A dental surface detection method, device, equipment and readable storage medium

Abstract

The application provides a tooth surface detection method, device and equipment and a readable storage medium. The method comprises the following steps: controlling an image sensor to image tooth surface fluorescence spectra transmitted by different optical filters respectively to obtain multiple tooth surface fluorescence images; connecting the multiple tooth surface fluorescence images in a channel dimension to obtain a multi-channel tooth surface fluorescence image; combining the multi-channel tooth surface fluorescence image and a trained tooth surface detection neural network to obtain a tooth surface detection result; and the tooth surface detection result comprises a tooth surface pixel point abnormal type and a corresponding pixel point position. Through the implementation of the application, the fluorescence spectrum is divided into bands and then imaged on the image sensor to obtain a multi-channel fluorescence image, and then a neural network model is used to realize the classification and identification of different tooth surface abnormal states, thereby improving the accuracy of tooth surface health diagnosis and effectively reducing the diagnosis cost.

Description

Technical Field

[0001] This application relates to the field of image processing technology, and in particular to a method, apparatus, device and readable storage medium for tooth surface detection. Background Technology

[0002] Accurate early detection of dental caries can effectively prevent disease progression and even reverse it. Traditional clinical methods for caries detection include visual inspection, probes, and X-rays, but these are subjective, invasive, and involve radiation, and are not always effective in detecting early caries. Caries detection methods based on fluorescence spectroscopy and fluorescence imaging offer advantages over traditional methods, including objectivity, accuracy, and zero radiation. While some commercial products exist, such as DIAGNOdent, they are relatively expensive and not suitable for community or home-based early caries screening.

[0003] The mechanism of detecting dental caries based on fluorescence spectroscopy and fluorescence imaging technology is to detect the difference in fluorescence spectra between normal tooth surfaces and carious teeth. Furthermore, as dental caries progresses, the content of bacterial metabolites porphyrins in teeth at different stages gradually increases, and the fluorescence spectrum also changes. However, traditional single-point fluorescence spectroscopy or three-channel fluorescence imaging cannot accurately distinguish the subtle differences in the fluorescence spectra of teeth at different stages of caries.

[0004] Dental calculus consists of calcified deposits on the tooth surface, containing a large amount of plaque. It also emits a fluorescence spectrum similar to that of cavities. When both calculus and cavities are present on the tooth surface, traditional fluorescence spectroscopy techniques struggle to distinguish the subtle differences in their spectra. Therefore, the detection methods offered by these technologies cannot accurately identify different abnormalities on the tooth surface, resulting in low precision in diagnosing tooth health. Summary of the Invention

[0005] This application provides a tooth surface detection method, apparatus, device, and readable storage medium, which can at least solve the problem that the detection methods provided in the related art cannot accurately identify different abnormalities on the tooth surface, resulting in low accuracy in tooth surface health diagnosis.

[0006] The first aspect of this application provides a method for detecting tooth surfaces, including:

[0007] The image sensors are controlled to image the fluorescence spectrum of the tooth surface transmitted through different filters, resulting in multiple fluorescence images of the tooth surface; wherein the transmission wavelengths of the different filters are different.

[0008] Multiple tooth surface fluorescence images are concatenated along the channel dimension to obtain a multi-channel tooth surface fluorescence image;

[0009] By combining the multi-channel tooth surface fluorescence image and the trained tooth surface detection neural network, tooth surface detection results are obtained; wherein, the tooth surface detection results include the abnormality type of tooth surface pixels and the corresponding pixel position.

[0010] A second aspect of this application provides a tooth surface inspection device, comprising:

[0011] An imaging module is used to control the image sensor to image the fluorescence spectrum of the tooth surface transmitted through different filters, thereby obtaining multiple fluorescence images of the tooth surface; wherein, the transmission wavelengths of the different filters are different;

[0012] A connection module is used to connect multiple tooth surface fluorescence images in the channel dimension to obtain a multi-channel tooth surface fluorescence image;

[0013] The acquisition module is used to combine the multi-channel tooth surface fluorescence image and the trained tooth surface detection neural network to obtain tooth surface detection results; wherein, the tooth surface detection results include the abnormality type of tooth surface pixels and the corresponding pixel position.

