Tooth data model reconstruction method and device, electronic equipment and storage medium

Abstract

The invention discloses a tooth data model reconstruction method and device, electronic equipment and a storage medium. Comprises: acquiring image information of a target tooth; performing segmentation processing on the image information of the target tooth to obtain an image segmentation result of the target tooth; constructing a three-dimensional shape model of the target tooth based on an image segmentation result; performing boundary extraction on the three-dimensional shape model of the target tooth to obtain binary contour information of the target tooth; and constructing a three-dimensional mesh model of the target tooth based on the binary contour information of the target tooth. According to the scheme, the three-dimensional shape model is determined according to the image information of the target tooth, then boundary extraction is performed to obtain the accurate binary contour information of the target tooth, and the three-dimensional mesh model of the target tooth is reconstructed according to the two-dimensional contour information, so that the accuracy and rapidness of reconstructing the three-dimensional mesh model of the target tooth are improved, and the reconstruction efficiency of the three-dimensional mesh model of the target tooth is improved. And an accurate tooth three-dimensional model is provided for generation of a dental crown.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to a tooth data model reconstruction method, device, electronic equipment and storage medium. Background Art

[0002] During the entire treatment process of teeth, while solving problems such as tooth misalignment, tooth unevenness and tooth missing, accurate and comprehensive tooth three-dimensional data is obtained to provide important information basis for the generation of crowns.

[0003] Nowadays, computer-aided design (CAD) has been widely used in the diagnosis, restoration and treatment of dentistry, which solves the problem that traditional methods are time-consuming and labor-intensive due to manual processing and over-reliance on doctors' experience. However, most patients have abnormal teeth such as missing teeth, crowded teeth, and malocclusion, all of which will lead to unclear boundaries of teeth and increase the difficulty of segmentation. Existing methods have problems such as inaccurate tooth segmentation, inaccurate reconstruction of tooth 3D data, and poor accuracy of crown generation in operation. Summary of the invention

[0004] The present invention provides a tooth data model reconstruction method, device, electronic device and storage medium to solve the problems of inaccurate tooth segmentation and inaccurate tooth model reconstruction, as well as inaccurate tooth crown generation.

[0005] According to one aspect of the present invention, a method for reconstructing a tooth data model is provided, comprising:

[0006] Acquire image information of target teeth;

[0007] Segmenting the image information of the target teeth to obtain the image segmentation result of the target teeth;

[0008] Construct a three-dimensional shape model of the target tooth based on the image segmentation result;

[0009] Perform boundary extraction on the three-dimensional shape model of the target tooth to obtain binary contour information of the target tooth;

[0010] A three-dimensional mesh model of the target tooth is constructed based on the binary contour information of the target tooth.

[0011] Optionally, before obtaining the image information of the target tooth, it also includes: obtaining oral three-dimensional environment data of the target object, the oral three-dimensional environment data including three-dimensional data of multiple teeth; preprocessing the oral three-dimensional environment data to obtain a grayscale image corresponding to the oral three-dimensional environment data; traversing the pixel value of each pixel point in the grayscale image, calculating the difference between adjacent pixel points in turn, obtaining a pixel point difference value set of each tooth, and determining the target tooth based on the pixel point difference value set.

[0012] Optionally, the image information of the target teeth is segmented to obtain the image segmentation result of the target teeth, including: calling a pre-trained image segmentation model, segmenting the image information of the target teeth based on the pre-trained image segmentation model, and determining the image segmentation result of the target teeth.

[0013] Optionally, the pre-trained image segmentation model includes multiple convolutional layers, pooling layers, at least one upsampling layer and a feature fusion layer; the image information of the target teeth is segmented based on the pre-trained image segmentation model to determine the image segmentation result of the target teeth, including: inputting the image information into the multiple convolutional layers for convolution processing to obtain an intermediate feature map, wherein the convolution kernel sizes corresponding to the multiple convolutional layers are the same, and the output result of each convolutional layer is used as the input data of the next convolutional layer; transmitting the intermediate feature map to the pooling layer for pooling processing to obtain a first feature map; transmitting the first feature map to the upsampling layer for upsampling processing to obtain a second feature map; transmitting the second feature map and the image information to the feature fusion layer for feature fusion processing to obtain a fused feature map, and using the fused feature map as the image segmentation result of the target teeth.

[0014] Optionally, the method also includes: calling a pre-trained tooth reconstruction model, the pre-trained tooth reconstruction model includes a tooth shape construction sub-model, a tooth boundary extraction sub-model and a tooth fitting sub-model; processing the image segmentation result based on the tooth shape construction sub-model to construct a three-dimensional shape model of the target tooth; performing boundary extraction on the three-dimensional shape model of the target tooth based on the tooth boundary extraction sub-model to obtain binary contour information of the target tooth; processing the binary contour information of the target tooth based on the tooth fitting sub-model to construct a three-dimensional mesh model of the target tooth.

[0015] Optionally, the tooth boundary extraction sub-model includes an Attention-U-net neural network model, wherein the Attention-U-net neural network model includes an encoding module, a decoding module and an attention module, the encoding module includes multiple downsampling convolution blocks and a spatial pyramid pooling layer, the decoding module includes multiple upsampling convolution blocks, the attention module includes multiple attention gates, multiple downsampling convolution blocks, spatial pyramid pooling layers and multiple upsampling convolution blocks are connected in sequence, and each downsampling layer and the upper upsampling layer of the same upsampling layer are connected to the corresponding upsampling layer of the same layer through an attention gate.

