Tooth orthodontic analysis method, device and equipment based on multi-scale convolution and medium

Abstract

The invention relates to the technical field of tooth point cloud data processing, and discloses a tooth orthodontic analysis method, device and equipment based on multi-scale convolution and a storage medium, and the method comprises the following steps: extracting tooth pose features and tooth geometric shape features from tooth point cloud data obtained in advance by using point cloud dynamic graph convolution; performing multi-channel graph convolution on the tooth pose features and the tooth geometric shape features to obtain multi-scale features of the tooth point cloud data; and generating a target pose of the multi-scale feature by using a preset multi-layer perceptron. Through the implementation of the method, the accuracy of target position analysis of tooth orthodontics can be effectively improved by means of point cloud dynamic graph convolution and a multi-layer perceptron.

Description

Technical Field

[0001] The present application relates to the technical field of tooth point cloud data processing, and in particular to a tooth orthodontic analysis method, device, equipment and storage medium based on multi-scale convolution. Background Art

[0002] The existing methods for predicting the target position of teeth after orthodontics can be divided into the following three categories:

[0003] 1) Orthodontic target position prediction based on expert rules: This method usually requires a lot of manual intervention from doctors and usually requires some prior knowledge such as tooth key points. However, this method requires a lot of pre-labeled tooth key point information and is too dependent on prior knowledge, resulting in certain problems with the accuracy of target position analysis.

[0004] 2) Orthodontic target position prediction based on generative models: This method is mainly aimed at 2D images. Given a 2D face image before orthodontic treatment, it generates a predicted 2D face image after orthodontic treatment. There are also some methods for 3D models that do not require advance segmentation of teeth. The desired target position is achieved by editing the reconstructed 3D model. However, the accuracy of target position analysis when this method is used for 3D processing is uncontrollable.

[0005] 3) Orthodontic target position prediction based on regressing the rotational and translational transformation matrix of each tooth. However, this method requires paired tooth models before and after orthodontics for training, and it is difficult to handle some cases with severe tooth deformities, which will ultimately result in inaccurate orthodontic target position analysis. Summary of the invention

[0006] The present application provides a method, device, equipment and medium for orthodontic analysis based on multi-scale convolution, the main purpose of which is to solve the problem of inaccurate target position analysis in orthodontic treatment.

[0007] To achieve the above objectives, the present application provides a tooth orthodontic analysis method based on multi-scale convolution, comprising:

[0008] Use point cloud dynamic graph convolution to extract tooth posture features and tooth geometric shape features from pre-acquired tooth point cloud data;

[0009] Perform multi-channel graph convolution on tooth posture features and tooth geometric shape features to obtain multi-scale features of tooth point cloud data;

[0010] The preset multi-layer perceptron is used to generate target poses with multi-scale features.

[0011] In order to solve the above problems, the present application also provides a tooth orthodontic analysis device based on multi-scale convolution, comprising:

[0012] A feature extraction module, used to extract tooth posture features and tooth geometric shape features from pre-acquired tooth point cloud data using point cloud dynamic graph convolution;

[0013] Multi-channel analysis module, used to perform multi-channel graph convolution on tooth posture features and tooth geometric shape features to obtain multi-scale features of tooth point cloud data;

[0014] The multi-layer perception module is used to generate target poses with multi-scale features using a preset multi-layer perceptron.

[0015] In order to solve the above problems, the present application also provides a device, including:

[0016] at least one processor; and,

[0017] a memory communicatively connected to at least one processor; wherein,

[0018] The memory stores a computer program that can be executed by at least one processor, and the computer program is executed by at least one processor so that the at least one processor can perform the above-mentioned multi-scale convolution-based orthodontic analysis method.

[0019] In order to solve the above problems, the present application also provides a storage medium, in which at least one computer program is stored, and the at least one computer program is executed by a processor in the device to implement the above-mentioned orthodontic analysis method based on multi-scale convolution.