[0014] A third aspect of this application provides an electronic device, including a memory and a processor, wherein the processor is used to execute a computer program stored in the memory, and when the processor executes the computer program, it implements the steps in the tooth surface detection method provided in the first aspect of this application.

[0015] The fourth aspect of this application provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, it implements the steps of the tooth surface detection method provided in the first aspect of this application.

[0016] As can be seen from the above, according to the tooth surface detection method, device, equipment, and readable storage medium provided in this application, the image sensor is controlled to image the tooth surface fluorescence spectrum transmitted through different filters, resulting in multiple tooth surface fluorescence images. These multiple tooth surface fluorescence images are then connected along the channel dimension to obtain a multi-channel tooth surface fluorescence image. Combining the multi-channel tooth surface fluorescence image with the trained tooth surface detection neural network, the tooth surface detection result is obtained. The tooth surface detection result includes the type of abnormality of tooth surface pixels and the corresponding pixel location. Through the implementation of this application, the fluorescence spectrum is segmented into bands and imaged on the image sensor to obtain a multi-channel fluorescence image. Then, a neural network model is used to classify and identify different tooth surface abnormalities, improving the accuracy of tooth surface health diagnosis and effectively reducing diagnostic costs. Attached Figure Description

[0017] Figure 1 This is a basic flowchart of a tooth surface inspection method provided in the first embodiment of this application;

[0018] Figure 2 A schematic diagram illustrating an imaging method provided in the first embodiment of this application;

[0019] Figure 3 A schematic diagram of the transmission spectrum of a filter provided in the first embodiment of this application;

[0020] Figure 4 A schematic diagram of the fluorescence spectrum after passing through a filter, provided for the first embodiment of this application;

[0021] Figure 5 A schematic diagram illustrating another imaging method provided in the first embodiment of this application;

[0022] Figure 6 A detailed flowchart illustrating a tooth surface inspection method provided in the second embodiment of this application;

[0023] Figure 7 This is a schematic diagram of the program modules of the tooth surface detection device provided in the third embodiment of this application;

[0024] Figure 8 This is a schematic diagram of the structure of an electronic device provided in the fourth embodiment of this application. Detailed Implementation

[0025] To make the inventive objectives, features, and advantages of this application more apparent and understandable, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0026] In the description of the embodiments of this application, it should be understood that the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings. They are only for the convenience of describing the embodiments of this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limiting the present invention.

[0027] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of the embodiments of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0028] In the embodiments of this application, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. For those skilled in the art, the specific meaning of the above terms in the embodiments of this application can be understood according to the specific circumstances.

[0029] The above description is merely a preferred embodiment of this application and is not intended to limit the invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the protection scope of this application.

[0030] To address the problem that existing detection methods in related technologies cannot accurately identify different abnormalities on the tooth surface, the first embodiment of this application provides a tooth surface detection method, such as... Figure 1 This is a basic flowchart of the tooth surface inspection method provided in this embodiment. The tooth surface inspection method includes the following steps:

[0031] Step 101: Control the image sensor to image the fluorescence spectrum of the tooth surface transmitted through different filters to obtain multiple fluorescence images of the tooth surface.

[0032] Specifically, in this embodiment, an excitation light source (e.g., an ultraviolet or near-infrared light source) irradiates the tooth surface, exciting it to generate a fluorescence spectrum. Different types of tooth surface abnormalities (e.g., dental caries, tartar, etc.) exhibit different fluorescence spectra. Different filters in this embodiment have different transmission wavelengths; the function of the filters is to segment the fluorescence spectrum into bands before imaging it on the image sensor. It should also be noted that the fluorescence image in this embodiment can be either an RGB image or an xyY image; the appropriate format can be flexibly selected based on detection requirements in practical applications, and this embodiment does not impose a unique limitation on this.

[0033] In this embodiment, multi-channel fluorescence images can be obtained through different combinations of filters and sensors. It should be noted that the specific implementation methods for controlling the image sensor to image the tooth surface fluorescence spectrum transmitted through different filters to obtain multiple tooth surface fluorescence images are, but are not limited to, the following two:

[0034] Method 1 involves controlling the same image sensor to image the fluorescence spectrum of the tooth surface transmitted through different filters in a time-division manner, thereby obtaining multiple fluorescence images of the tooth surface.

