolar point n), where n is the number of teeth; b. Calculate the tangent direction of the curve at each of the above points from tempANormal1 to tempANormaln (alveolar bone normal 1 to alveolar bone normal n), construct planes (tempAPointi, tempANormali) respectively, calculate the intersection of the plane with the buccle border line upBuccleBorderLine (upper buccle border line) to obtain tempCBucclePointi (temporary buccle intersection point i), calculate the intersection of the plane with the lingual border line upBongueBorderLine (upper lingual border line) to obtain tempCBonguePointi (temporary lingual intersection point i) (i∈[1,n]); c. Calculate the length components of each tempAPointi and the corresponding tempCBucclePointi in the occlusal direction: distVertBucclei (buccal vertical distance i) and distHorBucclei (buccal horizontal distance i). Calculate the length components of each tempAPointi and the corresponding tempCBonguePointi in the occlusal direction: distVertBonguei (lingual vertical distance i) and distHorBonguei (lingual horizontal distance i). If the lingual intersection point tempCBonguePointi does not exist, then distVertBonguei = 0, distHorBonguei = c, where c is a set constant. d. Translate each tempAPointi above to the corresponding buccal direction by a distance distHorBucclei, and to the occlusal direction by a distance distVertBucclei / 2, to obtain n points on the buccal line of the alveolar bone; e. Fit curve L2 together with the points on the buccal side line of the alveolar bone mentioned above and tempSide1P2 and tempSide2P2; f. When the dentition type is mandibular, each tempAPointi is translated by distHorBonguei distance in the opposite direction to the corresponding buccal side and by distVertBonguei / 2 distance in the occlusal direction to obtain n points on the alveolar bone lingual line. These points are then fitted together with tempSide1P6 and tempSide2P6 to obtain curve L6. g. When the dentition type is maxillary, each tempAPointi is translated by distHorBonguei distance in the opposite direction to the corresponding buccal side. The intersection of the ray from the translated tempAPointi point in the opposite direction to the occlusal direction with the maxillary model is calculated. Points above the intersection points with gingivaThickness are selected. These points are then fitted together with tempSide1P7 and tempSide2P7 to obtain curve L7. h. When the dental type is maxillary, take the fitting points of L5 and L7 without the beginning and end endpoints, calculate the midpoint of these points respectively, and obtain n points. Fit these points together with tempSide1P6 and tempSide2P6 to obtain curve L6. (3) Calculation of gingival boundaries L1 and L7 a. The buccal gingival boundary line L1 is the above-mentioned upBorderBuccleLine. b. When the dentition type is mandibular, the lingual gingival boundary line L7 is the above-mentioned upBorderBongueLine. c. When the dentition type is maxillary, the lingual gingival boundary line L7 is collected.

[0033] (4) Curve fitting and sampling a. Sample n1 parts of curves L1 to L7 evenly. In the case of full mouth, n1 can be 400. Take one sample point from each curve in sequence and fit a new curve tempC (temporary curve C) to these 7 sample points. b. Uniformly sample the n1 curves tempC mentioned above to obtain all points on the upper surface and construct the corresponding upper surface regions.

[0034] Further, the tongue-side boundary curve segment is determined based on the curve of the tongue-side position. Input data: the tongue-side curve generated in S121 (a cubic B-spline curve containing 38 control points); curve segment division method: the tongue-side curve is divided into multiple curve segments using curvature analysis; curvature threshold: 0.1mm⁻¹, exceeding this threshold is considered a critical change point; minimum segment length: 5mm, to avoid excessively short curve segments; boundary condition processing: ensure G1 continuity (tangent continuity) between adjacent curve segments; high curvature areas have high sampling density, and low curvature areas have low sampling density; basic sampling interval: 2mm; high curvature area sampling interval: 0.5mm; sampling point attribute record: three-dimensional coordinates (x,y,z), parameter position (t), normal vector, curvature value.

