Learning device for dental sculpting under augmented reality guidance
By integrating AR modules and deep learning technology into the carving workbench, the problem of three-dimensional morphological correspondence in traditional dental carving teaching has been solved, realizing low-cost three-dimensional morphological guidance and real-time feedback, thus improving teaching effectiveness.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- THE AFFILIATED STOMATOLOGICAL HOSPITAL OF KUNMING MEDICAL UNIV
- Filing Date
- 2026-02-28
- Publication Date
- 2026-06-09
Smart Images
Figure FT_1 
Figure FT_2 
Figure FT_3
Abstract
Description
Technical Field
[0001] This invention relates to the field of dental sculpting technology, specifically to an augmented reality-guided dental sculpting learning device. Background Technology
[0002] Dental sculpting is one of the essential basic skills that dental students must master. In traditional teaching, students typically use carving tools on plaster, wax, or resin dental blocks based on two-dimensional diagrams in textbooks and teacher demonstrations. This approach has the following shortcomings: Students need to mentally transform the two-dimensional standard tooth shape from books or screens into a three-dimensional structure and then apply it to the cutting of tooth blocks. Beginners often find it difficult to accurately grasp the spatial shape. During the carving process, students cannot establish a precise and intuitive one-to-one correspondence between the current carving shape and the target standard shape. They can only rely on rough comparison with the naked eye, and errors are not easy to detect. Some existing systems based on virtual reality or instrument tracking can provide three-dimensional guidance, but they require expensive tracking hardware and professional instrument modifications. Moreover, students lack tactile feedback from real materials when operating in a virtual environment, which limits large-scale promotion. Many high-end systems require the use of special equipment or virtual joysticks, which are quite different from the traditional dental carving tools used in clinical practice, and are not conducive to the continuation of students' operating habits.
[0003] In addition, existing solutions attempt to overlay dental training objects or dental arch models with augmented reality content. However, common implementations rely on feature point / texture tracking of the tooth surface or the use of external optical tracking with multiple cameras. These methods are easily affected by occlusion, lighting, and surface materials, and the calibration and deployment costs are high. Therefore, there is an urgent need for a technical solution that retains the experience of using traditional carving tools and physical tooth blocks, while accurately overlaying standard digital images of teeth onto actual tooth blocks using augmented reality technology. This would provide students with an intuitive three-dimensional morphological reference, and would also be simple in structure, low in cost, and easy to promote in conventional laboratories. Summary of the Invention
[0004] To address the shortcomings of existing technologies, this invention provides an augmented reality-guided dental carving learning device. This device features the advantage of fitting and superimposing standard digital images of teeth with carved blocks, allowing for real-time observation of the spatial correspondence between standard shapes and actual carvings when using traditional dental carving tools, thus solving the aforementioned technical problems.
[0005] To achieve the above objectives, the present invention provides the following technical solution: an augmented reality-guided dental sculpting learning device, comprising a sculpting block, a sculpting worktable, a data processing system, and AR glasses. The sculpting block is detachably installed inside the sculpting worktable. The AR glasses are used to identify the sculpting block on the sculpting worktable and provide sculpting prompts based on the data processing system, which are displayed on the AR glasses.
[0006] As a preferred embodiment of the present invention, the carving workbench includes a mounting base and an AR recognition module. The carving tooth block is detachably installed in a groove opened on the top of the mounting base, and the AR recognition module is a QR code.
[0007] As a preferred technical solution of the present invention, the data processing system includes a perception and pose estimation unit, a registration and fitting unit, and a rendering and augmented reality overlay unit. The data processing system performs registration after selecting a preset standard model.
[0008] As a preferred embodiment of the present invention, the perception and pose estimation unit performs image acquisition and identification of AR-recognizable modules and pose calculation based on AR glasses, and the specific steps are as follows: AR marker recognition uses a pre-trained convolutional neural network (CNN) as input to extract marker locations from AR scene images acquired by AR glasses (4). The specific expression is as follows: in, Indicates the pixel coordinates of the center point of the AR marker; This represents the captured AR scene image. This represents a pre-trained convolutional neural network (CNN); g() represents the position decoding regression function. Pose calculation involves reading the output of a pre-trained convolutional neural network (CNN) and using the PnP algorithm to estimate the pose. The specific expression is as follows: in, This represents the intrinsic parameter matrix of the camera being captured; Represents the rotation matrix; It is a translation vector; Represents homogeneous pixel coordinates; Represents the homogeneous coordinates of a three-dimensional point; As a scale factor, Indicates to and splicing.