[0020] The embodiment of the present application extracts tooth posture features and tooth geometry features from pre-acquired tooth point cloud data through point cloud dynamic graph convolution, avoiding data denoising and other operations in common methods, improving data analysis efficiency, and at the same time, using multi-channel graph convolution to analyze and process tooth posture features and tooth geometry features to obtain multi-scale features of each tooth in the tooth point cloud data, thereby providing more accurate data support for subsequent analysis, and further, multi-layer perceptrons are used to process multi-scale features to improve the accuracy of the target posture finally analyzed. Therefore, the orthodontic analysis method, device, equipment and medium based on multi-scale convolution proposed in this application can solve the problem of inaccurate target position analysis of orthodontic teeth. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 A schematic diagram of a process flow of a tooth orthodontic analysis method based on multi-scale convolution provided in an embodiment of the present application;

[0022] Figure 2 A schematic diagram of the structure of a three-head DGCNN architecture provided in an embodiment of the present application;

[0023] Figure 3-1A schematic diagram of a global adjacency matrix provided in one embodiment of the present application;

[0024] Figure 3-2 A schematic diagram of a local adjacency matrix provided in one embodiment of the present application;

[0025] Figure 3-3 A schematic diagram of an intersecting adjacency matrix provided in an embodiment of the present application;

[0026] Figure 4 A schematic diagram of the network structure of a multi-layer perceptron provided in one embodiment of the present application;

[0027] Figure 5 A functional module diagram of a tooth orthodontic analysis device based on multi-scale convolution provided in one embodiment of the present application;

[0028] Figure 6 A schematic diagram of the structure of a device for implementing a tooth orthodontic analysis method based on multi-scale convolution provided in one embodiment of the present application.

[0029] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0030] It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0031] Existing methods for predicting the target position of teeth after orthodontic treatment mainly include orthodontic target position prediction based on expert rules, orthodontic target position prediction based on generative models, and orthodontic target position prediction based on regressing the rotation and translation transformation matrix of each tooth. However, all of the above methods have the problem that the orthodontic target position analysis is not accurate enough. An embodiment of the present invention provides an orthodontic analysis method based on multi-scale convolution, which is a method for effectively improving the accuracy of orthodontic target position analysis based on point cloud dynamic graph convolution and multi-layer perceptron.

[0032] Reference Figure 1 FIG. 1 is a flow chart of a method for orthodontic analysis based on multi-scale convolution provided in an embodiment of the present application. In this embodiment, the method for orthodontic analysis based on multi-scale convolution includes:

[0033] S1. Use point cloud dynamic graph convolution to extract tooth posture features and tooth geometric shape features from the pre-acquired tooth point cloud data.

[0034] In an embodiment of the present application, the tooth point cloud data includes point cloud data of the teeth in a local coordinate system, namely, a world coordinate system, collected by a preset sensor in the user's mouth.

[0035] In detail, since the data directly collected by the sensor contains noise, the data is unstable. Therefore, the commonly used convolutional network needs to denoise the data or reconstruct the network itself before processing such point cloud data. This method will result in low data processing efficiency and high processing cost.

[0036] Therefore, the embodiment of the present application uses point cloud dynamic convolution to directly extract tooth posture features and tooth geometric shape features from pre-acquired tooth point cloud data, avoiding data denoising and other operations and improving data analysis efficiency.

[0037] In an embodiment of the present application, point cloud dynamic graph convolution is used to extract tooth posture features and tooth geometric shape features from pre-acquired tooth point cloud data, including: extracting the local coordinate system and world coordinate system of each tooth from the tooth point cloud data; calculating the tooth posture features of each tooth based on the local coordinate system, the world coordinate system and a preset rotation matrix; and using a pre-trained deep autoencoder to perform edge convolution on the tooth point cloud data to obtain the geometric shape features of each tooth.

[0038] In detail, Figure 2 As shown, the present application can utilize the pre-acquired three-head DGCNN architecture to extract the local coordinates of each tooth in the three-dimensional space from the tooth point cloud data. Figure 2 In the figure, the three-head DGCNN architecture identifies the local coordinates of each tooth in three dimensions in x, y, and z.

