A method for determining the occlusal relationship of upper and lower dentitions on a digital intraoral scan model
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHANGHAI JIAOTONG UNIV
- Filing Date
- 2023-08-28
- Publication Date
- 2026-07-21
Smart Images
Figure CN117137660B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of digital medical technology, and in particular to a method for determining the occlusal relationship of the upper and lower teeth in a digital intraoral scanning model. Background Technology
[0002] Orthodontics focuses on correcting dental and maxillofacial misalignments. It involves diagnosing, preventing, and treating various abnormalities in the dental and maxillofacial structures to improve a patient's occlusal function, dental aesthetics, and oral health. It is primarily suitable for cases where teeth are misaligned but the jawbone is normal. Common occlusal relationships include normal occlusion, deep overbite, open bite, and cross bite. During orthodontic treatment, dentists typically examine the patient's occlusal relationship using clinical examinations, dental or oral impressions, and X-rays, and develop appropriate treatment plans based on the specific situation. Intraoral scanners are advanced digital technologies that can acquire three-dimensional model data of the patient's oral cavity. They can provide detailed information on the dental and maxillofacial structures; however, the model data from an intraoral scanner alone often cannot directly determine the complete occlusal relationship. Intraoral flare photographs can show the detailed structure of teeth and jawbones, including problems such as crowding, misalignment, and missing teeth. Dentists can use these photographs to assess the patient's oral health, tooth position, and occlusal relationship to determine whether orthodontic treatment is necessary and the appropriate treatment method. Therefore, there is an urgent need to propose a method that combines the occlusal information of teeth in intraoral malocclusion photographs with the digital upper and lower dentition models obtained by intraoral scanners to determine the occlusal relationship of the digital upper and lower dentition models in three-dimensional space, so as to facilitate dentists in diagnosis, analysis and subsequent treatment. Summary of the Invention
[0003] The purpose of this invention is to provide a method for determining the occlusal relationship of the upper and lower dentition in a digital intraoral scanning model. By using the occlusal information from three intraoral flare photographs taken at different angles, the occlusal relationship of the digital three-dimensional model of the upper and lower dentition obtained by the intraoral scanner is determined, thus assisting orthodontists in judging the patient's occlusal condition.
[0004] The objective of this invention can be achieved through the following technical solutions:
[0005] A method for determining the occlusal relationship of the upper and lower teeth in a digital intraoral scanning model includes the following steps:
[0006] Step 1) Obtain three intraoral dilation photos from different angles: front, left, and right.
[0007] Step 2) Use a deep learning-based tooth semantic segmentation model to perform tooth semantic segmentation on intraoral flaring photos from different angles, and extract tooth contour segmentation maps with tooth number information.
[0008] Step 3) Obtain the upper and lower dentition triangular patch files of the digital intraoral scan model with tooth number information in two different coordinate systems;
[0009] Step 4) Initialize camera parameters, relative position of upper and lower teeth rows, and orientation parameters;
[0010] Step 5) Based on the current camera parameters, the relative position and orientation parameters of the upper and lower teeth, and the standard pinhole camera model, the upper and lower teeth models obtained by digital intraoral scanning are projected according to intraoral swollen photos from different angles to extract the visible tooth outline projection map with tooth number information.
[0011] Step 6) Match the corresponding points based on the tooth contour segmentation map and the tooth contour projection map, define the loss function, and iteratively optimize the camera parameters, the relative position of the upper and lower teeth, and the orientation parameters by calculating the loss function value of the matching results. Repeat steps 5)-6) until convergence and the optimal solution is obtained.
[0012] Step 7) Determine the occlusal relationship of the upper and lower teeth based on the calculated optimal relative position and orientation parameters of the upper and lower teeth, and convert the triangular patches of the upper and lower teeth in different coordinate systems in the digital intraoral scanning model to the same coordinate system based on the optimal relative position and orientation parameters of the upper and lower teeth, and generate the corresponding files.
[0013] Step 2) includes the following steps:
[0014] Step 2-1) Construct a tooth semantic segmentation model based on the encoder and decoder structure of U-Net3+ in deep learning, the pooling pyramid module of multi-scale dilated spatial convolution, and the dual-branch structure of multi-task learning. Take an intraoral flaring photo as input and output a tooth semantic segmentation map.