[0009] As a preferred embodiment of the present invention, the registration and fitting unit acquires different images from multiple camera devices, converts them into point cloud data, and minimizes the error from the perspective of each camera device. The specific expression is as follows: in, Indicates the first Three-dimensional points obtained from a single perspective; Represents the standard model; Indicates the first Weight of each perspective; Indicates the total error; For the first The squared Euclidean distance between a 3D point from a given viewpoint and the standard model; ||| represents the Euclidean norm; Indicates the total number of viewpoints; The registration and fitting unit adopts the thin plate spline method, and optimizes the three-dimensional model of the carving block (2) through deformation function. The specific expression is as follows: in, This indicates the optimization result. This represents the deformation function based on thin plate splines. Indicates the first Three-dimensional points obtained from a single perspective. Indicates the first Individual perspective deformation weights This represents the three-dimensional points output by the pre-trained convolutional neural network (CNN).
[0010] As a preferred embodiment of the present invention, the rendering and augmented reality overlay unit provides dynamic feedback during the sculpting process by combining real-time depth information and a lighting model to help adjust the sculpting path. The depth information calculation expression is as follows: in, This represents the shortest Euclidean distance between a pixel and the edge of the sculpted content. )≥0; ln() is the natural logarithm function; depth(x,y) is the normalized depth information; the lighting model adopts the Phong lighting model, and the expression is as follows: in, , , These represent ambient light, diffuse light, and specular light, respectively. This represents the lighting rendering value; The specific expression for adjusting the carving path is as follows: in, This represents the three-dimensional error vector of the sculpted area. Xactual refers to the actual three-dimensional form sculpted. Represents the standard model; This is the inverse of the camera intrinsic parameter matrix; , s is the rotation matrix and translation vector; s is the scale factor; This is an adjustment amount for the engraving path on a two-dimensional image plane, used to guide the operator in correcting the engraving direction.
[0011] Compared with the prior art, the present invention provides an augmented reality-guided dental sculpting learning device, which has the following beneficial effects: This invention integrates an AR-recognizable module into the carving workbench and uses augmented reality display to achieve spatial overlay of standard digital images of teeth with physical carving blocks. This allows students to receive 3D morphological guidance in a familiar operating environment. It eliminates the need for tracking sensors on the instruments and avoids reliance on expensive 3D tracking hardware. Only low-cost graphic markers need to be set on the workbench, and common camera equipment and computers / mobile terminals are required. This makes it suitable for widespread use in large-scale teaching classes. Furthermore, it combines multi-view deep learning non-rigid registration to achieve accurate fitting between the standard model and the carving blocks. Real-time lighting rendering and dynamic depth feedback ensure guidance stability in complex operating environments. It is also compatible with head-mounted devices and projection devices, adapting to diverse teaching scenarios. Attached Figure Description
[0012] Figure 1 This is a schematic diagram of the overall invention; Figure 2 This is a three-dimensional schematic diagram of the engraving worktable of the present invention; Figure 3 This is a schematic diagram of the engraving worktable of the present invention.
[0013] The components include: 1. Carving workbench; 11. Mounting base; 12. AR recognition module; 2. Carving block; 3. Data processing system; 4. AR glasses. Detailed Implementation
[0014] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0015] Please see Figures 1-3 The engraving worktable 1 has a mounting base 11 and an AR recognition module 12 on its surface; The carved tooth block 2 is detachably installed in the mounting base 11, and the unique correct placement direction is achieved through the limiting groove and the positioning block; Data processing system 3 runs on a single computer; AR Glasses 4, as an augmented reality display terminal, uses a front-facing camera to capture and project images of the control panel and the carved ivory block. An augmented reality-guided dental sculpting learning device includes a sculpting block 2, a sculpting worktable 1, a data processing system 3, and AR glasses 4. The sculpting block 2 is detachably installed inside the sculpting worktable 1. The AR glasses 4 are used to identify the sculpting block 2 on the sculpting worktable 1 and provide sculpting prompts based on the data processing system 3, which are displayed on the AR glasses 4.
