Method and device for collecting 3D scanning image before dental cleaning

Through 3D scanning image acquisition method and PS-OCT technology, combined with deep learning network, the dental image is automatically analyzed, which solves the problem of tedious dental images manually analyzing dental images, achieving more accurate calculus detection and judgment, and improving the efficiency of dental examination .

CN119014814AInactive Publication Date: 2024-11-26WATERCARE TECH (WENZHOU) CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410915258.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-09
Publication Date
2024-11-26
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

In the prior art, dentists need to manually analyze patient teeth images to determine whether teeth cleaning is needed. The process is cumbersome and there is a risk of misdiagnosis of small lesions. How to accurately segment and diagnose tartars becomes a challenge.

Method used

Using the 3D scan image acquisition method, the PS-OCT system emits near-infrared light of different wavelengths to collect teeth images, perform denoising and contour extraction, and combines Stokes parameters and deep convolution neural networks and long and short-term memory networks to automatically Analyze the coverage of tartars and determine whether cleaning is needed.

Benefits of technology

Provide more accurate dental structure images, reduce the dentist's dependence on subjective judgment, improve the accuracy and efficiency of calculus detection, reduce the risk of missed diagnosis, and save the dentist's diagnosis time.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119014814A_ABST
    Figure CN119014814A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of image processing, in particular to a 3D scanning image acquisition method and device before dental cleaning. Comprising the following steps: emitting near-infrared light with different wavelengths through a PS-OCT system to collect different tooth images; performing denoising processing on the tooth image to generate initial contours of teeth and dental calculus; comparing the initial contours of the teeth and the dental calculus, if the coverage range of the dental calculus exceeds a preset value, sending out a cleaning notice, and adopting 3D scanning and PS-OCT technologies to provide detail images of the tooth structure and the dental calculus which are more accurate than those of the traditional 2D X-ray; the higher image accuracy is beneficial to the determination of the size of a dental calculus area, and the judgment on whether cleaning is needed or not is automatically realized.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a method and device for acquiring 3D scanning images before dental cleaning. Background Art

[0002] As part of the "facade" project, teeth are naturally very important. The World Health Organization attaches great importance to oral health care and has set "clean teeth, no cavities, no pain, normal gum color, and no bleeding" as the standard for oral health. In recent years, with the continuous improvement of people's living standards, the awareness of the importance of dental care has been continuously enhanced, and "teeth washing" has been accepted by more and more people; teeth washing is oral supragingival cleaning. Its treatment method is to use an oral ultrasonic vibration tooth cleaning machine, which vibrates up and down after contacting with dental calculus and plaque, and removes them through mechanical stimulation, reducing the probability of periodontal disease; if you do not wash your teeth frequently, you may suffer from periodontal disease, which is an infection of the gum tissue and bones, which will cause teeth to shift and is the culprit for tooth loss in adults. If diagnosed early, periodontal disease can be treated and reversed. If not treated, it will lead to more serious gum disease later. Regular teeth washing and examination can effectively prevent gum disease. Dental calculus is a common oral disease. It is calcified or calcified plaque or soft scale deposited on the teeth. If it is not removed in time, it will constantly irritate the periodontal tissue and compress the gums, leading to oral diseases such as periodontitis and gingivitis. Therefore, a comprehensive understanding of the condition of dental calculus is very important for clinicians to formulate treatment plans. Relying on dentists to analyze patients' oral images is a tedious and time-consuming task, and due to factors such as the different shapes of dental calculus and visual fatigue, the potential risk of missing small lesions will increase. Therefore, the use of automatic detection and segmentation of dental calculus to assist treatment and improve the accuracy of screening has important research value and practical application prospects. However, how to accurately segment dental calculus from oral images and diagnose and evaluate it has become a challenging problem.

[0003] At present, before teeth cleaning, doctors usually check the patient's tartar condition and subjectively judge whether the patient needs teeth cleaning. In order to reduce the workload of doctors, the present invention proposes a 3D scanning image acquisition method and device before dental cleaning. When the patient undergoes dental examination, the tooth image is analyzed and the tartar is extracted to determine whether teeth cleaning is needed. Summary of the invention

[0004] The present invention aims at the technical problems existing in the prior art and provides a 3D scanning image acquisition method and device before dental cleaning.