[0035] In a preferred embodiment, the filter includes a first filter and a second filter. Thus, in this embodiment, at a first moment, the image sensor is controlled to image the tooth surface fluorescence spectrum transmitted through the first filter to obtain a first tooth surface fluorescence image; then, at a second moment, the image sensor is controlled to image the tooth surface fluorescence spectrum transmitted sequentially through the first filter and the second filter to obtain a second tooth surface fluorescence image; wherein, the optical axes of the first filter and the second filter coincide.

[0036] like Figure 2 The diagram illustrates an imaging method provided in this embodiment. Initially, the first filter is coaxial with the image sensor. The fluorescence spectrum generated by the light source module on the tooth surface passes through the first filter (i.e., filter 1) and is then imaged by the image sensor to obtain a first tooth surface fluorescence image. Next, at the next moment, the second filter is moved to a position coaxial with the first filter, and the fluorescence spectrum is excited again, allowing it to pass through the first filter and the second filter (i.e., filter 1 and filter 2) respectively, and then imaged by the image sensor to obtain a second tooth surface fluorescence image. In other words, this embodiment achieves band-segmented imaging through a serial imaging method.

[0037] like Figure 3 The diagram shown is a schematic of the transmission spectrum of a filter provided in this embodiment. Figure 3 The left image shows the transmission spectrum of the first filter, and the right image shows the transmission spectrum of the second filter. The transmission wavelengths of the first and second filters are approximately 470 nm and 600 nm, respectively. Figure 4 The image shown is a schematic diagram of the fluorescence spectrum after passing through a filter, as provided in this embodiment. Figure 4 The left image shows the fluorescence spectrum after passing through only the first filter, while the right image shows the fluorescence spectrum after passing through the first and second filters in sequence. The fluorescence spectrum after passing through the first filter has a spectral range of 470nm-780nm, and the fluorescence spectrum after passing through the first filter has a spectral range of 600nm-780nm, which are divided into two bands for imaging.

[0038] Method 2 involves controlling different image sensors to simultaneously image the fluorescence spectrum of the tooth surface transmitted through the corresponding filters, thereby obtaining multiple fluorescence images of the tooth surface.

[0039] Specifically, unlike method one, this implementation method achieves band-segmented imaging through parallel imaging. During image acquisition, it does not require moving the filters. Taking two filters as an example, that is, an image sensor is set up for each of the first and second filters, such as... Figure 5 The diagram shows another imaging method provided in this embodiment. The fluorescence spectrum generated by the light source module on the tooth surface passes in parallel through the first filter (i.e., filter 1) and the second filter (i.e., filter 2), and is then imaged by image sensor 1 and image sensor 2 respectively to obtain the first tooth surface fluorescence image and the second tooth surface fluorescence image.

[0040] Step 102: Connect multiple tooth surface fluorescence images in the channel dimension to obtain a multi-channel tooth surface fluorescence image.

[0041] Specifically, traditional single-point fluorescence spectroscopy or three-channel fluorescence imaging cannot accurately distinguish the subtle differences in fluorescence spectra of different stages of caries, caries and tartar. In this embodiment, multiple fluorescence images acquired by combinations of different sensors and filters are connected in the channel dimension to obtain a fluorescence image with six or more channels, which effectively increases the number of channels and provides more detailed information for image recognition.

[0042] Step 103: Combine multi-channel tooth surface fluorescence images with the trained tooth surface detection neural network to obtain tooth surface detection results.

[0043] Specifically, the tooth surface detection results in this embodiment include the abnormal pixel types (e.g., caries pixels, tartar pixels) and their corresponding locations. After obtaining the tooth surface detection results, the results and corresponding visualized images can be output. In this embodiment, a deep learning algorithm is used to detect different types of abnormal pixels in multi-channel tooth surface fluorescence images. The neural network used can include any one of deep neural networks (DNN), convolutional neural networks (CNN), and recurrent neural networks (RNN). It should be noted that this embodiment classifies and identifies tooth surface objects on a pixel-by-pixel basis, resulting in higher recognition accuracy in practical applications.