[0015] Step 2-2) Use a post-processing algorithm to adjust the output tooth semantic segmentation map and number the teeth;
[0016] Steps 2-3) Based on the relative area relationship of the upper and lower dentition tooth regions, determine the extraction order of visible tooth contours, classify the extracted tooth contours according to their tooth numbers, and obtain a tooth contour segmentation map with tooth number information.
[0017] The network structure of the tooth semantic segmentation model is that the output of a standard U-Net3+ image encoder is simultaneously input to a standard U-Net3+ tooth semantic segmentation decoder and a standard U-Net3+ tooth binary contour segmentation decoder. The outputs of the two decoders are stacked and input to a region-contour fusion module based on a multi-scale dilated spatial convolution pooling pyramid. The region-contour fusion module consists of three convolutional layers connected in series with a multi-scale dilated convolutional module, and then three more convolutional layers connected in series.
[0018] The post-processing algorithm performs the following operations on the tooth semantic segmentation map: determine the connected regions of the tooth semantic segmentation map, unify the tooth numbers within the same connected region, extract the largest connected region of tooth numbers at different tooth positions, adjust the tooth numbers of the connected regions in a specific order, and ensure the uniqueness of the connected regions under the same tooth number, and use a morphological algorithm to smooth the results.
[0019] The specific order for extracting visible tooth contours based on the relative area relationship between the upper and lower dentition tooth regions is as follows:
[0020] If the segmented area of the upper dentition teeth is larger than that of the lower dentition teeth, the visible contours of the upper dentition teeth are extracted first; otherwise, the contours of the lower dentition teeth are extracted first. When extracting the contours, the order is from the middle to the left and right sides, ignoring the contours of the occluded teeth.
[0021] The camera parameters include internal and external parameters. The internal parameters include the camera's focal length, principal point coordinates, and the physical dimensions of the pixels on the horizontal and vertical axes. The external parameters include the camera's position and orientation in the world coordinate system.
[0022] The matching relationship for corresponding point matching based on the tooth contour segmentation map and the tooth contour projection map is calculated by the following formula:
[0023]
[0024] Among them, c i τ This represents the coordinates of the i-th point of the tooth contour line of tooth position τ extracted from the image in step 2) in the pixel coordinate system. This represents the coordinates of the i-th point in the pixel coordinate system of the visible contour line of the corresponding tooth position τ obtained by projection in step 5). n i τ This represents the plane normal vector in the pixel coordinate system of the i-th point of the tooth contour line of tooth position τ extracted from the photo in step 2). This represents the plane normal vector in the pixel coordinate system of the i-th point of the visible contour line of the corresponding tooth position τ obtained by projection in step 5). Let σ represent the square of the vector's 2-norm, where σ is an adjustable hyperparameter.
[0025] The loss function includes contour-corresponding point distance matching loss and contour-corresponding point normal vector matching loss. The total loss function is the sum of the loss functions corresponding to photos from different angles. The loss function to be optimized for each image is... Represented as:
[0026]
[0027] Among them, L p L is the loss function for matching the distance between corresponding points on the contour. n The loss function for matching the normal vectors of corresponding points on the contour is λ. n For hyperparameters,
[0028] The contour corresponding point distance matching loss function L p Represented as:
[0029]
[0030] Where N is the total number of tooth contour points segmented from the photograph, T is the number of tooth categories segmented from the photograph, τ represents the tooth number, and n τ c represents the number of contour points of the τth tooth. i τ This represents the coordinates of the i-th point of the tooth contour line of tooth position τ extracted from the image in step 2) in the pixel coordinate system. This represents the coordinates of the i-th point in the pixel coordinate system of the visible contour line of the corresponding tooth position τ obtained by projection in step 5).
[0031] The loss function L for matching the normal vectors of the contour points n Represented as:
[0032]
[0033] Where N is the total number of tooth contour points segmented from the photograph, T is the number of tooth categories segmented from the photograph, τ represents the tooth number, and n τ c represents the number of contour points of the τth tooth. i τ This represents the coordinates of the i-th point of the tooth contour line of tooth position τ extracted from the image in step 2) in the pixel coordinate system. This represents the coordinates of the i-th point in the pixel coordinate system of the visible contour line of the corresponding tooth position τ obtained by projection in step 5). This represents the plane normal vector in the pixel coordinate system of the i-th point of the visible contour line of the corresponding tooth position τ obtained by projection in step 5). <·,·> represents the vector dot product operation.