[0035] The initial triangular mesh is constructed using the Delaunay triangulation algorithm. Boundary constraints: tongue-shaped curve segments are used as fixed boundaries. Internal point generation: internal control points are generated based on sampling point interpolation. Frame optimization: the Laplacian smoothing algorithm is applied. The number of iterations is 10. Boundary points are fixed, and only internal points are optimized to ensure mesh manifoldness and prevent self-intersection.

[0036] A UV-spline surface was constructed based on the geometric framework of the palatal region and the curves at the lingual position. UV parameterization: U direction: along the lingual curve, parameter range [0,1]; V direction: extending inward perpendicular to the lingual curve, parameter range [0,0.8]; parameterization method: harmonic mapping; spline surface construction: bicubic B-spline surface; control point mesh: 20×15; surface continuity: C2 continuity; surface fitting: least squares fitting, weight function: inverse distance ratio; iterative optimization: 5 times; convergence threshold: 0.05mm.

[0037] It can construct anatomically accurate and geometrically continuous maxillary regions. In particular, when dealing with the transition region from the lingual boundary to the maxilla, the smoothness and accuracy of the surface are ensured through adaptive sampling and UV parameterization technology. The generated maxillary region model not only conforms to individualized anatomical features, but also provides a key geometric basis for the subsequent overall construction of the gingival model.

[0038] In the real-time example of this application, the corresponding maxillary region is constructed as follows: In the case of a maxillary model, the gingiva will generally also include the maxillary region, which is the area enclosed by L7 and the up-Bongue Border Line of the teeth.

[0039] a. Divide the above curve L7 into three segments: S1, S2, and S3; b. Uniformly sample 7 points from upBongueBorderLine excluding both ends, namely tempBPoint1 to tempBPoint7 (temporary points on the tongue side 1 to 7). Then uniformly sample 7 points from S2 excluding both ends, namely tempSPoint1 to tempSPoint7 (temporary points on S2 1 to 7). c. Sample 30 points tempUPoint (temporary points of the upper jaw) uniformly from the line segments from tempBPointi to tempSPointi (i∈[1,7]); d. Calculate the intersection of the ray passing through tempUPoint and the ray in the opposite direction of the occlusal direction with the maxillary model, and take the point above the intersection by the distance of gingivaThickness to obtain the straight line bongueL1 to bongueL7 (lingual line 1 to lingual line 7). e. Construct a UV spline surface using S1, bondL1 to bondL7, S3, and upBongueBorderLine, S2; f. Perform uniform sampling on the above curved surface to obtain the upper surface of the maxillary region, and locate the corresponding maxillary region.

[0040] Furthermore, a transition curve for the side profile is constructed based on the set of equidistant line points, and multiple side vertices are determined based on the identification of the transition curve of the side profile. The side region is constructed based on the multiple side vertices, which takes into account the overall consideration of the identification of the transition curve of the side profile and ensures the accuracy of the multiple side vertices.

[0041] At this point, a transition curve for the side profile is constructed based on the set of points on the upper equidistant line. Input data: the set of points on the upper equidistant line output in S111 (e.g., 1,423 points). Transition curve construction method: fitting with a cubic B-spline curve. Control point spacing: 1.5mm (closer than S121 due to more complex side changes). Curve smoothness control: Laplacian smoothing with a smoothing factor of 0.2. At the same time, it maintains G1 continuity with the upper surface region and C0 continuity with the bottom region (positional continuity is sufficient).

[0042] Meanwhile, the vertex recognition algorithm identifies key vertices based on curvature analysis; curvature calculation uses the five-point difference method; curvature threshold: 0.15mm (higher than S121, due to more drastic changes on the sides); Simultaneously, side regions are constructed based on multiple side vertices, using the Delaunay triangulation algorithm; boundary constraints ensure consistency with the boundaries of the upper and lower surfaces; mesh optimization includes edge swapping to optimize triangle quality; vertex adjustment uses Laplacian smoothing, iterating 5 times; patch size control sets the target side length to 2mm; and it is continuous with the upper surface region (G1 continuity) and the lower surface region (C0 continuity).