[0005] The technical solution of the present invention to solve the above technical problems is as follows: a method and device for collecting 3D scanning images before dental cleaning;

[0006] A 3D scanning image acquisition method before dental cleaning, comprising the following steps:

[0007] The PS-OCT system emits near-infrared light of different wavelengths to collect different dental images;

[0008] De-noise the tooth image to generate the initial outline of the teeth and tartar;

[0009] The initial contours of the teeth and tartar are compared and if the coverage of tartar exceeds the preset value, a cleaning notification is issued.

[0010] Furthermore, the 3D scanning image acquisition method before dental cleaning in the PS-OCT system includes:

[0011] OCT intensity images and registered phase delay, optical axis and Stokes parameters Q, U, V data are acquired simultaneously from cross-sectional scans, where the delay is the difference in the optical path length between two orthogonal linear polarization states:

[0012]

[0013] Where I0 is the intensity of vertically polarized light, I1 is the intensity of horizontally polarized light; and the optical axis and Stokes parameters are:

[0014]

[0015] where r0 and r1 are the complex reflectances of vertically polarized light and horizontally polarized light as a function of depth, respectively.

[0016] Furthermore, the 3D scanning image acquisition method before dental cleaning averages the Stokes vectors between adjacent pixels by calculating the Gaussian weighted average of the Stokes parameters, and the Stokes vector is closely related to the degree of polarization:

[0017]

[0018] Among them, Q m*m , U m*m , V m*m It represents the Stokes parameter value after Gaussian weighted averaging, m represents the size of the average sum, and when the DOPU value is 1, it means that the polarization states of adjacent pixel positions are consistent. When it is less than 1, it means that polarization disturbance exists.

[0019] Furthermore, in the method for acquiring 3D scanning images before dental cleaning, the step of performing denoising on the tooth image and generating the initial contours of the teeth and dental calculus includes:

[0020] Initialization: First, initialize the image and determine the initial value of the active contour;

[0021] Calculate energy function: Construct an energy function, part of which is provided by the image itself and part of which is provided by active contours;

[0022] Optimization: Use optimization algorithms to minimize the energy function;

[0023] Update active contours: Continuously update the position of active contours through an iterative algorithm;

[0024] Output: The final updated result is output as a binary image, where the pixels inside the contour are marked as 1 and the pixels outside the contour are marked as 0.

[0025] Furthermore, the method for collecting 3D scanning images before dental cleaning includes, in the step of calculating the energy function:

[0026] Perform two-dimensional discrete wavelet transform on the image and decompose it into a low-frequency molecular weight image and three high-frequency component sub-images. The three high-frequency component sub-images include texture information in the horizontal, vertical and diagonal directions respectively. The sum of the energy values ​​of the three high-frequency component sub-images constitutes a composite image.

[0027] W CB (x,y)=|LH CB (x,y) 2 +|HL CB (x,y) 2 +HH CB (x,y) 2 ;

[0028] W Fn (x,y)=|LH Fn (x,y) 2 +|HL Fn (x,y) 2 +HH Fn (x,y) 2 ;

[0029] W CB (x,y)=W Fn (x,y),if(object(x,y)=0);

[0030] Where LH, HL, and HH represent three high-frequency molecular images, respectively. CB (x, y) is the energy value of the current image itself, W Fn(x, y) The energy value of the current active contour, set a threshold difference W, if the difference between the energy value of the current background image and the energy value of the current active contour is greater than the threshold difference W, iterate the active contour.

[0031] Furthermore, the 3D scanning image acquisition method before dental cleaning provides a formula to describe the evolution process of the active contour curve;

[0032] For a given active contour, the partial calculus equation for the evolution of the entire active contour within the time range t is:

[0033]

[0034] Where V(C) is the velocity of change, N is the normal vector, , d is the calculus parameter of the active contour, and C(p,t) is the active contour.