[0034] Compared with the prior art, the present invention has the following beneficial effects:
[0035] (1) This invention uses a deep learning-based tooth semantic segmentation model to perform tooth semantic segmentation on intraoral flaring photos from different angles, and accurately extracts tooth contour segmentation maps with tooth number information.
[0036] (2) This invention projects the upper and lower dentition of the digital intraoral scanning model onto intraoral swirling photographs from different angles, extracts visible tooth contour projections with tooth number information, and iteratively optimizes camera parameters and relative position parameters of the upper and lower dentition based on the defined loss function between contours until the optimal solution is obtained. This achieves the calculation of the optimal relative position parameters of the upper and lower dentition and transforms the triangular facets of the upper and lower dentition in different coordinate systems in the digital model to the same coordinate system to predict the occlusal relationship of the upper and lower dentition. This allows for the intuitive determination of the occlusal relationship of teeth from the digital model, assisting dentists in making decisions. Attached Figure Description
[0037] Figure 1 This is a flowchart of the method of the present invention;
[0038] Figure 2 This is a schematic diagram of the network structure of the tooth semantic segmentation model of the present invention. Detailed Implementation
[0039] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.
[0040] This embodiment provides a method for determining the occlusal relationship of the upper and lower dentition in a digital intraoral scanning model. The input consists of preprocessed triangular facet files of the upper and lower dentition from a patient's intraoral scanning model, obtained from two different coordinate systems and containing tooth identification information, along with three intraoral flare photographs from different angles (front, left, and right). The output consists of triangular facet files of the upper and lower dentition from the digital intraoral scanning model, within the same coordinate system, and the relative positional relationship of the upper and lower dentition, i.e., the occlusal relationship. Specifically, as shown... Figure 1 As shown, it includes the following steps:
[0041] Step 1) Obtain three intraoral diaphragm photos from different angles: front, left, and right.
[0042] Step 2) Use a deep learning-based tooth semantic segmentation model to perform tooth semantic segmentation on the three intraoral flare photos taken from different angles, and extract the tooth contour segmentation map with tooth number information.
[0043] Step 2-1) Construct a tooth semantic segmentation model based on the encoder and decoder structure of U-Net3+ in deep learning, the pooling pyramid module of multi-scale dilated spatial convolution, and the dual-branch structure of multi-task learning. Take three intraoral flaring photos of the case from different angles as input and output the corresponding three tooth semantic segmentation images.
[0044] like Figure 2 As shown, the network structure of the tooth semantic segmentation model used in this embodiment is that the output of a standard U-Net3+ image encoder is simultaneously input to a standard U-Net3+ tooth semantic segmentation decoder and a standard U-Net3+ tooth binary contour segmentation decoder. The outputs of the two decoders are stacked and input to a region-contour fusion module based on a multi-scale dilated spatial convolution pooling pyramid. The region-contour fusion module consists of three convolutional layers connected in series with a multi-scale dilated convolutional module, and then three more convolutional layers connected in series. Its output is a tooth semantic segmentation map.
[0045] This embodiment trains the tooth semantic segmentation model on a training set of approximately 15,000 images, saves the model that performs best on the test set, and saves the model's structure and parameters as a local file for easy access next time. The input image size of the tooth semantic segmentation model is (256, 256, 3), and the output tooth semantic segmentation map size is (256, 256, 33). The tooth semantic segmentation map uses one-hot encoding and has a total of 33 categories, including 1 background class and 32 tooth categories. The convolutional layers in the model have a 3*3 kernel size and use the ReLU function as the activation function for each layer. The training and evaluation process is implemented using the PyTorch framework.
[0046] Step 2-2) Use a post-processing algorithm to adjust the three output tooth semantic segmentation images and number the teeth respectively;
[0047] Specifically, the post-processing algorithm performs the following operations on the tooth semantic segmentation map: It determines the connected regions of the tooth semantic segmentation map; it unifies the tooth numbers within a connected region based on the tooth number with the most pixels; it extracts the largest connected region for tooth numbers at different tooth positions; it adjusts the tooth numbers within the connected regions in a specific order, ensuring the uniqueness of connected regions under the same tooth number; it modifies the tooth numbers in redundant connected regions; and it smooths the result using morphological algorithms such as erosion and dilation. In this embodiment, the post-processing algorithm is implemented based on the OpenCV algorithm library.