[0043] It can construct anatomically accurate and geometrically continuous lateral regions. Especially when dealing with the complex morphology of the gingival lateral surface, it ensures the accuracy and computational efficiency of the model through adaptive vertex distribution and optimized mesh generation technology. The generated lateral region model can not only accurately reflect the lateral morphological features of the gingiva, but also provide a key geometric foundation for the subsequent overall construction of the gingival model.

[0044] In the real-time example of this application, the corresponding side vertices are constructed as follows: a. Sample the upBorderLine (upper boundary line) and downBorderLine (lower boundary line) curves described in 1.1 evenly to obtain points P1 and P0 on the upper and lower equidistant lines of the gingival boundary, respectively, and calculate the direction dirBorderUp (boundary direction) from P0 to P1. b. Calculate point P2 after shifting P0 by the distance gingivaThickness (gingival thickness) in the dirBorderUp direction; c. Sample 4 points uniformly on line segment P0P2, and add point P1 for a total of 5 points, and fit a curve tempC (temporary curve C). d. Sample 10 points uniformly on the curve tempC and add these sampled points to the side vertex set.

[0045] Therefore, by collecting dental models, determining the gingival boundary line based on the recognition of dental models, and marking multiple dental model vertices, the bottom surface region of the gingiva is constructed based on the multiple dental model vertices. This approach takes into account the overall consideration of dental model recognition and ensures the accuracy of the gingival boundary line.

[0046] At this point, a dental model is acquired, and the boundary recognition algorithm is used: the average curvature of each vertex is calculated, and the area with curvature > 0.05 mm is regarded as the gingival boundary. Starting from the high curvature point, the boundary is extended to form a continuous boundary; boundary smoothing is performed: moving least squares (MLS) is applied for smoothing, with a smoothing factor of 0.3; boundary closure is performed: the opening boundary is detected and automatically closed, with a closure error of < 0.1 mm.

[0047] Vertex selection strategy: Uniform sampling: Select a vertex every 1mm along the boundary line; Feature point preservation: Force the preservation of curvature maxima; Density adaptation: Increase sampling density in regions with large curvature changes; Vertex attribute labeling: Spatial coordinates (x,y,z), normal vector (nx,ny,nz), curvature value, and region identifier.

[0048] The gingival base region is constructed based on multiple vertices of the dental model. The surface construction method is Coons surface interpolation. Boundary conditions are achieved by using the vertices from step 3 as boundary control points. Internal point generation is achieved through radial basis function interpolation based on boundary points. Mesh generation is achieved through Delaunay triangulation. Mesh optimization is performed by edge flipping to improve triangle quality. Surface continuity is ensured by maintaining G0 continuity with the lateral surface region. Local smoothing is achieved by applying Laplacian smoothing.

[0049] It can construct anatomically accurate and geometrically continuous gingival floor regions. Especially when dealing with complex dental models, it ensures the accuracy and computational efficiency of the model through adaptive sampling and advanced surface interpolation techniques. The generated floor region model can not only accurately reflect the contact relationship between the gingiva and alveolar bone, but also provide a key geometric basis for the subsequent overall construction of the gingival model.