[0044] It should be noted that when constructing a neural network model, the first step is to obtain a training sample set. The steps for obtaining the training sample set include: acquiring tooth surface images with different pixel types, then traversing all tooth surface images and labeling different pixel types in all sample images, that is, labeling pixels in normal enamel areas, pixels in caries areas, and pixels in tartar areas to obtain sample images, which are also training samples. All training samples constitute the training sample set. Then, supervised learning is performed on the obtained training sample set to train the neural network model.

[0045] In one embodiment of this example, the step of obtaining tooth surface detection results by combining multi-channel tooth surface fluorescence images and a trained tooth surface detection neural network includes: inputting the multi-channel tooth surface fluorescence images into the trained spectral reconstruction neural network to obtain the corresponding hyperspectral data cube; and inputting the hyperspectral data cube into the trained tooth surface detection neural network to obtain the tooth surface detection results.

[0046] Specifically, in order to further improve the accuracy of classification and recognition by the tooth surface detection neural network, this embodiment can first input the multi-channel tooth surface fluorescence image into a pre-trained spectral reconstruction neural network. The reconstruction model increases the number of channels to a hyperspectral dimension, and then inputs it into the tooth surface detection neural network to realize the classification of caries and tartar at different stages of normal teeth.

[0047] Furthermore, in one embodiment of this example, the step of inputting the multi-channel tooth surface fluorescence image into a trained spectral reconstruction neural network to obtain the corresponding hyperspectral data cube includes: flattening the multi-channel tooth surface fluorescence image into a two-dimensional matrix form and inputting it into the trained spectral reconstruction neural network; extracting regional features through a feature extraction branch network; calculating a weight matrix guided by regional features through a weight calculation network; and calculating an adaptive threshold based on regional features through a threshold calculation network; and performing calculations on the two-dimensional matrix form of the multi-channel tooth surface fluorescence image based on the weight matrix and the adaptive threshold to obtain the corresponding hyperspectral data cube.

[0048] Specifically, in this embodiment, the multi-channel tooth surface fluorescence image is first flattened into a two-dimensional matrix form, and then used as the input to the reconstruction model. The reconstructed model after training contains a series of weight matrices and thresholds. When the image data in the two-dimensional matrix form is input into the reconstruction model, it will be processed with the weight matrices and thresholds. After the operation is completed, it will undergo shape restoration, and its channel number will be increased to the hyperspectral dimension to form a hyperspectral image cube.

[0049] In one embodiment of this example, after the step of obtaining the tooth surface detection result by combining multi-channel tooth surface fluorescence images and the trained tooth surface detection neural network, the method further includes: obtaining the corresponding visualization features for the abnormal types of tooth surface pixels at each pixel position in the tooth surface detection result; and performing visualization representation on the corresponding pixel positions in the tooth surface detection result based on the visualization features to obtain a visualization image.

[0050] Specifically, this embodiment pre-sets different visualization features for different types of abnormal pixel points on the tooth surface. Then, based on the correspondence of the pre-set features, it determines the visualization features corresponding to the position of each pixel point in the tooth surface detection result. Then, based on the determined visualization features, it characterizes the tooth surface detection result image to obtain a visualization image, which allows users to intuitively understand the health / abnormality of the tooth surface.

[0051] Furthermore, in one embodiment of this example, the aforementioned visualization features are basic color features. After the step of visually representing the corresponding pixel positions in the tooth surface detection results based on the visualization features to obtain a visualization image, the method further includes: for each pixel anomaly type, calculating the area of ​​the connected regions formed by each same-color tooth surface pixel in the visualization image, and determining the corresponding weight coefficient based on the tooth surface position of the connected region; for each connected region, calculating the corresponding anomaly degree evaluation index based on the area and weight coefficient; determining the color adjustment coefficient of each connected region based on the anomaly degree evaluation index; and correcting the basic color features of the corresponding connected regions in the visualization image based on the color adjustment coefficient.