[0048] Steps 2-3) Based on the relative area relationship of the upper and lower dentition tooth regions, determine the extraction order of visible tooth contours. If the segmented area of the upper dentition tooth region is larger than that of the lower dentition, the visible contours of the upper dentition teeth are extracted first, and vice versa. When extracting contours, follow the order from the middle to the left and right sides, ignoring the occluded tooth contours. The extracted tooth contours are classified according to their tooth numbers to obtain a tooth contour segmentation map with tooth number information.
[0049] Step 3) Obtain the upper and lower dental triangular patch files of the digital intraoral scan model with tooth number information in two different coordinate systems.
[0050] Step 4) Initialize camera parameters, relative position of upper and lower teeth rows, and orientation parameters.
[0051] For intraoral slit photographs taken from different angles, empirical values were used to initialize the camera's internal and external parameters. The internal parameters included the camera's focal length, principal point coordinates, and the physical dimensions of the pixels on the horizontal and vertical axes. The external parameters included the camera's position and orientation in the world coordinate system. The relative position and orientation parameters of the upper and lower dentition were initialized to ensure that the 3D model presented a standard normal occlusion. Specifically, the initial relative rotation vector of the upper and lower dentition was [0,0,0], and the relative position relationship was that the upper dentition was located 7mm above and 2mm in front of the lower dentition, with a left-right offset of 0.
[0052] Step 5) Based on the current camera parameters, the relative position and orientation parameters of the upper and lower teeth, and the standard pinhole camera model, the vertices of the upper and lower teeth models obtained by the preprocessed digital intraoral scan with tooth number information are projected according to intraoral flaring photos from different angles, and the visible edge contour points of each tooth are extracted to obtain a visible tooth contour projection map with tooth number information.
[0053] Step 6) Match the corresponding points based on the tooth contour segmentation map and the tooth contour projection map, define the loss function, and iteratively optimize the camera parameters, the relative position of the upper and lower teeth and the orientation parameters by calculating the loss function value of the matching results. Repeat steps 5)-6) until convergence and the optimal solution is obtained.
[0054] Based on the tooth contour points in the tooth contour segmentation map and the tooth contour projection map, the corresponding point relationships are matched. The matching relationship is calculated by the following formula:
[0055]
[0056] Among them, c i τ This represents the coordinates of the i-th point of the tooth contour line of tooth position τ extracted from the image in step 2) in the pixel coordinate system. This represents the coordinates of the i-th point in the pixel coordinate system of the visible contour line of the corresponding tooth position τ obtained by projection in step 5). n i τ This represents the plane normal vector in the pixel coordinate system of the i-th point of the tooth contour line of tooth position τ extracted from the photo in step 2). This represents the plane normal vector in the pixel coordinate system of the i-th point of the visible contour line of the corresponding tooth position τ obtained by projection in step 5). Let σ represent the square of the vector's 2-norm, where σ is an adjustable hyperparameter.
[0057] In this embodiment, the loss function includes contour-corresponding point distance matching loss and contour-corresponding point normal vector matching loss. The total loss function is the sum of the loss functions corresponding to photos from different angles. The loss function to be optimized for each image is... Represented as:
[0058]
[0059] Among them, L p L is the loss function for matching the distance between corresponding points on the contour. n The loss function for matching the normal vectors of corresponding points on the contour is λ. n For hyperparameters, In this embodiment, λ n The value is 0.05.
[0060] Specifically, the loss function L for matching the distance between corresponding points of the contour. p Represented as:
[0061]
[0062] Where N is the total number of tooth contour points segmented from the photograph, T is the number of tooth categories segmented from the photograph, τ represents the tooth number, and n τ c represents the number of contour points of the τth tooth. i τ This represents the coordinates of the i-th point of the tooth contour line of tooth position τ extracted from the image in step 2) in the pixel coordinate system. This represents the coordinates of the i-th point in the pixel coordinate system of the visible contour line of the corresponding tooth position τ obtained by projection in step 5).