[0035] Furthermore, in the method for acquiring 3D scanning images before dental cleaning, the level set function φ(x, y) is a surface function representing contour lines, φ=0 is a zero level set, and time t is introduced in the evolution process, and the continuous function φ(x, y, t) is the variation form of the active contour C(p, t) at time t. The active contour evolution process is:

[0036]

[0037] The zero level set of the active contour of φ at time t corresponds to:

[0038] φ(C(t),t)=0;

[0039] Performing total differential operation yields:

[0040]

[0041] Furthermore, in the method for acquiring 3D scanning images before dental cleaning, in the step of acquiring different tooth images by emitting near-infrared light of different wavelengths through the PS-OCT system, the different tooth images include tooth images without tartar and tooth images with tartar, and the tooth images without tartar are extracted using 1310nm near-infrared light.

[0042] Furthermore, the 3D scanning image acquisition method before dental cleaning calculates the near-infrared light of the calculus image and matches the wavelength of the near-infrared light according to the bacterial agglomerates and calcium content in the calculus:

[0043] The different proportions of bacterial clumps and calcium contents were input into a convolutional long short-term memory network constructed by combining a deep convolutional neural network with a long short-term memory network.

[0044] Divide into training data set and test data set in a ratio of 8:2;

[0045] Complete model training of the convolutional neural network model on the training set;

[0046] Load the trained model weights to test model performance;

[0047] After the test performance passes, the neural network model is deployed to the PS-OCT system.

[0048] A 3D scanning image acquisition device before dental cleaning, applied to the 3D scanning image acquisition method before dental cleaning, the device comprising:

[0049] Scanning unit: The PS-OCT system emits near-infrared light of different wavelengths to collect different tooth images;

[0050] Processing unit: denoise the tooth image and generate the initial contours of the teeth and tartar;

[0051] Comparison unit: Compare the initial contours of teeth and tartar. If the coverage of tartar exceeds the preset value, a cleaning notification is issued. Otherwise, the surface teeth do not need to be cleaned.

[0052] The beneficial effects of the present invention are:

[0053] The use of 3D scanning and PS-OCT technology can provide more accurate detailed images of tooth structure and tartar than traditional 2D X-rays. The higher image accuracy helps to determine the size of the tartar area and automatically determine whether cleaning is needed, reducing dentists' reliance on their experience and subjective judgment during the diagnosis process and potentially saving them time in busy clinical environments.

[0054] Displaying detailed 3D images can help patients better understand their oral health status, strengthen communication between doctors and patients, enhance patients' self-care awareness, and prompt them to pay more attention to oral hygiene habits and regular check-ups. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 A schematic diagram of a process of a 3D scanning image acquisition method before dental cleaning according to the present invention;

[0056] Figure 2 A schematic diagram of a 3D scanning image acquisition device before dental cleaning according to the present invention;

[0057] Figure 3 FIG. 4 is a flow chart of image processing in one embodiment of the present invention. DETAILED DESCRIPTION

[0058] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of this application.

[0059] In the description of this application, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of this application, "plurality" means two or more, unless otherwise clearly and specifically defined.

[0060] In the description of the present application, the term "for example" is used to mean "used as an example, illustration or description". Any embodiment described as "for example" in the present application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any technician in the field to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art can recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes will not be elaborated in detail to avoid unnecessary details to obscure the description of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed in the present application.

[0061] A 3D scanning image acquisition method before dental cleaning, comprising the following steps:

[0062] S1: The PS-OCT system emits near-infrared light of different wavelengths to collect different tooth images;

[0063] S2: De-noising the tooth image to generate the initial contours of the teeth and dental calculus;

[0064] S3: Compare the initial contours of the teeth and tartar. If the coverage of the tartar exceeds the preset value, a cleaning notification is issued.