[0052] Specifically, in practical applications, color representation can be used to visualize different types of tooth surface pixel anomalies. However, considering that the degree of anomaly varies for the same type of tooth surface pixel anomaly, using the same base color cannot provide users with more detailed results. Therefore, this embodiment calculates the area and location of each connected region of the same color for each type of tooth surface pixel anomaly (i.e., the joint region composed of multiple adjacent pixels of the same type of tooth surface pixel anomaly). The degree of anomaly of the connected regions of the same color is evaluated by combining the area and location. Then, based on the anomaly degree evaluation index, a corresponding color adjustment coefficient is assigned to each connected region. Finally, based on the color adjustment coefficient, the base color is adjusted, such as adjusting the color purity, so that tooth surface pixel regions with different degrees of anomaly under the same type of tooth surface pixel anomaly have different subdivided colors, and the visualization results are more detailed.

[0053] Based on the technical solution of the above embodiments of this application, the image sensor is controlled to image the fluorescence spectrum of the tooth surface transmitted through different filters to obtain multiple tooth surface fluorescence images; the multiple tooth surface fluorescence images are connected in the channel dimension to obtain a multi-channel tooth surface fluorescence image; the multi-channel tooth surface fluorescence image and the trained tooth surface detection neural network are combined to obtain the tooth surface detection result; wherein, the tooth surface detection result includes the abnormality type of tooth surface pixels and the corresponding pixel position. Through the implementation of the solution of this application, the fluorescence spectrum is segmented into bands and imaged on the image sensor to obtain a multi-channel fluorescence image, and then the classification and recognition of different tooth surface abnormalities are realized through a neural network model, which improves the accuracy of tooth surface health diagnosis and effectively reduces the diagnostic cost.

[0054] Figure 6 The method described in the second embodiment of this application is a refined tooth surface inspection method, which includes:

[0055] Step 601: At the first moment, control the image sensor to image the fluorescence spectrum of the tooth surface transmitted through the first filter to obtain the fluorescence image of the first tooth surface.

[0056] Specifically, in this embodiment, the tooth surface fluorescence spectrum is the fluorescence spectrum excited by an excitation light source (e.g., an ultraviolet or near-infrared light source) irradiating the tooth surface.

[0057] Step 602: At the second moment, control the image sensor to image the fluorescence spectrum of the tooth surface transmitted through the first filter and the second filter in sequence to obtain the second tooth surface fluorescence image.

[0058] In this embodiment, the first moment is the moment after the first moment, the optical axes of the first filter and the second filter coincide, and the transmission wavelengths of the two are different.

[0059] Step 603: Connect the two tooth surface fluorescence images in the channel dimension to obtain a six-channel tooth surface fluorescence image.

[0060] Specifically, compared to single-point fluorescence spectroscopy or three-channel fluorescence spectroscopy, the six-channel tooth surface fluorescence image in this embodiment can provide more detailed information, which is beneficial to the comprehensiveness and accuracy of subsequent image recognition.

[0061] Step 604: Input the six-channel tooth surface fluorescence image into the trained spectral reconstruction neural network to obtain the corresponding hyperspectral data cube.

[0062] Specifically, in this embodiment, the six-channel tooth surface fluorescence image can be fed into a pre-trained spectral reconstruction neural network. The reconstruction model increases the number of channels to a hyperspectral dimension, further improving the accuracy of subsequent tooth surface detection neural network classification and recognition.

[0063] Step 605: Input the cubic hyperspectral data into the trained tooth surface detection neural network to obtain the tooth surface detection results.

[0064] Specifically, the tooth surface detection results in this embodiment include the abnormal types of tooth surface pixels and the corresponding pixel locations. The abnormal types of tooth surface pixels include caries pixels, tartar pixels, etc.

[0065] Step 606: For the abnormality type of each pixel in the tooth surface detection results, obtain the corresponding visualization features.

[0066] Step 607: Based on the visualization features, visualize the positions of the corresponding pixels in the tooth surface detection results and output a visualization image.

[0067] Specifically, in this embodiment, the visualization features corresponding to the position of each pixel in the tooth surface detection result are determined according to the abnormality type of the tooth surface pixel. Then, the tooth surface detection result image is characterized based on the determined visualization features to obtain a visualization image, which allows users to intuitively understand the health / abnormality of the tooth surface.

[0068] It should be understood that the sequence number of each step in this embodiment does not imply the order in which the steps are executed. The execution order of each step should be determined by its function and internal logic, and should not constitute a unique limitation on the implementation process of this application embodiment.