[0039] In the embodiment of the present application, the tooth posture is the position and posture of the local coordinate system of the tooth relative to the world coordinate system, which is described by rotation and translation in three-dimensional space, wherein the tooth posture feature of each tooth is calculated according to the local coordinate system, the world coordinate system and the preset rotation matrix, including:

[0040] The following formula is used to express the transformation of the coordinate system, and then the tooth posture characteristics of each tooth are calculated:

[0041]

[0042] in, is the tooth posture feature of the i-th tooth in the world coordinate system, R -1 represents the rotation matrix, is the local coordinate of the ith tooth in the local coordinate system, and C is the center of mass of the ith tooth.

[0043] Therefore, the tooth's position and posture can be represented by the quaternion four-dimensional vector of the rotation matrix and the three-dimensional vector of the tooth's center of mass.

[0044] Furthermore, in the process of analyzing the final target position of teeth for orthodontic treatment, in order to improve the accuracy of the final analysis results, it is necessary to consider not only the tooth position characteristics of the teeth, but also the geometric shape data of the teeth such as the shape and size, surface texture and occlusal groove of different teeth.

[0045] Therefore, in order to effectively extract the geometric shape features of the tooth surface, the present application can use a pre-trained deep autoencoder to analyze the tooth point cloud data. The deep autoencoder is used to compress and encode the geometric shape information of each tooth in the tooth point cloud data.

[0046] In an embodiment of the present application, a pre-trained deep autoencoder is used to perform edge convolution on tooth point cloud data to obtain geometric shape features of each tooth, including: using a pre-trained deep autoencoder to perform convolution encoding on the tooth point cloud data to obtain encoded data; using a mapping matrix in the deep autoencoder to perform latent vector mapping on the encoded data to obtain a mapping matrix; performing maximum pooling on the mapping matrix, and using the attention mechanism in the deep autoencoder to calculate the mapping matrix after maximum pooling to obtain the geometric shape features of each tooth.

[0047] In detail, using a pre-trained deep autoencoder to convert tooth point cloud data into bit-coded data can improve the efficiency of subsequent extraction of each tooth's geometric shape features.

[0048] Specifically, the mapping matrix is ​​a pre-set weight matrix or bias matrix in the deep autoencoder, which can be used to map the encoded data of the current layer in the deep autoencoder into a latent vector, which is then used as the input of the subsequent layer. Through the projection of multiple layers of latent vectors and maximum pooling processing, the geometric shape features of each tooth in the tooth point cloud data can be roughly extracted.

[0049] Furthermore, by performing attention calculation on the mapping matrix after maximum pooling through the attention mechanism in the deep autoencoder, the refined geometric shape features of each tooth in the tooth point cloud data can be generated.

[0050] S2. Perform multi-channel graph convolution on the tooth posture features and the tooth geometric shape features to obtain multi-scale features of the tooth point cloud data.

[0051] In the embodiment of the present application, in order to achieve the final analysis of the target posture of orthodontic teeth, it is necessary to further extract the tooth posture characteristics and tooth geometric shape characteristics to obtain the multi-scale characteristics of each tooth in the tooth point cloud data, thereby providing more accurate data support for subsequent analysis.

[0052] In an embodiment of the present application, multi-channel graph convolution is performed on the tooth posture features and tooth geometric shape features to obtain multi-scale features of the tooth point cloud data, including: constructing a global adjacency matrix, a local adjacency matrix and an intersection adjacency matrix of each tooth according to the tooth posture features and tooth geometric shape features; using a preset graph convolutional neural network to perform graph convolution on the global adjacency matrix, local adjacency matrix and intersection adjacency matrix of each tooth to obtain a multi-channel convolution graph; splicing the multi-channel convolution graphs corresponding to the global adjacency matrix, local adjacency matrix and intersection adjacency matrix corresponding to each tooth to obtain the multi-scale features of the tooth point cloud data.

[0053] In detail, see Figure 3-1 As shown, it is a schematic diagram of the global adjacency matrix in the present application. The schematic diagram of the global adjacency matrix is ​​used to represent a fully connected graph composed of each tooth in the tooth point cloud data. The nodes represented by each tooth in the graph are connected to each other, so that the overall shape information of the tooth can be extracted from the global adjacency matrix later.