[0016] Optionally, the image segmentation model is trained, including: obtaining sample data, dividing the sample data into a training set, a validation set and a test set; initializing the model parameters of the image segmentation model to be trained through an adaptive moment estimation optimizer, inputting the training set into the image segmentation model to be trained for training, determining the loss function value of the image segmentation model to be trained through a joint loss function, and feeding the loss function value back to the adaptive moment estimation optimizer to determine the model parameters of the next training round, wherein the joint loss function is obtained by combining a boundary loss function and a cross entropy loss function; after each round of training, inputting the validation set into the image segmentation model to be trained for verification, and determining the evaluation index data of the output result corresponding to the image segmentation model to be trained, the evaluation index data including the Dice coefficient and the average intersection-over-union ratio; dynamically adjusting the learning rate of the image segmentation model based on the evaluation index data; and when the preset training rounds are completed, ending the training process of the image segmentation model to be trained to obtain a trained image segmentation model.

[0017] According to another aspect of the present invention, there is provided a tooth data model reconstruction device, comprising:

[0018] An image information acquisition module, used to acquire image information of target teeth;

[0019] An image segmentation result determination module is used to perform segmentation processing on the image information of the target teeth to obtain the image segmentation result of the target teeth;

[0020] A three-dimensional shape model determination module, used to construct a three-dimensional shape model of the target tooth based on the image segmentation result;

[0021] A binary contour information determination module is used to extract the boundary of the three-dimensional shape model of the target tooth to obtain the binary contour information of the target tooth;

[0022] The three-dimensional mesh model determination module is used to construct a three-dimensional mesh model of the target tooth based on the binary contour information of the target tooth.

[0023] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0024] at least one processor; and

[0025] a memory communicatively connected to at least one processor; wherein,

[0026] The memory stores a computer program executable by at least one processor, and the computer program is executed by at least one processor so that the at least one processor can execute the tooth data model reconstruction method according to any embodiment of the present invention.

[0027] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the tooth data model reconstruction method of any embodiment of the present invention when executed.

[0028] The technical solution of the embodiment of the present invention obtains the image information of the target tooth; performs segmentation processing on the image information of the target tooth to obtain the image segmentation result of the target tooth; constructs a three-dimensional shape model of the target tooth based on the image segmentation result; performs boundary extraction on the three-dimensional shape model of the target tooth to obtain the binary contour information of the target tooth; and constructs a three-dimensional mesh model of the target tooth based on the binary contour information of the target tooth. This solution achieves the goal of obtaining a three-dimensional shape model of the target tooth by segmenting and reconstructing the image information of the target tooth, and then performs boundary extraction to obtain accurate binary contour information of the target tooth, reconstructs the three-dimensional mesh model of the target tooth based on the two-dimensional contour information, solves the problems of inaccurate tooth segmentation and inaccurate tooth model reconstruction, improves the accuracy and speed of reconstructing the three-dimensional mesh model of the target tooth, provides an accurate three-dimensional tooth model for the generation of the crown, helps to improve the accuracy of the crown generation, and makes the generated crown more consistent with the morphology of the corresponding target tooth, so as to better solve the problem of tooth abnormality.

[0029] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0031] Figure 1 is a flow chart of a tooth data model reconstruction method provided in Embodiment 1 of the present invention;

[0032] Figure 2 is a flow chart of a tooth data model reconstruction method provided in Embodiment 1 of the present invention;

[0033] Figure 3 It is a structural schematic diagram of a tooth data model reconstruction device provided by Embodiment 3 of the present invention;

[0034] Figure 4 It is a schematic diagram of the structure of an electronic device for implementing the tooth data model reconstruction method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0035] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0036] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0037] Embodiment 1

[0038] Figure 1 This is a flow chart of a method for reconstructing a tooth data model provided in the first embodiment of the present invention. This embodiment is applicable to the reconstruction of a tooth data model. The method can be executed by a tooth data model reconstruction device. The tooth data model reconstruction device can be implemented in the form of hardware and / or software. The tooth data model reconstruction device can be configured in a computer and an electronic device provided in the present application. Figure 1 As shown, the method includes:

[0039] S110, acquiring image information of target teeth.

[0040] The target teeth can be specifically understood as teeth to be reconstructed into a three-dimensional model. The target teeth can be one or more teeth with abnormal problems of the target object, which are determined by identifying the oral environment information of the target object. The target object can be a patient with abnormal teeth or an object that needs to reconstruct a tooth model. The target teeth can also be one or more teeth in a specified tooth scan image, which are determined according to actual needs. The image information specifically refers to an image containing the target teeth. The image information includes an image of at least one cropping angle of the target teeth. In this embodiment, in order to better segment the target teeth, the original tooth image uploaded by the oral scanning device is cropped at multiple angles to obtain image information of multiple cropping angles of the target teeth.

[0041] Specifically, the original three-dimensional image information containing the target teeth uploaded by the oral scanning device can be received in real time, or the original three-dimensional image information containing the target teeth can be retrieved from the target storage space according to the patient's information. Furthermore, the original three-dimensional image information of the target teeth is cropped according to the preset image rotation angle to obtain images of the target teeth at multiple cropping angles. It should be noted that in order to better obtain the images of the target teeth at each cropping angle, multiple images can be continuously collected at each cropping angle to reduce the impact of image noise on the image processing results. During the application process, one of the multiple images corresponding to each cropping angle can be randomly selected to ensure that each cropping angle has a corresponding image, wherein the image cropping ratio can be a random cropping ratio or can be set according to actual needs. The images of each cropping angle obtained by cropping are used as the image information of the target teeth.

[0042] Optionally, before obtaining the image information of the target tooth, it also includes: obtaining oral three-dimensional environment data of the target object, the oral three-dimensional environment data including three-dimensional data of multiple teeth; preprocessing the oral three-dimensional environment data to obtain a grayscale image corresponding to the oral three-dimensional environment data; traversing the pixel value of each pixel point in the grayscale image, calculating the difference between adjacent pixel points in turn, obtaining a pixel point difference value set of each tooth, and determining the target tooth based on the pixel point difference value set.