[0065] In this embodiment, PS-OCT technology is used to image teeth. PS-OCT is a non-invasive imaging technology that uses near-infrared light (usually light with a wavelength of about 1300nm, which is transparent to teeth) to obtain structural information inside teeth. This system can recognize different wavelengths, so it can distinguish different components of teeth, such as healthy tooth tissue and tartar (tartar formed after hardening of tartar). Each tissue reacts differently to light of different wavelengths, so by analyzing the reflection or scattering of these lights, different structures inside teeth can be identified. Once the original scanned image of the teeth is obtained, the next step is to remove the noise in the image. Noise may come from many aspects, including noise from the equipment itself, blurring caused by patient movement, or light interference during imaging. Denoising is a common step in image processing, which is used to improve image quality and make the boundaries between teeth and tartar clearer. Clear boundaries help the subsequent steps to more accurately extract the contours of teeth and tartar. After extracting the initial contours, the contours of teeth and tartar need to be compared and analyzed. This step involves quantitative analysis, the purpose of which is to measure the extent to which tartar covers the teeth. If it is found that the tartar coverage exceeds the set standard, the system will make a cleaning recommendation, that is, notify the dentist or patient that professional dental cleaning is needed. This quantitative method can help dentists determine the severity of tartar and make decisions on whether immediate intervention is needed. The above method provides dental professionals with an automated tool to evaluate the condition of tartar before dental cleaning operations and prompt intervention when necessary. This method not only improves the efficiency of dental examinations, but also helps prevent dental problems in advance and ensure better oral hygiene.

[0066] In one embodiment, the PS-OCT system includes:

[0067] OCT intensity images and registered phase delay, optical axis and Stokes parameters Q, U, V data are acquired simultaneously from cross-sectional scans, where the delay is the difference in the optical path length between two orthogonal linear polarization states:

[0068]

[0069] Where I0 is the intensity of vertically polarized light, I1 is the intensity of horizontally polarized light; and the optical axis and Stokes parameters are:

[0070]

[0071]

[0072] where r0 and r1 are the complex reflectances of vertically polarized light and horizontally polarized light as a function of depth, respectively.

[0073] The Stokes vectors between adjacent pixels are averaged by computing the Gaussian-weighted average of the Stokes parameters, which are closely related to the degree of polarization:

[0074]

[0075] Among them, Q m*m , U m*m , V m*m It represents the Stokes parameter value after Gaussian weighted averaging, m represents the size of the average sum, and when the DOPU value is 1, it means that the polarization states of adjacent pixel positions are consistent. When it is less than 1, it means that polarization disturbance exists.

[0076] In this embodiment, in the PS-OCT system, OCT intensity images and Stokes parameters are simultaneously acquired by cross-sectional scanning. The OCT intensity image provides a two-dimensional image of the tooth structure based on the intensity of the reflected or scattered light. The Stokes parameters (Q, U, V) describe the change in polarization state, and these parameters can provide polarization-sensitive information, such as the polarization characteristics of tooth tissue. Phase delay refers to the difference in optical path length between two orthogonal linear polarization states. In this embodiment, the phase delay calculation involves light intensity, which is the light intensity of vertically polarized light and horizontally polarized light. In the actual calculation process, the light intensity difference is used to determine the optical path difference. The optical axis and Stokes parameters (Q), (U), (V) are used to describe the state of polarized light. These parameters are based on the complex reflectivity of vertically and horizontally polarized light. The complex reflectivity contains information about amplitude and phase, which can be used to describe how light waves interact with tooth tissue. In order to more accurately estimate the polarization state between pixels, the Gaussian weighted average of the Stokes parameters (Q), (U), and (V) is calculated. This Gaussian weighted average helps to smooth pixel values, reduce the impact of local noise, and make the polarization information in the scanned image more continuous. The degree of polarization uniformity (DOPU) is a measurement parameter that reflects the consistency of the polarization state between adjacent pixels. If the DOPU value is 1, it means that the polarization state between adjacent pixels is completely consistent, while a DOPU value less than 1 indicates that there is polarization disturbance, which may be caused by the characteristics of the internal tooth tissue.

[0077] In one embodiment, the step of performing denoising on the tooth image to generate the initial contours of the teeth and dental calculus includes:

[0078] S21: Initialization: First, the image is initialized and the initial value of the active contour is determined; the initial stage involves estimating the approximate location of the contour of the object of interest in the image. This initial value setting is crucial for the subsequent contour evolution because it provides a starting point. For example, in dental image processing, the active contour may be specified around the area where tartar is estimated to exist based on the dentist's expertise.