[0050] In the real-time example of this application, the bottom surface region of the gingiva is constructed as follows: Calculate the vertices of the jaw model surrounded by the gingival boundary, and translate each vertex along its normal direction by a distance of gingivaThickness / 2 (half the gingival thickness) to obtain the vertex set of the bottom region of the gingiva. Combine this with the following boundary recognition algorithm and vertex selection strategy: Boundary recognition algorithm: Calculate the average curvature of each vertex, and the region with curvature > 0.05 mm is regarded as the gingival boundary; start from the high curvature point and expand to form a continuous boundary; Boundary smoothing: Apply moving least squares (MLS) smoothing, smoothing factor: 0.3; Boundary closure processing: Detect the opening boundary and close it automatically, closure error: < 0.1 mm; Vertex selection strategy: Uniform sampling: Select a vertex every 1mm along the boundary line; Feature point preservation: Force the preservation of curvature maxima; Density adaptation: Increase sampling density in areas with large curvature changes; Vertex attribute labeling: Spatial coordinates (x,y,z), normal vector (nx,ny,nz), curvature value, and region identifier; Construct the bottom surface region of the gingiva based on multiple vertices of the dental model, using the following surface construction method: Coons surface interpolation; Boundary conditions: Use the vertices from step 3 as boundary control points; Internal point generation: Radial basis function interpolation based on boundary points; Mesh generation: Triangulation method: Delaunay triangulation; Mesh optimization: Edge flipping optimization to improve triangle quality; Surface continuity guarantee: Maintain G0 continuity with the lateral surface region; Local smoothing: Apply Laplacian smoothing.

[0051] It can construct anatomically accurate and geometrically continuous gingival floor regions. Especially when dealing with complex dental models, it ensures the accuracy and computational efficiency of the model through adaptive sampling and advanced surface interpolation techniques. The generated floor region model can not only accurately reflect the contact relationship between the gingiva and alveolar bone, but also provide a key geometric basis for the subsequent overall construction of the gingival model.

[0052] refer to Figure 4 In step S13, the specific steps are as follows: S131: Synthesize the upper surface region, upper palatal region, lateral region, and bottom region to form a complete initial set of gingival vertices, which contains all the key points that constitute the shape of the gingiva; S132: Transform the initial set of gingival vertices into an initial mesh model, which consists of multiple triangles or polygons to represent the geometry of the gingiva.

[0053] In the embodiments of this application, before proceeding to S131, four regions have been constructed respectively through the following steps: Upper surface region (S121): including buccal, central, lingual curves and alveolar bone boundary curves, forming the geometry of the upper surface of the gingiva; Maxillary region (S122): composed of lingual boundary curve segments and alveolar bone boundary curves, generated by UV spline surface; Lateral region (S123): constructed based on the transition curve of the lateral contour and multiple lateral vertices; Bottom region (S124): constructed based on the gingival boundary line identified by the dental model and multiple dental model vertices. Each of these regions contains an independent set of vertices, and there may be overlap or gaps at the junction.

[0054] To combine the four regions into a single initial set of gingival vertices, the following method is used: The vertex deduplication and merging method uses a spatial hash table for vertex deduplication. The principle is to map all vertices to the hash table according to their spatial positions. A merging threshold is set (e.g., 0.05mm). If the distance between two vertices is less than this threshold, they are considered the same vertex and merged. The actual operation involves: traversing all vertices and mapping their coordinates to the hash table; comparing the distances of vertices within each hash bucket and merging duplicate vertices; and updating the indexes of adjacent regions to ensure that the merged vertices can be correctly connected.

[0055] Boundary alignment method: At the boundary of adjacent regions (such as the top surface and the side surface), the nearest neighbor search algorithm is used to align the boundary points; Principle: Match the vertices at the boundary to ensure that the boundary points of adjacent regions are consistent in spatial position; If there are gaps, new vertices are generated by interpolation (such as linear interpolation) to fill the gaps; Practical operation: Extract the boundary points of the top surface and the side surface; Calculate the nearest neighbor distance between the boundary points; For point pairs with a distance greater than a threshold, new vertices are generated by interpolation.

[0056] Normal consistency check method: After merging vertices, perform a consistency check on the normal vectors of adjacent triangles; Principle: Ensure that the normal vectors of adjacent triangles are in the same direction to avoid normal flipping; If inconsistency in normal vectors is found, adjust the vertex order or regenerate the triangles; Practical operation: Traverse all triangles and calculate their normal vectors; Check the angle between the normal vectors of adjacent triangles. If the angle is greater than a threshold (such as 90 degrees), adjust the vertex order.