[0043] Among them, the target object specifically refers to an object with tooth reconstruction needs. The three-dimensional environment data specifically refers to the data that characterizes the three-dimensional environment of the oral cavity of the target object, including the surface shape and color texture information of soft and hard tissues such as teeth, gums, and mucosa. This information is obtained through a high-precision optical scanner or laser scanner, which can provide more accurate oral images. In this embodiment, the acquired oral three-dimensional environment data includes three-dimensional data of multiple teeth. In order to quickly segment the target teeth, it is necessary to preprocess the oral three-dimensional environment data to obtain a grayscale image, wherein the preprocessing includes but is not limited to the farthest point sampling algorithm, the K nearest neighbor algorithm, the adaptive median filter and the histogram equalization algorithm.

[0044] Specifically, the oral environment of the target object is scanned by an external oral scanning device to obtain the original three-dimensional environment data of the oral cavity of the target object, and the original three-dimensional environment data is uploaded to the host computer module matched with the external oral scanning device. The original three-dimensional environment data is processed by the data processing method pre-set in the host computer module to obtain the three-dimensional environment data of the oral cavity of the target object. The obtained three-dimensional environment data of the oral cavity of the target object can be understood as a three-dimensional data model of the oral cavity of the target object, wherein the scanning frequency, scanning mode and scanning direction of the oral scanning device can be set by the host computer module. The three-dimensional environment data of the oral cavity of the target object can be obtained from the host computer module. The three-dimensional environment data of the oral cavity of the target object is processed by pre-processing algorithms such as the farthest point sampling algorithm, the K nearest neighbor algorithm, the adaptive median filter and the histogram equalization algorithm to obtain a grayscale image corresponding to the three-dimensional environment data of the oral cavity. Traverse the pixel value of each pixel in the grayscale image, calculate the difference between adjacent pixels in turn, and store the related pixels whose calculated difference between adjacent pixels is not 0. If the pixel information includes the corresponding tooth position information, then the pixel information and the difference of the tooth position are stored in the information storage space corresponding to the tooth position, wherein each information storage space is provided with the corresponding tooth position information, and the corresponding information storage space can be matched according to the tooth position information. It should be noted that after traversing all the pixels in the grayscale image, the pixel difference set of each tooth is obtained, and it is determined whether the differences in the pixel difference set of each tooth are the same, and the tooth corresponding to the pixel difference set with the same difference is obtained, and the tooth is used as the target tooth.

[0045] In this embodiment, by acquiring the image data of the target teeth, a method for subsequently carrying out targeted reconstruction of the data model of the target teeth is provided. There is no need to reconstruct the data models of all teeth, thereby reducing the amount of data processing, improving data processing efficiency, and solving dental problems in a targeted manner.

[0046] S120, performing segmentation processing on the image information of the target teeth to obtain an image segmentation result of the target teeth.

[0047] It is understandable that the acquired image information of the target tooth includes information of the target tooth and surrounding tissues, and the target tooth needs to be segmented separately to obtain an accurate segmented image of the target tooth, so that the data model of the target tooth can be better reconstructed, so as to provide accurate reference data for subsequent crown design work. In this embodiment, the segmentation method used for the segmentation processing of the image information of the target tooth can be an image segmentation model constructed based on a deep learning network, or it can be an image segmentation algorithm called for image segmentation, which is not limited here.

[0048] Specifically, the pre-trained image segmentation model may be called to perform segmentation processing on the images of each cropping angle in the image information of the target teeth to obtain the image segmentation result of the target teeth.

[0049] In this embodiment, the segmentation result of the target tooth is obtained by segmenting the image information of the target tooth, which is used for the subsequent reconstruction of the three-dimensional data model of the target tooth, reducing the interference of the image information of non-target teeth, and helping to improve the accuracy and speed of the reconstruction of the three-dimensional data model of the target tooth.

[0050] S130: construct a three-dimensional shape model of the target tooth based on the image segmentation result.

[0051] Among them, the three-dimensional shape model of the target teeth can be understood as a model that can reflect the morphology and structure of the teeth. It is obtained by splicing the image segmentation results representing the target teeth. The image segmentation results can be processed by a pre-built tooth shape model to obtain the three-dimensional shape model of the target teeth. It can also be obtained by splicing according to the arrangement order corresponding to the image segmentation results. The arrangement order can be determined according to the cropping order of the image corresponding to the image segmentation result, or it can be determined according to the cropping angle of the image corresponding to the image segmentation result, which is not limited here.

[0052] Specifically, the arrangement order corresponding to the image segmentation results is obtained, wherein the arrangement order is the cropping order of the image corresponding to the image segmentation results, any segmented image in the image segmentation results is randomly selected as the reference image, and the corresponding segmented image is distributed forward or backward on both sides of the reference image according to the arrangement order to complete the splicing of the image segmentation results and construct a three-dimensional shape model of the target tooth.

[0053] In this embodiment, the segmented image information of the target teeth is spliced ​​to construct a three-dimensional shape model of the target teeth, which is used to further complete the three-dimensional data model of the target teeth. This helps the subsequently constructed three-dimensional data model of the target teeth to be more consistent with the actual shape of the target teeth, thereby improving the accuracy of the three-dimensional data model of the target teeth.

[0054] S140, performing boundary extraction on the three-dimensional shape model of the target tooth to obtain binary contour information of the target tooth.

[0055] The binary contour information refers to information that uses two values ​​to represent the pixel points on the image that represent the contour. In this embodiment, the contour information of the target tooth in the three-dimensional shape model of the target tooth is represented by the binary contour information, and the two values ​​of "1" and "0" respectively represent the tooth contour area and the non-tooth contour area of ​​the three-dimensional shape data model of the target tooth. The boundary extraction can be performed by a pre-built tooth boundary extraction model, or the boundary of the three-dimensional shape model of the target tooth can be identified according to a boundary recognition algorithm to obtain the boundary of the three-dimensional shape model of the target tooth and extract it.