[0079] S22: Calculate Energy Function: Construct an energy function, partly from information provided by the image itself and partly from information provided by active contours; the energy function combines information about the image data and the shape of the contour. The function contains multiple terms designed to capture important features of the image, such as edge strength, and the smoothness and continuity of the contour. The energy function usually includes internal energy (related to contour smoothness) and external energy (related to the image data).

[0080] S23: Optimization: Use an optimization algorithm to minimize the energy function; Use the optimization algorithm to minimize the energy function to adjust the position of the contour. This optimization process causes the contour to be attracted to the real boundary of the tooth and tartar in the image.

[0081] S24: Update active contour: The position of the active contour is continuously updated through an iterative algorithm; using an iterative algorithm such as gradient descent, the contour will gradually move in the image until the energy is minimized. The position of the contour at this time represents the edge of the tooth and tartar in the image.

[0082] S25: Output result: Output the final update result as a binary image, in which the pixels inside the contour are marked as 1, and the pixels outside the contour are marked as 0.

[0083] The steps in calculating the energy function include:

[0084] Perform two-dimensional discrete wavelet transform on the image and decompose it into a low-frequency molecular weight image and three high-frequency component sub-images. The three high-frequency component sub-images include texture information in the horizontal, vertical and diagonal directions respectively. The sum of the energy values ​​of the three high-frequency component sub-images constitutes a composite image.

[0085] W CB (x,y)=|LH CB (x,y) 2 +|HL CB (x,y) 2 +|HH CB (x,y) 2 ;

[0086]

[0087] Where LH, HL, and HH represent three high-frequency molecular images, respectively.CB (x,y) is the energy value of the current image itself, The energy value of the current active contour sets a threshold difference W. If the difference between the energy value of the current background image and the energy value of the current active contour is greater than the threshold difference W, the active contour is iterated.

[0088] In one embodiment, the evolution process of the active contour curve is described by a formula;

[0089] For a given active contour, the partial calculus equation for the evolution of the entire active contour within the time range t is:

[0090]

[0091] Where V(C) is the velocity of change, N is the normal vector, , d is the calculus parameter of the active contour, and C(p,t) is the active contour.

[0092] The level set function φ(x, y) is a surface function representing the contour line. When φ=0, it is a zero level set. When time t is introduced in the evolution process, the continuous function φ(x, y, t) is the variation form of the active contour C(p, t) at time t. The evolution process of the active contour is:

[0093]

[0094] The zero level set of the active contour of φ at time t corresponds to:

[0095] φ(C(t),t)=0;

[0096] Performing total differential operation yields:

[0097]

[0098] In another embodiment, the evolution of the active contour can be described by the following partial differential equation (PDE):

[0099] [\frac{\partial\vec{X}}{\partial t}=\alpha\cdot\vec{N}\cdot(\kappa\cdot\vec{N})+\beta\cdot\vec{N}\cdot g]

[0100] This equation means that the movement of each point on the contour line is composed of two parts: the internal force is controlled by the curvature and smoothness parameters, and the external force is determined by the image edge function (g). This setting allows the contour to move along the edge of the image while maintaining smoothness.

[0101] The level set method is a numerical technique used to solve the evolution of curves and surfaces with various topologies. This method can be described by the following equation:

[0102] [\frac{\partial\phi}{\partial t}+\vec{V}\cdot\nabla\phi=0]

[0103] where (\phi) is the level set function, which implicitly defines the contour (e.g., when (\phi = 0), the contour is the zero level set).

[0104] The evolution equation states that the evolution of the contour is a function of time (t), and the inner product of the velocity (\vec{V}) and the gradient (\nabla\phi) results in a (\phi) function that changes over time. This allows contours to evolve freely in the image, including splitting and merging, without maintaining continuity or closure.