[0054] In detail, see Figure 3-2 Shown is a schematic diagram of the local adjacency matrix in the present application. The schematic diagram of the local adjacency matrix is ​​used to indicate that each tooth is not only connected to the opposite tooth, but also to its own adjacent teeth, and to the teeth adjacent to the opposite teeth, so that the local crowding or sparse information of the dentition composed of teeth can be extracted from the local adjacency matrix later.

[0055] In detail, see Figure 3-3 As shown, it is a schematic diagram of the intersection adjacency matrix in the present application. The schematic diagram of the intersection adjacency matrix is ​​used to indicate that each tooth is connected with the opposite tooth and the adjacent tooth, which can reflect the collision relationship between the teeth.

[0056] In an embodiment of the present application, a preset graph convolutional neural network (GCN) is used to perform graph convolution on the global adjacency matrix, local adjacency matrix and intersection adjacency matrix of each tooth, respectively, to obtain multi-channel convolution graphs corresponding to the global adjacency matrix, local adjacency matrix and intersection adjacency matrix.

[0057] Specifically, the multi-channel convolution images corresponding to the obtained global adjacency matrix, local adjacency matrix and intersection adjacency matrix are spliced ​​to obtain the multi-scale features of the tooth point cloud data. When splicing the multi-channel convolution images, a feature pyramid structure can be used to splice the multi-channel convolution images to achieve fast and accurate splicing of different convolution images.

[0058] S3. Generate the target pose of the multi-scale features using a preset multi-layer perceptron.

[0059] In an embodiment of the present application, in order to more accurately realize the analysis of the target posture of orthodontics for each tooth in the tooth point cloud data, a preset multi-layer perceptron is used to process multi-scale features to improve the accuracy of the target posture finally analyzed.

[0060] In an embodiment of the present application, a preset multi-layer perceptron is used to generate a target pose of multi-scale features, including: using a pre-acquired shared convolution kernel to perform a preset round of convolution on the multi-scale features to obtain convolution features; performing a global information embedding operation on the convolution features to obtain embedded features; adaptively recalibrating the embedded features to obtain corrected features; and generating a target pose of the multi-scale features based on the corrected features.

[0061] Ginseng Figure 4 As shown, Figure 4 This is a schematic diagram of the network structure of a multi-layer perceptron. After multi-channel graph convolution obtains the multi-scale features of the tooth point cloud data, the multi-scale features pass through the multi-level network of the multi-layer perceptron and are calculated using the same shared convolution kernel as parameters, and finally output the target pose of the multi-scale features.

[0062] In detail, the shared convolution kernel is a pre-set convolution kernel matrix for performing point multiplication operations with all data in the multi-scale features. By performing a preset round of convolution operations on the multi-scale features using the shared convolution kernel, the convolution features corresponding to the multi-scale features can be obtained.

[0063] In the embodiment of the present application, a shared convolution kernel (i.e., SharedMLP) is used to convolve multi-scale features, which is different from the traditional MLP (Multilayer Perceptron) processing method. In the traditional MLP processing method, each data has an independent MLP network, and each data has an independent weight and bias matrix, which will lead to a bloated network structure and low computational efficiency.

[0064] The shared convolution kernel (i.e., SharedMLP) adopted in this application realizes feature representation by sharing the same convolution kernel, that is, all data in the multi-scale feature are point-multiplied with the shared convolution kernel to realize parameter sharing. This sharing greatly reduces the number of parameters, making the multi-layer perceptron network have better generalization ability and improving the efficiency of analyzing multi-scale features.

[0065] Furthermore, since the above analysis of convolutional features uses the multi-channel graph convolution method, it may lead to channel dependency problems. In order to solve the channel dependency problem in the subsequent analysis and processing of convolutional features, Figure 4As shown, the SE block (Squenze-Excitation block) is used in the multi-layer perceptron of the present application. The SE block can compress the global information in the tooth point cloud data into a preset channel descriptor, that is, through the global average pooling of the channel, the feature data containing the global information is compressed into a feature vector, so that the generated vector is embedded in the convolution feature to obtain the embedded feature, thereby alleviating the channel dependency.

[0066] In the present application examples, Figure 4 As shown in the figure, after the SE block performs a global information embedding operation on the convolutional features to generate embedded features, it is still necessary to perform an Excitation (adaptive recalibration) operation to fully capture the channel dependency.