[0056] Specifically, by calling a pre-built tooth boundary extraction model, the three-dimensional shape model of the target tooth is input into the tooth boundary extraction model, and boundary extraction processing is performed to obtain binary contour information of the target tooth.

[0057] In this embodiment, by extracting the boundary information of the target tooth in the three-dimensional shape model of the target tooth and obtaining the binary contour information of the target tooth, the information of the pixel points on the non-target tooth contour can be further eliminated. At the same time, the amount of data processing can also be reduced, and the accuracy and speed of target tooth boundary extraction can be improved, which helps to improve the accuracy of the three-dimensional data model of the target tooth.

[0058] S150: construct a three-dimensional mesh model of the target tooth based on the binary contour information of the target tooth.

[0059] Among them, the three-dimensional mesh model can be specifically understood as a three-dimensional model structure composed of polygons, which is used to represent the topology and spatial structure of the surface contour of the three-dimensional model. The three-dimensional mesh model includes but is not limited to a triangular mesh model and a quadrilateral mesh model.

[0060] Specifically, the binary contour information of the target tooth is processed by calling a surface reconstruction algorithm to obtain a three-dimensional mesh model of the target tooth. The surface reconstruction algorithm includes but is not limited to Delaunay triangulation, Poisson surface reconstruction algorithm, octree reconstruction algorithm and marching cube method. The pre-built tooth fitting model can also be called for reconstruction.

[0061] On the basis of the above-mentioned embodiment, the method also includes: calling a pre-trained tooth reconstruction model, the pre-trained tooth reconstruction model includes a tooth shape construction sub-model, a tooth boundary extraction sub-model and a tooth fitting sub-model; processing the image segmentation result based on the tooth shape construction sub-model to construct a three-dimensional shape model of the target tooth; performing boundary extraction on the three-dimensional shape model of the target tooth based on the tooth boundary extraction sub-model to obtain the binary contour information of the target tooth; processing the binary contour information of the target tooth based on the tooth fitting sub-model to construct a three-dimensional mesh model of the target tooth.

[0062] Among them, the tooth shape construction sub-model refers to an algorithm model used to construct a three-dimensional shape model of teeth. The tooth shape construction sub-model can be stored in a preset storage space in advance and can be directly called when constructing a three-dimensional shape model. The tooth boundary extraction sub-model refers to a model used for boundary extraction, which is constructed based on a neural network. Optionally, the tooth boundary extraction sub-model includes an Attention-U-net neural network model, which is used to accelerate the training convergence speed of the model and enhance the robustness of the results. Among them, the Attention-U-net neural network model includes an encoding module, a decoding module and an attention module. The encoding module includes multiple downsampling convolution blocks and a spatial pyramid pooling layer. The decoding module includes multiple upsampling convolution blocks. The attention module includes multiple attention gates. Multiple downsampling convolution blocks, spatial pyramid pooling layers and multiple upsampling convolution blocks are connected in sequence. Each downsampling layer and the upper upsampling layer of the same upsampling layer are connected to the corresponding upsampling layer of the same layer through an attention gate. The tooth fitting sub-model refers to a model used for data fitting processing to determine the three-dimensional mesh model of the target tooth. It can be a deep learning network model. The trained tooth fitting sub-model can be stored in a preset storage space in advance and can be directly called when constructing the three-dimensional mesh model.

[0063] Specifically, after determining the image segmentation result of the target tooth, call the pre-trained tooth reconstruction model, input the image segmentation result of the target tooth into the tooth shape construction sub-model, process the image segmentation result, and obtain the three-dimensional shape model of the target tooth; input the three-dimensional shape model of the target tooth into the tooth boundary extraction sub-model, perform boundary extraction, and obtain the binary contour information of the target tooth; input the binary contour information of the target tooth into the tooth fitting sub-model, perform model reconstruction processing, and construct a three-dimensional mesh model of the target tooth.

[0064] In a specific embodiment, after obtaining image information of a target tooth, the image information includes images at multiple cropping angles, calling an image segmentation model to perform image segmentation processing on the images at multiple cropping angles in sequence to obtain image segmentation results corresponding to the multiple images, serializing the image segmentation results according to the cropping angles or cropping orders corresponding to the image segmentation results to obtain an arrangement sequence of the image segmentation results, randomly extracting any segmented image and splicing the segmented images starting from the position to obtain a three-dimensional shape model of the target tooth, inputting the three-dimensional shape model of the target tooth into a tooth boundary extraction sub-model, and performing multiple downsampling convolution blocks in the tooth boundary extraction sub-model on the target tooth. The three-dimensional shape model of the tooth is downsampled to extract the core feature information in the three-dimensional shape model of the target tooth, and the result after downsampling is input into the spatial pyramid pooling layer, which can process images of different sizes to obtain a feature map containing multi-scale information, and the feature map is input into multiple upsampling convolution blocks for upsampling processing, wherein each downsampling layer and the upper upsampling layer of the same upsampling layer are connected to the corresponding upsampling layer of the same layer through an attention gate, and the feature maps of each downsampling layer and the upper upsampling layer of the same upsampling layer are connected in series through the attention gate and then input into the corresponding upsampling layer of the same layer for upsampling processing. After upsampling processing A binary feature map with the same resolution as the three-dimensional shape model of the target tooth is obtained, that is, the binary contour information of the target tooth is obtained, and the Poisson surface construction algorithm is called to construct the surface of the binary contour information of the target tooth. The binary contour information of the target tooth is represented as a surface point cloud. For any specified point in the point cloud, the KD-tree data structure is used to quickly find the 30 points closest to it, and the covariance analysis is used to calculate the principal axis of these 30 adjacent points. The eigenvector of the covariance matrix represents the direction of the principal axis, and the eigenvector corresponding to the minimum eigenvalue is the normal direction of the point. The centroid of the surface point cloud of the target tooth is calculated. For each point, its normal and the direction from the centroid to the The angle between the vectors of the tooth itself, if the angle is greater than 90 degrees, the normal is oriented in the opposite direction to ensure that all normals point outward, and the normal is further oriented according to the consistent tangent plane of the normal. For a point cloud of a tooth model, after normal calculation and adjustment, the normals of all points point correctly to the outside and have consistent directions in different areas of the tooth surface. The point normals of the point cloud of the target tooth are used to perform the Poisson surface reconstruction process. The Poisson reconstruction algorithm constructs a continuous triangular mesh surface based on the normal information of the point cloud and the position of the points, so that the surface is as close as possible to the input point cloud. After reconstruction, a smooth triangular mesh model of the tooth is obtained.