[0016] The carving workbench 1 includes a mounting base 11 and an AR recognition module 12. The carving block 2 is detachably installed in a groove opened on the top of the mounting base 11. The AR recognition module 12 is a QR code.
[0017] The data processing system 3 includes a perception and pose estimation unit, a registration and fitting unit, and a rendering and augmented reality overlay unit. The data processing system 3 performs registration after selecting a preset standard model.
[0018] The perception and pose estimation unit performs image acquisition and recognition based on AR glasses 4 and AR recognizable module 12 and pose calculation. The specific steps are as follows: AR marker recognition uses a pre-trained convolutional neural network (CNN) as input to extract marker locations from AR scene images captured by AR glasses 4. The specific expression is as follows: in, Indicates the pixel coordinates of the center point of the AR marker; This represents the captured AR scene image. This represents a pre-trained convolutional neural network (CNN); g() represents the position decoding regression function. Pose calculation involves reading the output of a pre-trained convolutional neural network (CNN) and using the PnP algorithm to estimate the pose. The specific expression is as follows: in, This represents the intrinsic parameter matrix of the camera being captured; Represents the rotation matrix; It is a translation vector; Represents homogeneous pixel coordinates; Represents the homogeneous coordinates of a three-dimensional point; As a scale factor, Indicates to and splicing.
[0019] The registration and fitting unit acquires different images from multiple camera devices, converts them into point cloud data, and minimizes the error from the viewpoint of each camera device. The specific expression is as follows: in, Indicates the first Three-dimensional points obtained from a single perspective; Represents the standard model; Indicates the first Weight of each perspective; Indicates the total error; For the first The squared Euclidean distance between a 3D point from a given viewpoint and the standard model; ||| represents the Euclidean norm; Indicates the total number of viewpoints; The registration and fitting unit adopts the thin plate spline method, and the 3D model of the engraved block 2 is optimized by the deformation function. The specific expression is as follows: in, This indicates the optimization result. This represents the deformation function based on thin plate splines. Indicates the first Three-dimensional points obtained from a single perspective. Indicates the first Individual perspective deformation weights This represents the three-dimensional points output by the pre-trained convolutional neural network (CNN).
[0020] During the sculpting process, the rendering and augmented reality overlay unit provides dynamic feedback by combining real-time depth information and lighting models to help adjust the sculpting path. The depth information calculation expression is as follows: in, Represents the shortest Euclidean distance between a pixel and the edge of the sculpted content (dist(x,y)≥0); ln( ) is the natural logarithm function; depth(x,y) is the normalized depth information (ensuring that depth(x,y)≥0 when dist(x,y)≥0, which conforms to spatial depth logic); the lighting model adopts the Phong lighting model, and the expression is as follows: in, , , These represent ambient light, diffuse light, and specular light, respectively. This represents the lighting rendering value; The specific expression for adjusting the carving path is as follows: in, This represents the three-dimensional error vector of the sculpted area. Xactual refers to the actual three-dimensional form sculpted. Represents the standard model; R is the inverse of the camera intrinsic parameter matrix; R and t are the rotation matrix and translation vector obtained from the pose calculation; s is the scale factor in the pose calculation; ΔP is the carving path adjustment amount on the two-dimensional image plane (to guide the operator to correct the carving direction). Before teaching, the teacher selects a standard digital image of the target tooth location in data processing system 3 and sets the corresponding teaching task. The system pre-calculates the initial scaling ratio of the standard tooth image based on the size and installation position parameters of the carved tooth block 2.
[0021] During instruction, the operator places the carving workbench 1 within the field of view of the AR glasses 4 camera. The AR glasses 4 transmits the captured video stream to the data processing system 3 for recognition. The recognition process includes: Detect the position and orientation of the AR-recognizable module 12 from the image; Calculate the spatial coordinates of the operating table and the carved tooth block based on the test results; Using the registration and fitting module, the standard tooth digital image is accurately superimposed on the carved tooth block 2 to generate the fitted standard image; The fitted image is overlaid with a real-time video frame and displayed on the AR glasses screen.