[0069] Figure 7 This application provides a tooth surface inspection device according to a third embodiment. This tooth surface inspection device can be applied to the aforementioned tooth surface inspection methods. For example... Figure 7 As shown, the tooth surface inspection device mainly includes:

[0070] The imaging module 701 is used to control the image sensor to image the fluorescence spectrum of the tooth surface transmitted through different filters, thereby obtaining multiple fluorescence images of the tooth surface; wherein, the transmission wavelengths of the different filters are different.

[0071] The connection module 702 is used to connect multiple tooth surface fluorescence images in the channel dimension to obtain a multi-channel tooth surface fluorescence image;

[0072] The acquisition module 703 is used to combine multi-channel tooth surface fluorescence images and trained tooth surface detection neural networks to acquire tooth surface detection results; wherein, the tooth surface detection results include the abnormality type of tooth surface pixels and the corresponding pixel position.

[0073] In some embodiments of this example, the imaging module is specifically used to: control the same image sensor to image the tooth surface fluorescence spectrum transmitted through different filters in a time-division manner to obtain multiple tooth surface fluorescence images; or, control different image sensors to image the tooth surface fluorescence spectrum transmitted through corresponding filters simultaneously to obtain multiple tooth surface fluorescence images.

[0074] Furthermore, in some embodiments of this example, the filter includes a first filter and a second filter. Correspondingly, when the imaging module performs the function of controlling the same image sensor to image the tooth surface fluorescence spectrum transmitted through different filters in a time-division manner to obtain multiple tooth surface fluorescence images, it is specifically used to: control the image sensor to image the tooth surface fluorescence spectrum transmitted through the first filter at a first moment to obtain a first tooth surface fluorescence image; and control the image sensor to image the tooth surface fluorescence spectrum transmitted sequentially through the first filter and the second filter at a second moment to obtain a second tooth surface fluorescence image; wherein the optical axes of the first filter and the second filter coincide.

[0075] In some embodiments of this example, the acquisition module is specifically used to: input multi-channel tooth surface fluorescence images into a trained spectral reconstruction neural network to acquire corresponding hyperspectral data cubes; input hyperspectral data cubes into a trained tooth surface detection neural network to acquire tooth surface detection results.

[0076] Furthermore, in some embodiments of this example, when the imaging module performs the function of inputting the multi-channel tooth surface fluorescence image into the trained spectral reconstruction neural network to obtain the corresponding hyperspectral data cube, it is specifically used to: flatten the multi-channel tooth surface fluorescence image into a two-dimensional matrix form and input it into the trained spectral reconstruction neural network; extract regional features through a feature extraction branch network; calculate a weight matrix guided by regional features through a weight calculation network; and calculate an adaptive threshold based on regional features through a threshold calculation network; and perform calculations on the two-dimensional matrix form of the multi-channel tooth surface fluorescence image based on the weight matrix and the adaptive threshold to obtain the corresponding hyperspectral data cube.

[0077] In some embodiments of this example, the tooth surface detection device further includes: a characterization module, used to obtain corresponding visualization features for the abnormal types of tooth surface pixels at each pixel position in the tooth surface detection result; and to perform visualization characterization on the corresponding pixel positions in the tooth surface detection result based on the visualization features to obtain a visualization image.

[0078] Furthermore, in some embodiments of this example, the characterization module is also used to: for each pixel anomaly type, respectively calculate the area of ​​the connected region formed by each same-color tooth surface pixel in the visualized image, and determine the corresponding weight coefficient according to the position of the connected region on the tooth surface; for each connected region, calculate the corresponding anomaly degree evaluation index based on the area and weight coefficient; determine the color adjustment coefficient of each connected region based on the anomaly degree evaluation index; and correct the basic color features of the corresponding connected regions in the visualized image based on the color adjustment coefficient.

[0079] It should be noted that the tooth surface detection methods in the first and second embodiments can be implemented based on the tooth surface detection device provided in this embodiment. Those skilled in the art can clearly understand that, for the sake of convenience and brevity, the specific working process of the tooth surface detection device described in this embodiment can be referred to the corresponding process in the aforementioned method embodiments, and will not be repeated here.