[0065] The technical solution of this embodiment is to obtain the image information of the target tooth; segment the image information of the target tooth to obtain the image segmentation result of the target tooth; construct a three-dimensional shape model of the target tooth based on the image segmentation result; extract the boundary of the three-dimensional shape model of the target tooth to obtain the binary contour information of the target tooth; and construct a three-dimensional mesh model of the target tooth based on the binary contour information of the target tooth. This solution realizes the segmentation and reconstruction of the image information of the target tooth to obtain the three-dimensional shape model of the target tooth, and then extracts the boundary to obtain the accurate binary contour information of the target tooth, reconstructs the three-dimensional mesh model of the target tooth according to the two-dimensional contour information, solves the problems of inaccurate tooth segmentation and inaccurate tooth model reconstruction, improves the accuracy and speed of reconstructing the three-dimensional mesh model of the target tooth, provides an accurate three-dimensional tooth model for the generation of the crown, helps to improve the accuracy of the crown generation, makes the generated crown more consistent with the morphology of the corresponding target tooth, and better solves the problem of tooth abnormality.

[0066] Embodiment 2

[0067] Figure 2 This is a flowchart of a tooth data model reconstruction method provided by the second embodiment of the present invention. This embodiment is a further optimization of the solution of the above embodiment. Optionally, a pre-trained image segmentation model is called, and the image information of the target tooth is segmented based on the pre-trained image segmentation model to determine the image segmentation result of the target tooth. Figure 2 As shown, the method includes:

[0068] S210, acquiring image information of target teeth.

[0069] S220, calling a pre-trained image segmentation model, performing segmentation processing on the image information of the target teeth based on the pre-trained image segmentation model, and determining the image segmentation result of the target teeth.

[0070] The pre-trained image segmentation model is constructed based on a deep learning model architecture. Optionally, the pre-trained image segmentation model includes multiple convolutional layers, pooling layers, at least one upsampling layer and a feature fusion layer.

[0071] Specifically, a pre-trained image segmentation model is called, and the image information of the target tooth is input into the pre-trained image segmentation model to perform segmentation processing to obtain an image segmentation result of the target tooth.

[0072] Optionally, the image information of the target teeth is segmented based on a pre-trained image segmentation model to determine the image segmentation result of the target teeth, including: inputting the image information into a multi-layer convolution layer for convolution processing to obtain an intermediate feature map, wherein the convolution kernel sizes corresponding to the multi-layer convolution layers are the same, and the output result of each convolution layer is used as the input data of the next convolution layer; transmitting the intermediate feature map to the pooling layer for pooling processing to obtain a first feature map; transmitting the first feature map to the upsampling layer for upsampling processing to obtain a second feature map; transmitting the second feature map and the image information to the feature fusion layer for feature fusion processing to obtain a fused feature map, and using the fused feature map as the image segmentation result of the target teeth.

[0073] Specifically, the image information is input into a multi-layer convolution layer for convolution processing. The convolution kernels corresponding to the multi-layer convolution layers have the same size. For example, the convolution kernel size can be 5×5. The output result of each convolution layer is used as the input data of the next convolution layer. After each layer of convolution processing, the corresponding image feature information can be extracted. After the multi-layer convolution processing, an intermediate feature map containing various feature information of the target tooth can be obtained, wherein the feature information includes but is not limited to tooth edge features, tooth shape features, tooth and surrounding tissue contrast features, and texture features of different parts of the tooth; the intermediate feature map is transmitted to the pooling layer for pooling processing, wherein the pooling layer can adopt 2× 2, and adopt the maximum pooling method to perform pooling processing to obtain a first feature map; the first feature map is transmitted to the upsampling layer for upsampling processing to restore the resolution of the target tooth image and obtain a second feature map with the same resolution as the image information. Preferably, each upsampling layer is connected to a convolution layer for further extracting and refining the features; the second feature map and the image information are transmitted to the feature fusion layer for feature fusion processing to obtain a fused feature map, and the fused feature map is used as the image segmentation result of the target tooth. While restoring the image resolution, the feature information at different levels can also be fused, thereby retaining the detailed features of the target tooth and enhancing the accuracy of the target tooth segmentation.

[0074] Optionally, the image segmentation model is trained, including: obtaining sample data, dividing the sample data into a training set, a validation set and a test set; initializing the model parameters of the image segmentation model to be trained through an adaptive moment estimation optimizer, inputting the training set into the image segmentation model to be trained for training, determining the loss function value of the image segmentation model to be trained through a joint loss function, and feeding the loss function value back to the adaptive moment estimation optimizer to determine the model parameters of the next training round, wherein the joint loss function is obtained by combining a boundary loss function and a cross entropy loss function; after each round of training, inputting the validation set into the image segmentation model to be trained for verification, and determining the evaluation index data of the output result corresponding to the image segmentation model to be trained, the evaluation index data including the Dice coefficient and the average intersection-over-union ratio; dynamically adjusting the learning rate of the image segmentation model based on the evaluation index data; and when the preset training rounds are completed, ending the training process of the image segmentation model to be trained to obtain a trained image segmentation model.