[0022] The operator can see the actual tooth carving block 2 and its surrounding environment on the screen, and at the same time see the standard tooth shape outline and recommended trimming areas in the corresponding area of the tooth carving block. Students wear AR glasses to observe and operate on the tooth carving block with traditional dental carving tools, completing the carving in real time by referring to the standard shape.
[0023] After the carving is completed, the teacher can use a 3D scanner to scan the carved tooth block 2, import the data into the data processing system 33, perform difference analysis with the standard model, and output error heat map, volume difference data, etc.
[0024] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.
Claims
1. An augmented reality-guided dental sculpting learning device, characterized in that: The device includes a carving block (2), a carving workbench (1), a data processing system (3), and AR glasses (4). The carving block (2) is detachably installed inside the carving workbench (1). The AR glasses (4) are used to identify the carving block (2) on the carving workbench (1) and provide carving prompts based on the data processing system (3), which are displayed on the AR glasses (4).
2. The augmented reality-guided dental sculpting learning device according to claim 1, characterized in that: The carving workbench (1) includes a mounting base (11) and an AR recognition module (12). The carving block (2) is detachably installed in a groove opened on the top of the mounting base (11). The AR recognition module (12) is a QR code.
3. The augmented reality-guided dental sculpting learning device according to claim 2, characterized in that: The data processing system (3) includes a perception and pose estimation unit, a registration and fitting unit, and a rendering and augmented reality overlay unit. The data processing system (3) performs registration after selecting a preset standard model.
4. The augmented reality-guided dental sculpting learning device according to claim 3, characterized in that: The perception and pose estimation unit performs image acquisition and recognition of the AR-recognizable module (12) and pose calculation based on the AR glasses (4). The specific steps are as follows: AR marker recognition uses a pre-trained convolutional neural network (CNN) as input to extract marker locations from AR scene images acquired by AR glasses (4). The specific expression is as follows: in, Indicates the pixel coordinates of the center point of the AR marker; This represents the captured AR scene image. This represents a pre-trained convolutional neural network (CNN); g() represents the position decoding regression function. Pose calculation involves reading the output of a pre-trained convolutional neural network (CNN) and using the PnP algorithm to estimate the pose. The specific expression is as follows: in, This represents the intrinsic parameter matrix of the camera being captured; Represents the rotation matrix; It is a translation vector; Represents homogeneous pixel coordinates; Represents the homogeneous coordinates of a three-dimensional point; As a scale factor, Indicates to and splicing.
5. The augmented reality-guided dental sculpting learning device according to claim 3, characterized in that: The registration and fitting unit acquires different images from multiple camera devices, converts them into point cloud data, and minimizes the error from the perspective of each camera device. The specific expression is as follows: in, Indicates the first Three-dimensional points obtained from a single perspective; Represents the standard model; Indicates the first Weight of each perspective; Indicates the total error; For the first The squared Euclidean distance between a 3D point from a given viewpoint and the standard model; ||| represents the Euclidean norm; Indicates the total number of viewpoints; The registration and fitting unit adopts the thin plate spline method, and optimizes the three-dimensional model of the carving block (2) through deformation function. The specific expression is as follows: in, This indicates the optimization result. This represents the deformation function based on thin plate splines. Indicates the first Three-dimensional points obtained from a single perspective. Indicates the first Individual perspective deformation weights This represents the three-dimensional points output by the pre-trained convolutional neural network (CNN).
6. The augmented reality-guided dental sculpting learning device according to claim 3, characterized in that: During the sculpting process, the rendering and augmented reality overlay unit provides dynamic feedback by combining real-time depth information and lighting models to help adjust the sculpting path. The depth information calculation expression is as follows: in, This represents the shortest Euclidean distance between a pixel and the edge of the sculpted content. )≥0; ln() is the natural logarithm function; depth(x,y) is the normalized depth information; the lighting model adopts the Phong lighting model, and the expression is as follows: in, , , These represent ambient light, diffuse light, and specular light, respectively. This represents the lighting rendering value; The specific expression for adjusting the carving path is as follows: in, This represents the three-dimensional error vector of the sculpted area. Xactual refers to the actual three-dimensional form sculpted. Represents the standard model; This is the inverse of the camera intrinsic parameter matrix; , s is the rotation matrix and translation vector; s is the scale factor; This is an adjustment amount for the engraving path on a two-dimensional image plane, used to guide the operator in correcting the engraving direction.