[0105] By solving these equations, we can track and extract the boundaries of objects of interest in digital images (such as the boundaries of teeth and tartar in dental images). This process is iterative and will eventually converge to the precise outline of the tracked object.

[0106] Furthermore, in the step of collecting different dental images by emitting near-infrared light of different wavelengths through the PS-OCT system, the different dental images include dental images without tartar and dental images with tartar, and the dental images without tartar are extracted using 1310nm near-infrared light.

[0107] Furthermore, the near-infrared light of the calculus image is calculated, and the wavelength of the near-infrared light is matched according to the bacterial clusters and calcium content in the calculus:

[0108] The different proportions of bacterial clumps and calcium contents were input into a convolutional long short-term memory network constructed by combining a deep convolutional neural network with a long short-term memory network.

[0109] Divide into training data set and test data set in a ratio of 8:2;

[0110] Complete model training of the convolutional neural network model on the training set;

[0111] Load the trained model weights to test model performance;

[0112] After the test performance passes, the neural network model is deployed to the PS-OCT system.

[0113] In one embodiment, a 3D scanning image acquisition device before dental cleaning is applied to the 3D scanning image acquisition method before dental cleaning, and the device comprises:

[0114] Scanning unit: The PS-OCT system emits near-infrared light of different wavelengths to collect different tooth images;

[0115] Processing unit: denoise the tooth image and generate the initial contours of the teeth and tartar;

[0116] Comparison unit: Compare the initial contours of teeth and tartar. If the coverage of tartar exceeds the preset value, a cleaning notification is issued. Otherwise, the surface teeth do not need to be cleaned.

[0117] In this embodiment, a hybrid model is first established, which combines a deep convolutional neural network (CNN) and a long short-term memory network (LSTM). CNN has a significant effect in the field of image recognition and can extract the spatial features of images. At the same time, LSTM is better at processing sequence data and can take into account the contextual information before and after the image processing process. The input of the hybrid model is an image signal adjusted according to different proportions of bacterial clumps and calcium content. A sufficient number of annotated dental calculus image data are collected, and the data set is divided into a training set and a test set at a ratio of 8:2. The training set is used for model learning, and the test set is used to evaluate the generalization ability of the model. The model is trained using the training data set. In this process, the model will optimize the internal parameters through continuous iteration so that the output results (for example, light sensitivity to different wavelengths) best match the features in the input image signal (the distribution of bacterial clumps and calcium content). After the training is completed, the model is evaluated for performance on the test data set. This step is critical, and it helps to confirm the generalization ability of the model, that is, the processing effect of the model on unseen data. If the model performs satisfactorily on the test set, it will be deployed to the actual PS-OCT system. This means that in actual applications, when teeth are scanned by the PS-OCT system, this deep learning model will automatically calculate the specific wavelength of near-infrared light that is most suitable for dental plaque. The powerful capabilities of deep learning in image processing and pattern recognition are combined with traditional spectral analysis technology to improve the accuracy of dental plaque detection.

[0118] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0119] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A 3D scanning image acquisition method before dental cleaning, It is characterized in that The following steps are involved: The PS-OCT system emits near-infrared light of different wavelengths to collect different dental images; De-noise the tooth image to generate the initial outline of the teeth and tartar; The initial contours of the teeth and tartar are compared and if the coverage of tartar exceeds the preset value, a cleaning notification is issued.

2. A 3D scanning image acquisition method before dental cleaning according to claim 1, It is characterized in that The PS-OCT system includes: OCT intensity images and registered phase delay, optical axis and Stokes parameters Q, U, V data are acquired simultaneously from cross-sectional scans, where the delay is the difference in the optical path length between two orthogonal linear polarization states: Among them I 0 is the intensity of vertically polarized light, I 1 is the intensity of horizontally polarized light; while the optical axis and Stokes parameters are: Among them, r 0 and r 1 are the complex reflectances of vertically polarized light and horizontally polarized light as a function of depth, respectively.