[0075] Specifically, a grid data set representing the oral three-dimensional environment of the target object uploaded by the host computer module is obtained, and a preprocessing sampling method is used to obtain sample data from the grid set, wherein the preprocessing sampling method includes but is not limited to the farthest point sampling method and the K nearest neighbor algorithm; the sample data is divided into a training set, a validation set and a test set according to a preset division ratio; preferably, the obtained sample data is preprocessed, and the preprocessing method includes but is not limited to an adaptive median filter and a histogram equalization algorithm, wherein an adaptive median filter is used to filter out tiny protrusions or depressions that may exist in the tooth model image, and a histogram equalization algorithm is used to redistribute the grayscale value distribution of the tooth image, thereby enhancing the image contrast, and the grayscale histogram of the image is stretched to make the grayscale value distribution of the teeth and surrounding tissues originally concentrated in a narrow grayscale range more uniform.

[0076] Initialize the model parameters of the image segmentation model to be trained by the adaptive moment estimation optimizer, wherein the model parameters include but are not limited to the weight data of each convolutional layer and the learning efficiency of the model; input the training set into the image segmentation model to be trained, and the adaptive moment estimation optimizer will dynamically adjust the weight data and the learning efficiency of the model according to the gradient information of the model parameters in each iterative training. For example, a smaller learning rate is assigned to the part with a larger gradient change to avoid excessive parameter update amplitude and exceeding the optimal value, and a larger learning rate is assigned to the part with a smaller gradient change to accelerate the convergence speed; after each round of training, the loss function value of the image segmentation model to be trained is determined by the joint loss function, and the loss function value is fed back to the adaptive moment estimation optimizer to determine the model parameters of the next training round, wherein the joint loss function is obtained by combining the boundary loss function and the cross entropy loss function to simultaneously optimize the accuracy of the pixel-level classified tooth model image and the clarity of the tooth boundary. During the training process, the loss function value of the current round is calculated and updated with the optimizer of the next round of training so that the loss function value shows a decreasing trend. It is necessary to find the appropriate weight through experiments and parameter adjustment. The coefficient combination is used to balance the proportion of the two loss functions; after each round of training, the validation set is input into the image segmentation model to be trained for verification, and the evaluation index data of the output result corresponding to the image segmentation model to be trained is determined according to the output result of the model. The evaluation index data includes the Dice coefficient and the average intersection-over-union ratio. The closer the Dice coefficient is to 1, the better the segmentation effect of the model is, which means that the model's prediction of the tooth area on the validation set and the real area have a high degree of overlap, and the average intersection-over-union ratio is the average of the ratio of the intersection and union of the tooth area predicted by the model and the real tooth area, which can comprehensively reflect the performance of the model in the segmentation of different categories of regions; the learning rate of the image segmentation model is dynamically adjusted based on the evaluation index data. When the Dice coefficient and the average intersection-over-union ratio on the validation set have not improved in multiple consecutive rounds, the learning rate needs to be reduced. By reducing the learning rate, the model can search in the parameter space with a smaller step size, thereby jumping out of the local optimal dilemma and finding a better parameter value; after completing the preset training rounds, the training process of the image segmentation model to be trained is ended to obtain a trained image segmentation model

[0077] S230: construct a three-dimensional shape model of the target tooth based on the image segmentation result.

[0078] S240: Extract the boundary of the three-dimensional shape model of the target tooth to obtain binary contour information of the target tooth.

[0079] S250: construct a three-dimensional mesh model of the target tooth based on the binary contour information of the target tooth.

[0080] The technical solution of this embodiment is to obtain the image information of the target tooth; call the pre-trained image segmentation model, perform segmentation processing on the image information of the target tooth based on the pre-trained image segmentation model, and determine the image segmentation result of the target tooth; construct the three-dimensional shape model of the target tooth based on the image segmentation result; perform boundary extraction on the three-dimensional shape model of the target tooth to obtain the binary contour information of the target tooth; and construct the three-dimensional mesh model of the target tooth based on the binary contour information of the target tooth. This solution realizes the determination of the image segmentation result of the target tooth by the image segmentation model, improves the image segmentation result of the target tooth, further, determines the three-dimensional shape model of the target tooth according to the image segmentation result, and then performs boundary extraction to obtain accurate binary contour information of the target tooth, reconstructs the three-dimensional mesh model of the target tooth according to the two-dimensional contour information, solves the problem of inaccurate tooth segmentation and inaccurate tooth model reconstruction, improves the accuracy and speed of reconstructing the three-dimensional mesh model of the target tooth, provides an accurate three-dimensional tooth model for the generation of the crown, helps to improve the accuracy of the crown generation, makes the generated crown more consistent with the morphology of the corresponding target tooth, and better solves the problem of tooth abnormality.

[0081] Embodiment 3

[0082] Figure 3 Schematic diagram of the structure of a tooth data model reconstruction device provided by the third embodiment of the present invention. Figure 3 As shown, the device comprises:

[0083] An image information acquisition module 310 is used to acquire image information of a target tooth;

[0084] An image segmentation result determination module 320 is used to perform segmentation processing on the image information of the target tooth to obtain an image segmentation result of the target tooth;

[0085] A three-dimensional shape model determination module 330 is used to construct a three-dimensional shape model of the target tooth based on the image segmentation result;

[0086] A binary contour information determination module 340 is used to extract the boundary of the three-dimensional shape model of the target tooth to obtain the binary contour information of the target tooth;

[0087] The three-dimensional mesh model determination module 350 is used to construct a three-dimensional mesh model of the target tooth based on the binary contour information of the target tooth.