[0057] By deduplicating vertices, aligning boundaries, and checking normal consistency, the four regions are combined into a complete initial set of gingival vertices. This method ensures the geometric accuracy and continuity of the model, laying a solid foundation for subsequent mesh generation (S132). In practical applications, this method can efficiently handle complex gingival morphology and adapt to individual differences among different patients.

[0058] Furthermore, the initial set of gingival vertices is transformed into an initial mesh model, which consists of multiple triangles or polygons to represent the geometry of the gingiva, referencing the geometry of the gingiva; At this point, the input is: initial set of gingival vertices (from S131); file format: initialvertices.obj; number of vertices: 8,312; vertex attributes: 3D coordinates (x, y, z); normal vector (nx, ny, nz); boundary markers (indicating whether a vertex is located on a boundary).

[0059] To transform the initial gingival vertex set into an initial mesh model, the following method is used: The Delaunay triangulation method: An initial mesh is generated using the 3D Delaunay triangulation algorithm. Principle: Delaunay triangulation is a method of connecting point sets into a triangular mesh, ensuring that the circumcircle of all triangles does not contain other vertices. In 3D space, this method can generate tetrahedral meshes, but by projecting onto a 2D plane or restricting the triangulation range, triangular meshes can be generated. Practical operation: Project the vertex set onto a 2D plane (such as the xy-plane); perform Delaunay triangulation on the projected vertices; map the triangulation result back to 3D space to generate the initial triangular mesh.

[0060] Mesh optimization methods: Edge flipping and edge collapsing are used to optimize mesh quality; Principle: Edge flipping: Improves the geometric quality of the mesh by swapping the connection of triangle edges (e.g., reducing narrow triangles); Edge collapsing: Merges short edges into a single vertex, reducing mesh complexity; Practical operation: Traverse all edges, calculate their length and the included angle of adjacent triangles; Collapsing edges with a length less than 0.1mm; Flipping edges of triangles with included angles less than 30° or greater than 150°.

[0061] Boundary handling method: Repair the mesh boundary to ensure the model's closure; Principle: Boundary vertices may have gaps or overlaps due to triangulation or optimization operations; Repair the boundary by detecting boundary edges (edges belonging to only one triangle) and filling in the missing triangles; Actual operation: Traverse all edges and mark the boundary edges; Fill the boundary edges to generate new triangles; Check whether the filled mesh is closed.

[0062] By employing Delaunay triangulation, mesh optimization, and boundary treatment, the initial gingival vertex set is transformed into a high-quality initial mesh model. This method ensures the geometric accuracy and mesh quality of the model, laying a solid foundation for subsequent aesthetic feature addition and model optimization. In practical applications, this method can efficiently handle complex gingival morphologies and adapt to individual differences among different patients. Please see Figure 7 , Figure 7 This is a schematic diagram of the structural composition of the initial mesh model generation system for the gingiva in an embodiment of the present invention; the initial mesh model generation system for the gingiva includes: Acquisition module 21 is used to acquire data from different boundary regions of the gingiva; Region module 22 is used to determine the upper surface region, maxillary region, lateral region and underside region based on the different boundary regions of the gingiva; Initial mesh model module 23 is used to determine the initial mesh model of the gingiva based on the synthesis of the upper surface region, the maxillary region, the lateral region, and the bottom region.

[0063] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

Claims

1. A method for generating an initial mesh model of the gingiva, characterized in that, include: Collect samples from different boundary areas of the gingiva; The upper surface region, upper palatal region, lateral region, and underside region are determined based on the different boundary areas of the gingiva; The initial mesh model of the gingiva is determined by synthesizing the upper surface region, the maxillary region, the lateral region, and the bottom region.