[0063] Contour Corresponding Point Normal Vector Matching Loss Function L n Represented as:
[0064]
[0065] Where N is the total number of tooth contour points segmented from the photograph, T is the number of tooth categories segmented from the photograph, τ represents the tooth number, and n τ c represents the number of contour points of the τth tooth. i τ This represents the coordinates of the i-th point of the tooth contour line of tooth position τ extracted from the image in step 2) in the pixel coordinate system. This represents the coordinates of the i-th point in the pixel coordinate system of the visible contour line of the corresponding tooth position τ obtained by projection in step 5). This represents the plane normal vector in the pixel coordinate system of the i-th point of the visible contour line of the corresponding tooth position τ obtained by projection in step 5). <·,·> represents the vector dot product operation.
[0066] Steps 5) and 6) are iterated multiple times until the loss function in step 6) converges and its value stabilizes, thus obtaining the optimized camera parameters and the relative positions and orientation parameters of the upper and lower teeth. Specifically, the camera parameters and the relative positions of the upper and lower teeth are iterated for 15 rounds. The gradient of the optimization function is determined by explicit derivation, and the sequential least squares method is used as the optimization algorithm.
[0067] Step 7) Determine the occlusal relationship of the upper and lower teeth based on the calculated optimal relative position and orientation parameters of the upper and lower teeth, and convert the triangular patches of the upper and lower teeth in different coordinate systems in the digital intraoral scanning model to the same coordinate system based on the optimal relative position and orientation parameters of the upper and lower teeth, and generate the corresponding files.
[0068] The specific operation involves transforming the lower dental row triangular facet model into the coordinate system of the upper dental row based on the position and orientation relationship of the upper and lower dental rows obtained in step 6), and then merging the triangular facets of the two models into one file to generate an obj format file as the final output.
[0069] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.
Claims
1. A method for determining the occlusal relationship of the upper and lower teeth in a digital intraoral scanning model, characterized in that, Includes the following steps: Step 1) Obtain three intraoral dilation photos from different angles: front, left, and right. Step 2) Use a deep learning-based tooth semantic segmentation model to perform tooth semantic segmentation on intraoral flaring photos from different angles, and extract tooth contour segmentation maps with tooth number information. Step 3) Obtain the upper and lower dentition triangular patch files of the digital intraoral scan model with tooth number information in two different coordinate systems; Step 4) Initialize camera parameters, relative position of upper and lower teeth rows, and orientation parameters; Step 5) Based on the current camera parameters, the relative position and orientation parameters of the upper and lower teeth, and the standard pinhole camera model, the upper and lower teeth models obtained by digital intraoral scanning are projected according to intraoral swollen photos from different angles to extract the visible tooth outline projection map with tooth number information. Step 6) Match the corresponding points based on the tooth contour segmentation map and the tooth contour projection map, define the loss function, and iteratively optimize the camera parameters, the relative position of the upper and lower teeth, and the orientation parameters by calculating the loss function value of the matching results. Repeat steps 5)-6) until convergence and the optimal solution is obtained. Step 7) Determine the occlusal relationship of the upper and lower teeth based on the calculated optimal relative position and orientation parameters of the upper and lower teeth, and convert the triangular patches of the upper and lower teeth in different coordinate systems in the digital intraoral scanning model to the same coordinate system based on the optimal relative position and orientation parameters of the upper and lower teeth, and generate the corresponding files.
2. The method for determining the occlusal relationship of the upper and lower teeth in a digital intraoral scanning model according to claim 1, characterized in that, Step 2) includes the following steps: Step 2-1) Construct a tooth semantic segmentation model based on the encoder and decoder structure of U-Net3+ in deep learning, the pooling pyramid module of multi-scale dilated spatial convolution, and the dual-branch structure of multi-task learning. Take an intraoral flaring photo as input and output a tooth semantic segmentation map. Step 2-2) Use a post-processing algorithm to adjust the output tooth semantic segmentation map and number the teeth; Steps 2-3) Based on the relative area relationship of the upper and lower dentition tooth regions, determine the extraction order of visible tooth contours, classify the extracted tooth contours according to their tooth numbers, and obtain a tooth contour segmentation map with tooth number information.
3. The method for determining the occlusal relationship of the upper and lower teeth in a digital intraoral scanning model according to claim 2, characterized in that, The network structure of the tooth semantic segmentation model is that the output of a standard U-Net3+ image encoder is simultaneously input to a standard U-Net3+ tooth semantic segmentation decoder and a standard U-Net3+ tooth binary contour segmentation decoder. The outputs of the two decoders are stacked and input to a region-contour fusion module based on a multi-scale dilated spatial convolution pooling pyramid. The region-contour fusion module consists of three convolutional layers connected in series with a multi-scale dilated convolutional module, and then three more convolutional layers connected in series.