[0088] The technical solution of this embodiment is to obtain the image information of the target tooth through the image information acquisition module; the image segmentation result determination module performs segmentation processing on the image information of the target tooth to obtain the image segmentation result of the target tooth; the three-dimensional shape model determination module constructs the three-dimensional shape model of the target tooth based on the image segmentation result; the binary contour information determination module performs boundary extraction on the three-dimensional shape model of the target tooth to obtain the binary contour information of the target tooth; the three-dimensional mesh model determination module constructs the three-dimensional mesh model of the target tooth based on the binary contour information of the target tooth. This solution realizes the segmentation and reconstruction of the image information of the target tooth to obtain the three-dimensional shape model of the target tooth, and then performs boundary extraction to obtain the accurate binary contour information of the target tooth, and reconstructs the three-dimensional mesh model of the target tooth according to the two-dimensional contour information, solves the problem of inaccurate tooth segmentation and inaccurate tooth model reconstruction, improves the accuracy and rapidity of the reconstructed three-dimensional mesh model of the target tooth, provides an accurate three-dimensional tooth model for the generation of the crown, helps to improve the accuracy of the crown generation, makes the generated crown more consistent with the morphology of the corresponding target tooth, and better solves the problem of tooth abnormality.

[0089] On the basis of the above embodiment, optionally, before obtaining the image information of the target tooth, the device is also used to: obtain oral three-dimensional environmental data of the target object, the oral three-dimensional environmental data including three-dimensional data of multiple teeth; pre-process the oral three-dimensional environmental data to obtain a grayscale image corresponding to the oral three-dimensional environmental data; traverse the pixel value of each pixel point in the grayscale image, calculate the difference between adjacent pixel points in turn, obtain a pixel point difference value set of each tooth, and determine the target tooth based on the pixel point difference value set.

[0090] Optionally, the image segmentation result determination module 320 is specifically configured to call a pre-trained image segmentation model, perform segmentation processing on the image information of the target teeth based on the pre-trained image segmentation model, and determine the image segmentation result of the target teeth.

[0091] Optionally, the pre-trained image segmentation model includes multiple convolutional layers, pooling layers, at least one upsampling layer and a feature fusion layer; the image segmentation result determination module 320 is further specifically used to input image information into the multiple convolutional layers for convolution processing to obtain an intermediate feature map, wherein the convolution kernel sizes corresponding to the multiple convolutional layers are the same, and the output result of each convolutional layer is used as the input data of the next convolutional layer; the intermediate feature map is transmitted to the pooling layer for pooling processing to obtain a first feature map; the first feature map is transmitted to the upsampling layer for upsampling processing to obtain a second feature map; the second feature map and the image information are transmitted to the feature fusion layer for feature fusion processing to obtain a fused feature map, and the fused feature map is used as the image segmentation result of the target tooth.

[0092] Optionally, the device is also specifically used to: call a pre-trained tooth reconstruction model, the pre-trained tooth reconstruction model includes a tooth shape construction sub-model, a tooth boundary extraction sub-model and a tooth fitting sub-model; based on the tooth shape construction sub-model, process the image segmentation result to construct a three-dimensional shape model of the target tooth; based on the tooth boundary extraction sub-model, perform boundary extraction on the three-dimensional shape model of the target tooth to obtain the binary contour information of the target tooth; based on the tooth fitting sub-model, process the binary contour information of the target tooth to construct a three-dimensional mesh model of the target tooth.

[0093] Optionally, the tooth boundary extraction sub-model includes an Attention-U-net neural network model, wherein the Attention-U-net neural network model includes an encoding module, a decoding module and an attention module, the encoding module includes multiple downsampling convolution blocks and a spatial pyramid pooling layer, the decoding module includes multiple upsampling convolution blocks, the attention module includes multiple attention gates, multiple downsampling convolution blocks, spatial pyramid pooling layers and multiple upsampling convolution blocks are connected in sequence, and each downsampling layer and the upper upsampling layer of the same upsampling layer are connected to the corresponding upsampling layer of the same layer through an attention gate.

[0094] Optionally, the image segmentation result determination module 320 is also specifically used to obtain sample data, divide the sample data into a training set, a validation set and a test set; initialize the model parameters of the image segmentation model to be trained through an adaptive moment estimation optimizer, input the training set into the image segmentation model to be trained for training, determine the loss function value of the image segmentation model to be trained through a joint loss function, and feed the loss function value back to the adaptive moment estimation optimizer to determine the model parameters of the next training round, wherein the joint loss function is obtained by combining a boundary loss function and a cross entropy loss function; after each round of training, input the validation set into the image segmentation model to be trained for verification, and determine the evaluation index data of the output result corresponding to the image segmentation model to be trained, the evaluation index data including the Dice coefficient and the average intersection-over-union ratio; dynamically adjust the learning rate of the image segmentation model based on the evaluation index data; when the preset training rounds are completed, end the training process of the image segmentation model to be trained to obtain a trained image segmentation model.

[0095] The tooth data model reconstruction device provided in the embodiment of the present invention can execute the tooth data model reconstruction method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0096] Embodiment 4

[0097] Figure 41 is a schematic diagram of the structure of an electronic device provided in Embodiment 4 of the present invention. The electronic device 10 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 may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0098] like Figure 4 As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0099] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0100] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as a tooth data model reconstruction method.

[0101] In some embodiments, the tooth data model reconstruction method can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the tooth data model reconstruction method described above can be performed. Alternatively, in other embodiments, the processor 11 can be configured to execute the tooth data model reconstruction method in any other appropriate manner (for example, by means of firmware).

[0102] Various implementations 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 chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0103] The computer program for implementing the tooth data model reconstruction method of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general-purpose computer, a special-purpose computer or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program can be executed entirely on the machine, partially on the machine, partially on the machine as an independent software package and partially on a remote machine, or entirely on a remote machine or server.