[0080] According to the tooth surface detection device provided in this embodiment, the image sensor is controlled to image the fluorescence spectrum of the tooth surface transmitted through different filters, resulting in multiple tooth surface fluorescence images. These multiple tooth surface fluorescence images are then connected along the channel dimension to obtain a multi-channel tooth surface fluorescence image. The multi-channel tooth surface fluorescence image is combined with a trained tooth surface detection neural network to obtain the tooth surface detection result. The tooth surface detection result includes the type of abnormality of tooth surface pixels and the corresponding pixel location. Through the implementation of this application's solution, the fluorescence spectrum is segmented into bands and imaged on the image sensor to obtain a multi-channel fluorescence image. Then, a neural network model is used to classify and identify different tooth surface abnormalities, improving the accuracy of tooth surface health diagnosis and effectively reducing diagnostic costs.

[0081] Figure 8 An electronic device is provided in the fourth embodiment of this application. This electronic device can be used to implement the tooth surface detection method in the foregoing embodiments, and mainly includes:

[0082] The system includes a memory 801, a processor 802, and a computer program 803 stored on the memory 801 and executable on the processor 802. The memory 801 and the processor 802 are connected via communication. When the processor 802 executes the computer program 803, it implements the method described in Embodiment 1 or 2 above. The number of processors can be one or more.

[0083] The memory 801 can be a high-speed random access memory (RAM) or a non-volatile memory, such as a disk storage device. The memory 801 is used to store executable program code, and the processor 802 is coupled to the memory 801.

[0084] Furthermore, embodiments of this application also provide a computer-readable storage medium, which may be disposed in the aforementioned electronic device, and the computer-readable storage medium may be as described above. Figure 8 The memory in the illustrated embodiment.

[0085] The computer-readable storage medium stores a computer program that, when executed by a processor, implements the tooth surface detection method described in the foregoing embodiments. Furthermore, the computer-readable storage medium can also be a USB flash drive, a portable hard drive, a read-only memory (ROM), RAM, a magnetic disk, or an optical disk, or any other medium capable of storing program code.

[0086] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0087] The modules described as separate components may or may not be physically separate. Similarly, the components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0088] Furthermore, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0089] If the integrated module is implemented as a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a readable storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned readable storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0090] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0091] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0092] The above is a description of the tooth surface detection method, apparatus, device and readable storage medium provided in this application. For those skilled in the art, based on the ideas of the embodiments of this application, there will be changes in the specific implementation and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for detecting tooth surfaces, characterized in that, include: The image sensors are controlled to image the fluorescence spectrum of the tooth surface transmitted through different filters, resulting in multiple fluorescence images of the tooth surface; wherein the transmission wavelengths of the different filters are different. Multiple tooth surface fluorescence images are concatenated along the channel dimension to obtain a multi-channel tooth surface fluorescence image; By combining the multi-channel tooth surface fluorescence images and the trained tooth surface detection neural network, tooth surface detection results are obtained; wherein, the tooth surface detection results include the abnormality type of tooth surface pixels and the corresponding pixel position; The step of combining the multi-channel tooth surface fluorescence image and the trained tooth surface detection neural network to obtain the tooth surface detection result includes: inputting the multi-channel tooth surface fluorescence image into the trained spectral reconstruction neural network to obtain the corresponding hyperspectral data cube; inputting the hyperspectral data cube into the trained tooth surface detection neural network to obtain the tooth surface detection result; The step of inputting the multi-channel tooth surface fluorescence image into a trained spectral reconstruction neural network to obtain the corresponding hyperspectral data cube includes: flattening the multi-channel tooth surface fluorescence image into a two-dimensional matrix form and inputting it into the trained spectral reconstruction neural network; extracting regional features through a feature extraction branch network; calculating a weight matrix guided by the regional features through a weight calculation network; and calculating an adaptive threshold based on the regional features through a threshold calculation network; performing operations on the two-dimensional matrix form of the multi-channel tooth surface fluorescence image based on the weight matrix and the adaptive threshold to obtain the corresponding hyperspectral data cube; wherein, the two-dimensional matrix form of the multi-channel tooth surface fluorescence image is used as the input to the trained spectral reconstruction neural network, the trained spectral reconstruction neural network contains a series of weight matrices and adaptive thresholds, the two-dimensional matrix form of the multi-channel tooth surface fluorescence image is processed with the weight matrix and the adaptive thresholds, and after the operation is completed, shape restoration is performed, and its channel number is increased to the hyperspectral dimension to form a hyperspectral data cube.