2. The generation of the initial mesh model of the gingiva according to claim 1, characterized in that, The different boundary regions of the gingiva collected include: Collect multiple image data of the gingiva, determine the equidistant line of the gingival boundary based on the recognition of multiple image data of the gingiva, and output the set of points on the equidistant line; The dentition region is determined based on multiple images of the gingiva, the dentition type features are determined based on the detection of the dentition region, and the gingival width distribution of the dentition type features is marked. The key point segments are determined based on the division of the upper equidistant line point set, and the buccal, lingual, and two-end boundary regions are divided according to the key point segments and the gingival width distribution of the dental and jaw type characteristics.

3. The generation of the initial mesh model of the gingiva according to claim 1, characterized in that, The determination of the upper surface region, upper palatal region, lateral region, and underside region based on different gingival boundary areas includes: Multiple curves are constructed based on the division of the gingival boundary region, representing the buccal, central, and lingual positions of the teeth. The boundary curves of the alveolar bone morphology are determined based on the multiple curves related to the alveolar bone. The corresponding upper surface regions are constructed based on the multiple curves representing the buccal, central, and lingual positions of the teeth, the boundary curves of the alveolar bone morphology, and the buccal gingival boundary line.

4. The generation of the initial mesh model of the gingiva according to claim 3, characterized in that, The method of determining the upper surface region, maxillary region, lateral region, and underside region based on different gingival boundary regions also includes: The tongue-side boundary curve segment is determined based on the curve of the tongue-side position, and multiple sampling points are marked. The geometric framework of the maxillary region is constructed based on the multiple sampling points. The UV spline surface is constructed based on the geometric framework of the maxillary region and the curve of the tongue-side position to output the corresponding maxillary region.

5. The generation of the initial mesh model of the gingiva according to claim 1, characterized in that, The method of determining the upper surface region, maxillary region, lateral region, and underside region based on different gingival boundary regions also includes: The transition curve of the side profile is constructed based on the set of points on the upper equidistant line, and multiple side vertices are determined based on the identification of the transition curve of the side profile. The side region is constructed based on the multiple side vertices.

6. The generation of the initial mesh model of the gingiva according to claim 1, characterized in that, The method of determining the upper surface region, maxillary region, lateral region, and underside region based on different gingival boundary regions also includes: A dental model is collected, the gingival boundary is determined based on the identification of the dental model, and multiple vertices of the dental model are marked. The bottom surface region of the gingiva is constructed based on the multiple vertices of the dental model.

7. The generation of the initial mesh model of the gingiva according to claim 1, characterized in that, The initial mesh model of the gingiva, determined by synthesizing the upper surface region, maxillary region, lateral region, and bottom surface region, includes: The upper surface region, upper palatal region, lateral region, and bottom region are synthesized to form a complete initial set of gingival vertices, which contains all the key points that constitute the shape of the gingiva.

8. The generation of the initial mesh model of the gingiva according to claim 7, characterized in that, The initial mesh model of the gingiva determined based on the synthesis of the upper surface region, the maxillary region, the lateral surface region, and the bottom surface region also includes: Transform the initial set of gingival vertices into an initial mesh model.

9. The generation of the initial mesh model of the gingiva according to claim 8, characterized in that, The initial mesh model of the gingiva determined based on the synthesis of the upper surface region, the maxillary region, the lateral surface region, and the bottom surface region also includes: The initial mesh model consists of multiple triangles or polygons to represent the geometry of the gums.

10. A system for generating an initial mesh model of the gingiva, characterized in that, The system for generating the initial mesh model of the gingiva is applied to the generation of the initial mesh model of the gingiva as described in any one of claims 1-9, wherein the system for generating the initial mesh model of the gingiva comprises: The acquisition module is used to acquire data from different boundary areas of the gingiva; The region module is used to determine the upper surface region, the maxillary region, the lateral region, and the underside region based on the different boundary regions of the gingiva; The initial mesh model module is used to determine the initial mesh model of the gingiva based on the synthesis of the upper surface region, maxillary region, lateral region, and bottom region.

Images