[0104] Embodiment 5

[0105] Embodiment 5 of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to execute a tooth data model reconstruction method, the method comprising:

[0106] Acquire image information of target teeth;

[0107] Segmenting the image information of the target teeth to obtain the image segmentation result of the target teeth;

[0108] Construct a three-dimensional shape model of the target tooth based on the image segmentation result;

[0109] Perform boundary extraction on the three-dimensional shape model of the target tooth to obtain binary contour information of the target tooth;

[0110] A three-dimensional mesh model of the target tooth is constructed based on the binary contour information of the target tooth.

[0111] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0112] To provide interaction with a user, the systems and techniques described herein may 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 a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0113] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0114] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0115] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0116] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for reconstructing a tooth data model, characterized in that: include: Acquire image information of target teeth; Segmenting the image information of the target tooth to obtain an image segmentation result of the target tooth; constructing a three-dimensional shape model of the target tooth based on the image segmentation result; Performing boundary extraction on the three-dimensional shape model of the target tooth to obtain binary contour information of the target tooth; A three-dimensional mesh model of the target tooth is constructed based on the binary contour information of the target tooth.

2. The method according to claim 1, characterized in that: Before acquiring the image information of the target tooth, it also includes: Acquire three-dimensional oral environment data of a target object, wherein the three-dimensional oral environment data includes three-dimensional data of a plurality of teeth; Preprocessing the oral three-dimensional environment data to obtain a grayscale image corresponding to the oral three-dimensional environment data; The pixel value of each pixel in the grayscale image is traversed, and the difference between adjacent pixels is calculated in sequence to obtain a pixel difference value set of each tooth, and the target tooth is determined based on the pixel difference value set.

3. The method according to claim 1, characterized in that The segmenting process of the image information of the target tooth to obtain the image segmentation result of the target tooth includes: A pre-trained image segmentation model is called, and image information of the target tooth is segmented based on the pre-trained image segmentation model to determine an image segmentation result of the target tooth.

4. The method according to claim 3, characterized in that The pre-trained image segmentation model includes multiple convolutional layers, pooling layers, at least one upsampling layer and a feature fusion layer; The performing segmentation processing on the image information of the target tooth based on the pre-trained image segmentation model to determine the image segmentation result of the target tooth includes: Input the image information into the multi-layer convolutional layer for convolution processing to obtain an intermediate feature map, wherein the convolution kernels corresponding to the multi-layer convolutional layers have the same size, and the output result of each convolutional layer is used as the input data of the next convolutional layer; Transmitting the intermediate feature map to the pooling layer for pooling processing to obtain a first feature map; Transmitting the first feature map to the upsampling layer for upsampling to obtain a second feature map; The second feature map and the image information are transmitted to the feature fusion layer for feature fusion processing to obtain a fused feature map, and the fused feature map is used as the image segmentation result of the target tooth.

5. The method according to claim 1, characterized in that: The method further comprises: Calling a pre-trained tooth reconstruction model, wherein the pre-trained tooth reconstruction model includes a tooth shape construction sub-model, a tooth boundary extraction sub-model and a tooth fitting sub-model; Constructing a sub-model based on the tooth shape to process the image segmentation result and construct a three-dimensional shape model of the target tooth; Performing boundary extraction on the three-dimensional shape model of the target tooth based on the tooth boundary extraction sub-model to obtain binary contour information of the target tooth; The binary contour information of the target tooth is processed based on the tooth fitting sub-model to construct a three-dimensional mesh model of the target tooth.

6. The method according to claim 5, characterized in that The tooth boundary extraction submodel includes an Attention-U-net neural network model, wherein the Attention-U-net neural network model includes an encoding module, a decoding module and an attention module, the encoding module includes multiple downsampling convolution blocks and a spatial pyramid pooling layer, the decoding module includes multiple upsampling convolution blocks, the attention module includes multiple attention gates, the multiple downsampling convolution blocks, the spatial pyramid pooling layer and the multiple upsampling convolution blocks are connected in sequence, and each of the downsampling layers and the upper upsampling layer of the same upsampling layer is connected to the corresponding upsampling layer of the same layer through the attention gate.

7. The method according to claim 3, characterized in that Training the image segmentation model includes: Obtaining sample data, and dividing the sample data into a training set, a validation set, and a test set; Initializing the model parameters of the image segmentation model to be trained by an adaptive moment estimation optimizer, inputting the training set into the image segmentation model to be trained, determining the loss function value of the image segmentation model to be trained by a joint loss function, and feeding the loss function value back to the adaptive moment estimation optimizer to determine the model parameters of the next training round, wherein the joint loss function is obtained by combining a boundary loss function and a cross entropy loss function; After each round of training, the verification set is input into the image segmentation model to be trained for verification, and evaluation index data of the output result corresponding to the image segmentation model to be trained is determined, wherein the evaluation index data includes the Dice coefficient and the average intersection-over-union ratio; Dynamically adjusting the learning rate of the image segmentation model based on the evaluation index data; When the preset training rounds are completed, the training process of the image segmentation model to be trained is ended to obtain a trained image segmentation model.

8. A tooth data model reconstruction device, characterized in that: include: An image information acquisition module, used to acquire image information of target teeth; An image segmentation result determination module is used to perform segmentation processing on the image information of the target tooth to obtain the image segmentation result of the target tooth; A three-dimensional shape model determination module, used to construct a three-dimensional shape model of the target tooth based on the image segmentation result; A binary contour information determination module, used for performing boundary extraction on the three-dimensional shape model of the target tooth to obtain binary contour information of the target tooth; The three-dimensional mesh model determination module is used to construct a three-dimensional mesh model of the target tooth based on the binary contour information of the target tooth.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can perform the tooth data model reconstruction method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the tooth data model reconstruction method according to any one of claims 1 to 7 when executed.

Images