3. The 3D scanning image acquisition method before dental cleaning according to claim 2, It is characterized in that The Stokes vectors between adjacent pixels are averaged by computing the Gaussian-weighted average of the Stokes parameters, which are closely related to the degree of polarization: Among them, Q m*m , U m*m , V m*m It represents the Stokes parameter value after Gaussian weighted averaging, m represents the size of the average sum, and when the DOPU value is 1, it means that the polarization states of adjacent pixel positions are consistent. When it is less than 1, it means that polarization disturbance exists.

4. The 3D scanning image acquisition method before dental cleaning according to claim 1, It is characterized in that The step of performing denoising on the tooth image and generating the initial contours of the teeth and dental calculus comprises: Initialization: First, initialize the image and determine the initial value of the active contour; Calculate energy function: Construct an energy function, part of which is provided by the image itself and part of which is provided by active contours; Optimization: Use optimization algorithms to minimize the energy function; Update active contours: Continuously update the position of active contours through an iterative algorithm; Output: The final updated result is output as a binary image, where the pixels inside the contour are marked as 1 and the pixels outside the contour are marked as 0.

5. The 3D scanning image acquisition method before dental cleaning according to claim 4, It is characterized in that The steps in calculating the energy function include: Perform two-dimensional discrete wavelet transform on the image and decompose it into a low-frequency molecular weight image and three high-frequency component sub-images. The three high-frequency component sub-images include texture information in the horizontal, vertical and diagonal directions respectively. The sum of the energy values ​​of the three high-frequency component sub-images constitutes a composite image. W CB (x,y)=|LH CB (x,y) 2 +|HL CB (x,y) 2 +|HH CB (x,y) 2 ; Where LH, HL, and HH represent three high-frequency molecular images, respectively. CB (x, y) is the energy value of the current image itself, W Fn (x, y) The energy value of the current active contour, set a threshold difference W, if the difference between the energy value of the current background image and the energy value of the current active contour is greater than the threshold difference W, iterate the active contour.

6. The 3D scanning image acquisition method before dental cleaning according to claim 4, It is characterized in that Formulate a description of the evolution of the active contour curve; For a given active contour, the partial calculus equation for the evolution of the entire active contour within the time range t is: Where V(C) is the velocity of change, N is the normal vector, , d is the calculus parameter of the active contour, and C(p,t) is the active contour.

7. The 3D scanning image acquisition method before dental cleaning according to claim 6, It is characterized in that The level set function φ(x, y) is a surface function representing the contour line. When φ=0, it is a zero level set. When time t is introduced in the evolution process, the continuous function φ(x, y, t) is the variation form of the active contour C(p, t) at time t. The evolution process of the active contour is: The zero level set of the active contour of φ at time t corresponds to: φ(C(t),t)=0; Performing total differential operation yields:

8. The 3D scanning image acquisition method before dental cleaning according to claim 1, It is characterized in that In the step of collecting different dental images by emitting near-infrared light of different wavelengths through the PS-OCT system, the different dental images include dental images without tartar and dental images with tartar, and the dental images without tartar are extracted using 1310nm near-infrared light.

9. The 3D scanning image acquisition method before dental cleaning according to claim 8, It is characterized in that The near-infrared light of the calculus image is calculated, and the wavelength of the near-infrared light is matched according to the bacterial clusters and calcium content in the calculus: The different proportions of bacterial clumps and calcium contents were input into a convolutional long short-term memory network constructed by combining a deep convolutional neural network with a long short-term memory network. Divide into training data set and test data set in a ratio of 8:2; Complete model training of the convolutional neural network model on the training set; Load the trained model weights to test model performance; After the test performance passes, the neural network model is deployed to the PS-OCT system.

10. A 3D scanning image acquisition device before dental cleaning, It is characterized in that A 3D scanning image acquisition method before dental cleaning applied to any one of claims 1 to 9, the device comprising: Scanning unit: The PS-OCT system emits near-infrared light of different wavelengths to collect different tooth images; Processing unit: denoise the tooth image and generate the initial contours of the teeth and tartar; Comparison unit: Compare the initial contours of teeth and tartar. If the coverage of tartar exceeds the preset value, a cleaning notification is issued. Otherwise, the surface teeth do not need to be cleaned.