2. The tooth surface inspection method according to claim 1, characterized in that, The step of separately controlling the image sensor to image the tooth surface fluorescence spectrum transmitted through different filters to obtain multiple tooth surface fluorescence images includes: The same image sensor is controlled separately to image the fluorescence spectrum of the tooth surface transmitted through different filters in a time-division manner, resulting in multiple fluorescence images of the tooth surface; Alternatively, different image sensors can be controlled simultaneously to image the fluorescence spectrum of the tooth surface transmitted through the corresponding filter, resulting in multiple fluorescence images of the tooth surface.

3. The tooth surface inspection method according to claim 2, characterized in that, The filter includes a first filter and a second filter; The step of controlling the same image sensor to image the tooth surface fluorescence spectrum transmitted through different filters in a time-division manner to obtain multiple tooth surface fluorescence images includes: At the first moment, the image sensor is controlled to image the fluorescence spectrum of the tooth surface transmitted through the first filter to obtain the fluorescence image of the first tooth surface; At a second moment, the image sensor is controlled to image the fluorescence spectrum of the tooth surface transmitted through the first filter and the second filter in sequence to obtain a second tooth surface fluorescence image; wherein the optical axes of the first filter and the second filter coincide.

4. The tooth surface inspection method according to any one of claims 1 to 3, characterized in that, After the step of combining the multi-channel tooth surface fluorescence image and the trained tooth surface detection neural network to obtain the tooth surface detection result, the method further includes: For the abnormality type of each pixel point in the tooth surface detection result, obtain the corresponding visualization features; Based on the visualization features, the corresponding pixel positions in the tooth surface detection results are visualized to obtain a visualization image.

5. The tooth surface inspection method according to claim 4, characterized in that, The visualization feature is a basic color feature; after the step of visualizing the corresponding pixel positions in the tooth surface detection result based on the visualization feature to obtain a visualization image, the method further includes: For each of the aforementioned pixel anomaly types, the area of ​​the connected region formed by each same-color tooth surface pixel in the visualized image is calculated, and the corresponding weight coefficient is determined based on the position of the connected region on the tooth surface. For each of the connected regions, a corresponding anomaly assessment index is calculated based on the area and the weighting coefficient. The color adjustment coefficient for each connected region is determined based on the anomaly assessment index. Based on the color adjustment coefficients, the basic color features of the corresponding connected regions in the visualized image are corrected respectively.

6. A tooth surface inspection device, characterized in that, include: An imaging module is used to control the image sensor to image the fluorescence spectrum of the tooth surface transmitted through different filters, thereby obtaining multiple fluorescence images of the tooth surface; wherein, the transmission wavelengths of the different filters are different; A connection module is used to connect multiple tooth surface fluorescence images in the channel dimension to obtain a multi-channel tooth surface fluorescence image; The acquisition module is used to flatten the multi-channel tooth surface fluorescence image into a two-dimensional matrix and input it into a trained spectral reconstruction neural network. It extracts regional features through a feature extraction branch network; calculates a weight matrix guided by the regional features through a weight calculation network; and calculates an adaptive threshold based on the regional features through a threshold calculation network. Based on the weight matrix and the adaptive threshold, it performs calculations on the two-dimensional matrix form of the multi-channel tooth surface fluorescence image to obtain a corresponding hyperspectral data cube. The hyperspectral data cube is then input into a trained tooth surface detection neural network to obtain tooth surface detection results. Specifically, the two-dimensional matrix form of the multi-channel tooth surface fluorescence image is used as input to the trained spectral reconstruction neural network. The trained spectral reconstruction neural network contains a series of weight matrices and adaptive thresholds. The two-dimensional matrix form of the multi-channel tooth surface fluorescence image is processed with the weight matrix and adaptive thresholds. After the calculations are completed, shape restoration is performed, and the number of channels is increased to the hyperspectral dimension, forming a hyperspectral data cube. The tooth surface detection results include the anomaly type of tooth surface pixels and the corresponding pixel location.

7. An electronic device, characterized in that, Includes memory and processor, of which: The processor is used to execute computer programs stored in the memory; When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 5.

Images