[0067] In an embodiment of the present application, the embedded features are adaptively recalibrated to obtain the corrected features, including: using a preset first activation function to fully connect the embedded features to obtain an intermediate data set; using a preset second activation function to fully connect the intermediate data set to obtain an adaptive weight; using the adaptive weight to perform weighted calculation on the embedded features to obtain the corrected features.

[0068] In detail, the first activation function includes a Relu activation function, and the second activation function includes a Sigmoid activation function.

[0069] Specifically, this embodiment uses two full connections to process the embedded features. The first fully connected layer uses the Relu activation function to compress the embedded features of multiple channels, thereby reducing the amount of data for analysis and calculation. The second fully connected layer uses the Sigmoid activation function to activate the output data of the first fully connected layer to obtain weights for characterizing the features of different channels, and then through weighted processing of the weights, adaptive correction of the embedded features is achieved.

[0070] In an embodiment of the present application, a residual network for predicting and analyzing the target posture of each tooth can be added after the SE block structure to achieve the prediction of the target posture of each tooth.

[0071] In detail, the residual network can be trained using the following loss function:

[0072] Loss=Loss rotation +Loss translation

[0073] Among them, Loss is the training loss value of the residual network, Loss rotation is the cosine similarity between the predicted tooth posture after orthodontics in the pre-acquired training data and the actual tooth posture after orthodontics corresponding to the training data, Loss translationThe distance between the predicted centroid of the orthodontic teeth in the training data and the actual centroid of the orthodontic teeth corresponding to the training data.

[0074] In the embodiment of the present application, the trained residual network can be used to calculate the correction features to obtain the target posture of each tooth in the multi-scale features.

[0075] like Figure 5 , is a functional module diagram of a dental orthodontic analysis device based on multi-scale convolution provided in one embodiment of the present application.

[0076] The orthodontic analysis device 500 based on multi-scale convolution of the present application can be installed in a device. According to the functions implemented, the orthodontic analysis device 500 based on multi-scale convolution can include a feature extraction module 501, a multi-channel analysis module 502 and a multi-layer perception module 503. The module of the present application can also be called a unit, which refers to a series of computer program segments that can be executed by a device processor and can complete fixed functions, which are stored in the memory of the device.

[0077] In this embodiment, the functions of each module / unit are as follows:

[0078] The feature extraction module 501 is used to extract tooth posture features and tooth geometric shape features from pre-acquired tooth point cloud data using point cloud dynamic graph convolution;

[0079] A multi-channel analysis module 502, used for performing multi-channel graph convolution on tooth posture features and tooth geometric shape features to obtain multi-scale features of tooth point cloud data;

[0080] The multi-layer perception module 503 is used to generate a target posture with multi-scale features using a preset multi-layer perceptron.

[0081] In detail, each module in the orthodontic analysis device 500 based on multi-scale convolution in the embodiment of the present application is used in the same manner as described above. Figures 1 to 4 The orthodontic analysis method based on multi-scale convolution has the same technical means and can produce the same technical effects, so it will not be repeated here.

[0082] like Figure 6 , is a schematic diagram of the structure of a device for implementing a tooth orthodontic analysis method based on multi-scale convolution provided in one embodiment of the present application.

[0083] The device may include a processor 601, a memory 602, a communication bus 603, and a communication interface 604, and may also include a computer program stored in the memory 602 and executable on the processor 601, such as a dental orthodontic analysis program based on multi-scale convolution.

[0084] In some embodiments, the processor 601 may be composed of an integrated circuit, for example, a single packaged integrated circuit, or a plurality of packaged integrated circuits with the same or different functions, including one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and combinations of various control chips. The processor 601 is the control core (Control Unit) of the device, and uses various interfaces and lines to connect the various components of the entire device, and executes or executes programs or modules stored in the memory 602 (for example, executing a dental orthodontic analysis program based on multi-scale convolution, etc.), and calls the data stored in the memory 602 to execute various functions of the device and process data.