4. The method for determining the occlusal relationship of the upper and lower teeth in a digital intraoral scanning model according to claim 2, characterized in that, The post-processing algorithm performs the following operations on the tooth semantic segmentation map: determine the connected regions of the tooth semantic segmentation map, unify the tooth numbers within the same connected region, extract the largest connected region of tooth numbers at different tooth positions, adjust the tooth numbers of the connected regions in a specific order, and ensure the uniqueness of the connected regions under the same tooth number, and use a morphological algorithm to smooth the results.
5. The method for determining the occlusal relationship of the upper and lower teeth in a digital intraoral scanning model according to claim 2, characterized in that, The specific order for extracting visible tooth contours based on the relative area relationship between the upper and lower dentition tooth regions is as follows: If the segmented area of the upper dentition teeth is larger than that of the lower dentition teeth, the visible contours of the upper dentition teeth are extracted first; otherwise, the contours of the lower dentition teeth are extracted first. When extracting the contours, the order is from the middle to the left and right sides, ignoring the contours of the occluded teeth.
6. The method for determining the occlusal relationship of the upper and lower teeth in a digital intraoral scanning model according to claim 1, characterized in that, The camera parameters include internal and external parameters. The internal parameters include the camera's focal length, principal point coordinates, and the physical dimensions of the pixels on the horizontal and vertical axes. The external parameters include the camera's position and orientation in the world coordinate system.
7. The method for determining the occlusal relationship of the upper and lower teeth in a digital intraoral scanning model according to claim 1, characterized in that, The matching relationship for corresponding point matching based on the tooth contour segmentation map and the tooth contour projection map is calculated by the following formula: Among them, c i τ This represents the coordinates of the i-th point of the tooth contour line of tooth position τ extracted from the image in step 2) in the pixel coordinate system. This represents the coordinates of the i-th point in the pixel coordinate system of the visible contour line of the corresponding tooth position τ obtained by projection in step 5). n i τ This represents the plane normal vector in the pixel coordinate system of the i-th point of the tooth contour line of tooth position τ extracted from the photo in step 2). This represents the plane normal vector in the pixel coordinate system of the i-th point of the visible contour line of the corresponding tooth position τ obtained by projection in step 5). Let σ represent the square of the vector's 2-norm, where σ is an adjustable hyperparameter.
8. The method for determining the occlusal relationship of the upper and lower teeth in a digital intraoral scanning model according to claim 1, characterized in that, The loss function includes contour-corresponding point distance matching loss and contour-corresponding point normal vector matching loss. The total loss function is the sum of the loss functions corresponding to photos from different angles. The loss function to be optimized for each image is... Represented as: Among them, L p L is the loss function for matching the distance between corresponding points on the contour. n The loss function for matching the normal vectors of corresponding points on the contour is λ. n For hyperparameters, 9. The method for determining the occlusal relationship of the upper and lower teeth in a digital intraoral scanning model according to claim 8, characterized in that, The contour corresponding point distance matching loss function L p Represented as: Where N is the total number of tooth contour points segmented from the photograph, T is the number of tooth categories segmented from the photograph, τ represents the tooth number, and n τ c represents the number of contour points of the τth tooth. i τ This represents the coordinates of the i-th point of the tooth contour line of tooth position τ extracted from the image in step 2) in the pixel coordinate system. This represents the coordinates of the i-th point in the pixel coordinate system of the visible contour line of the corresponding tooth position τ obtained by projection in step 5).
10. The method for determining the occlusal relationship of the upper and lower teeth in a digital intraoral scanning model according to claim 8, characterized in that, The loss function L for matching the normal vectors of the contour points n Represented as: Where N is the total number of tooth contour points segmented from the photograph, T is the number of tooth categories segmented from the photograph, τ represents the tooth number, and n τ c represents the number of contour points of the τth tooth. i τ This represents the coordinates of the i-th point of the tooth contour line of tooth position τ extracted from the image in step 2) in the pixel coordinate system. This represents the coordinates of the i-th point in the pixel coordinate system of the visible contour line of the corresponding tooth position τ obtained by projection in step 5). This represents the plane normal vector in the pixel coordinate system of the i-th point of the visible contour line of the corresponding tooth position τ obtained by projection in step 5). <·,·> represents the vector dot product operation.