[0085] The memory 602 includes at least one type of readable storage medium, and the readable storage medium includes a flash memory, a mobile hard disk, a multimedia card, a card-type memory (for example, SD or DX memory, etc.), a magnetic memory, a disk, an optical disk, etc. In some embodiments, the memory 602 may be an internal storage unit of the device, such as a mobile hard disk of the device. In other embodiments, the memory 602 may also be an external storage device of the device, such as a plug-in mobile hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the device. Further, the memory 602 may also include both an internal storage unit of the device and an external storage device. The memory 602 can not only be used to store application software and various types of data installed on the device, such as the code of a dental orthodontic analysis program based on multi-scale convolution, but also can be used to temporarily store data that has been output or is to be output.

[0086] The communication bus 603 may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. The bus is configured to realize connection and communication between the memory 602 and at least one processor 601, etc.

[0087] The communication interface 604 is used for communication between the above-mentioned device and other devices, including a network interface and a user interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is generally used to establish a communication connection between the device and other devices. The user interface may be a display (Display), an input unit (such as a keyboard (Keyboard)), and optionally, the user interface may also be a standard wired interface, a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, and an OLED (Organic Light-Emitting Diode, organic light-emitting diode) touch device, etc. Among them, the display may also be appropriately referred to as a display screen or a display unit, which is used to display information processed in the device and to display a visual user interface.

[0088] The figure only shows a device with components. Those skilled in the art will understand that the structure shown in the figure does not constitute a limitation on the device, and may include fewer or more components than shown in the figure, or combine certain components, or arrange the components differently.

[0089] For example, although not shown, the device may also include a power source (such as a battery) for supplying power to each component. Preferably, the power source may be logically connected to at least one processor 601 through a power management device, so that the power management device can realize functions such as charging management, discharging management, and power consumption management. The power source may also include any components such as one or more DC or AC power sources, recharging devices, power failure detection circuits, power converters or inverters, and power status indicators. The device may also include a variety of sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be described in detail here.

[0090] It should be understood that the embodiment is for illustration only and the scope of the patent application is not limited by this structure.

[0091] Specifically, the specific implementation method of the processor 601 for the above instructions can refer to the description of the relevant steps in the corresponding embodiment of the accompanying drawings, which will not be repeated here.

[0092] Furthermore, if the module / unit integrated in the device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a storage medium. The storage medium can be volatile or non-volatile. For example, computer-readable media can include: any entity or device that can carry computer program code, recording media, USB flash drives, mobile hard disks, magnetic disks, optical disks, computer memory, and read-only memory (ROM).

[0093] The present application also provides a storage medium, which stores a computer program. When the computer program is executed by a processor of a device, it can implement the orthodontic analysis method based on multi-scale convolution of any of the above embodiments. It should be noted that the storage medium can be volatile or non-volatile. For example, the storage medium may include: any entity or device capable of carrying a computer program code, a recording medium, a USB flash drive, a mobile hard disk, a magnetic disk, an optical disk, a computer memory, and a read-only memory (ROM).

[0094] In the several embodiments provided in this application, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of modules is only a logical function division, and there may be other division methods in actual implementation.

[0095] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0096] In addition, each functional module in each embodiment of the present application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0097] It is obvious to those skilled in the art that the present application is not limited to the details of the above exemplary embodiments, and that the present application can be implemented in other specific forms without departing from the spirit or essential features of the present application.

[0098] Therefore, no matter from which point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present invention is not limited only according to the above description, and it is intended that all changes within the meaning and scope of equivalent elements within the scope of protection are included in the present invention.

[0099] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0100] In addition, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in the system claim can also be implemented by one unit or device through software or hardware. The words first, second, etc. are used to indicate names, and do not indicate any particular order.

[0101] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present application and are not intended to limit it. Although the present application has been described in detail with reference to the preferred embodiments, a person of ordinary skill in the art should understand that the technical solution of the present application may be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present application.

Claims

1. A tooth orthodontic analysis method based on multi-scale convolution, characterized in that: The method comprises: Use point cloud dynamic graph convolution to extract tooth posture features and tooth geometric shape features from pre-acquired tooth point cloud data; Performing multi-channel graph convolution on the tooth posture features and the tooth geometric shape features to obtain multi-scale features of the tooth point cloud data; The target pose of the multi-scale feature is generated using a preset multi-layer perceptron.

2. The orthodontic analysis method based on multi-scale convolution according to claim 1, characterized in that: The method of extracting tooth posture features and tooth geometric shape features from pre-acquired tooth point cloud data using point cloud dynamic graph convolution includes: Extracting a local coordinate system and a world coordinate system of each tooth from the tooth point cloud data; Calculating the tooth posture feature of each tooth according to the local coordinate system, the world coordinate system and a preset rotation matrix; The tooth point cloud data is subjected to edge convolution using a pre-trained deep autoencoder to obtain the geometric shape features of each tooth.

3. The orthodontic analysis method based on multi-scale convolution according to claim 2, characterized in that: The step of calculating the tooth posture feature of each tooth according to the local coordinate system, the world coordinate system and a preset rotation matrix includes: The tooth posture characteristics of each tooth are calculated using the following formula: in, is the tooth posture feature of the i-th tooth in the world coordinate system, i is the preset rotation matrix, is the local coordinate of the ith tooth in the local coordinate system, and C is the center of mass of the ith tooth.

4. The orthodontic analysis method based on multi-scale convolution according to claim 1, characterized in that: The method of performing edge convolution on the tooth point cloud data using a pre-trained deep autoencoder to obtain geometric shape features of each tooth includes: Using a pre-trained deep autoencoder to perform convolution encoding on the tooth point cloud data to obtain encoded data; Using the mapping matrix in the deep autoencoder to perform latent vector mapping on the encoded data, to obtain a mapping matrix; The mapping matrix is ​​subjected to maximum pooling, and the mapping matrix after maximum pooling is calculated using the attention mechanism in the deep autoencoder to obtain the geometric shape features of each tooth.

5. The orthodontic analysis method based on multi-scale convolution according to claim 4, characterized in that: The multi-channel graph convolution is performed on the tooth posture feature and the tooth geometric shape feature to obtain the multi-scale feature of the tooth point cloud data, including: Constructing a global adjacency matrix, a local adjacency matrix and an intersection adjacency matrix of each tooth according to the tooth posture characteristics and tooth geometric shape; Using a preset graph convolutional neural network, graph convolution is performed on the global adjacency matrix, the local adjacency matrix and the intersection adjacency matrix of each tooth to obtain a multi-channel convolution graph; The multi-channel convolution graphs corresponding to the global adjacency matrix, the local adjacency matrix and the intersection adjacency matrix corresponding to each tooth are spliced ​​to obtain multi-scale features of the tooth point cloud data.

6. The orthodontic analysis method based on multi-scale convolution according to claim 1, characterized in that: The method of generating the target posture of the multi-scale feature by using a preset multi-layer perceptron includes: Using the pre-acquired shared convolution kernel to perform a preset round of convolution on the multi-scale features to obtain convolution features; Performing a global information embedding operation on the convolutional features to obtain embedded features; Adaptively recalibrating the embedded features to obtain calibrated features; A target pose of the multi-scale feature is generated according to the correction feature.

7. The orthodontic analysis method based on multi-scale convolution according to claim 1, characterized in that: The step of adaptively recalibrating the embedded features to obtain the calibrated features comprises: Performing full connection processing on the embedded features using a preset first activation function to obtain an intermediate data set; Performing full connection processing on the intermediate data set using a preset second activation function to obtain an adaptive weight; The embedded features are weighted calculated using the adaptive weights to obtain corrected features.

8. A tooth orthodontic analysis device based on multi-scale convolution, characterized in that: The device comprises: A feature extraction module, used to extract tooth posture features and tooth geometric shape features from pre-acquired tooth point cloud data using point cloud dynamic graph convolution; A multi-channel analysis module, used for performing multi-channel graph convolution on the tooth posture features and the tooth geometric shape features to obtain multi-scale features of the tooth point cloud data; The multi-layer perception module is used to generate the target posture of the multi-scale features using a preset multi-layer perceptron.

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 multi-scale convolution-based orthodontic analysis method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the orthodontic analysis method based on multi-scale convolution as described in any one of claims 1 to 7 is implemented.

Images