Tongue picture segmentation and classification system

Abstract

The invention discloses a tongue picture segmentation and classification system, which belongs to the field of image recognition and comprises a tongue body segmentation module, a tongue picture classification module, a user interface, a privacy protection and data security module and an intelligent recommendation and cloud service interface. Wherein the user interface is used for providing an interactive interface for a user and managing tongue picture health data of the user; the tongue body segmentation module is used for automatically segmenting a tongue picture input by the user interface to obtain a tongue body region; the tongue picture classification module is used for classifying the tongue body region to obtain a tongue picture type; the privacy protection and data security module is used for encrypting system data; and the intelligent recommendation and cloud service interface is used for providing personalized suggestions based on the tongue picture type and providing a health data transmission interface among multiple platforms for a user.

Description

Technical Field

[0001] The present invention belongs to the technical field of image recognition, and in particular relates to a tongue image segmentation and classification system. Background Art

[0002] Tongue segmentation is a key step in achieving automated tongue analysis. However, traditional tongue image annotation methods face multiple challenges. Manual annotation is not only time-consuming and labor-intensive, but also easily affected by human factors, especially when dealing with complex areas such as the tongue and its connection with the upper part of the mouth, resulting in inaccurate and inconsistent annotation results. These problems directly affect the accuracy and reliability of tongue analysis.

[0003] With the development of image processing and artificial intelligence technology, automated tongue analysis technology has gradually become a research hotspot. These technologies can provide more objective and accurate tongue analysis. However, existing tongue analysis technology still faces some challenges, such as the accuracy of tongue segmentation, the comprehensiveness of feature extraction, and the efficiency of classification algorithms. In order to overcome these challenges, the present invention proposes a tongue image segmentation and classification system that integrates advanced image processing and deep learning technology. Summary of the invention

[0004] In order to solve the above technical problems, the present invention proposes a tongue image segmentation and classification system to solve the above problems existing in the prior art.

[0005] To achieve the above object, the present invention provides a tongue image segmentation and classification system, comprising:

[0006] Tongue segmentation module, tongue image classification module, user interface, privacy protection and data security module, and intelligent recommendation and cloud service interface;

[0007] Wherein, the user interface is used to provide an interactive interface for the user and manage the user's tongue health data;

[0008] The tongue segmentation module is used to automatically segment the tongue image inputted in the interactive interface to obtain the tongue area;

[0009] The tongue image classification module is used to classify the tongue area to obtain the tongue image type;

[0010] The privacy protection and data security module is used to encrypt the user's tongue health data and transmission process;

[0011] The intelligent recommendation and cloud service interface provides personalized suggestions based on the tongue image type and provides users with a tongue image health data transmission interface between platforms.

[0012] Optionally, the user interface includes: an image uploading unit, a result display unit and a data management unit;

[0013] Wherein, the image uploading unit is used to support uploading tongue image pictures in different file formats and from different sources;

[0014] The result display unit is used to display the results of tongue image segmentation and classification in a graphical manner;

[0015] The data management unit is used for users to view and manage tongue health data.

[0016] Optionally, the tongue segmentation module includes: an image encoder, a prompt encoder, a mask decoder and an automatic mask generator;

[0017] Wherein, the image encoder is used to extract high-dimensional features of the tongue image;

[0018] The prompt encoder is used to receive and process the segmentation instruction obtained based on the user interface;

[0019] The mask decoder generates a segmentation mask based on the high-dimensional feature and the segmentation instruction, and obtains the tongue body region based on the segmentation mask;

[0020] The automatic mask generator is used to enlarge the number of segmentation masks.

[0021] Optionally, the tongue image classification module includes: a feature extractor, a deep learning network and a classifier;

[0022] Wherein, the feature extractor is used to extract key features of the tongue area;

[0023] The deep learning network is used to analyze the key features to obtain feature analysis results;

[0024] The classifier is used to receive the output of the deep learning network and determine the tongue image type.

[0025] Optionally, the privacy protection and data security module includes: a data encryption unit and an access control unit;

[0026] Wherein, the data encryption unit is used to encrypt the tongue health data and the transmission and storage process of the tongue health data through an end-to-end encryption algorithm;

[0027] The access control unit performs user access control based on the role-based access control policy.

[0028] Optionally, the tongue image segmentation and classification system further includes a remote update and performance optimization module and a multi-language support and user feedback mechanism module;

[0029] Wherein, the remote update and performance optimization module is used to optimize the classifier and update the hardware performance;

[0030] The multi-language support and user feedback mechanism module is used to provide users with a multi-language selection mechanism and a feedback mechanism.

[0031] The present invention also provides a computer terminal device, comprising:

[0032] one or more processors;

[0033] A memory, coupled to the processor, for storing one or more programs;

[0034] When the one or more programs are executed by the one or more processors, the one or more processors implement a tongue image segmentation and classification system.

[0035] The present invention also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, a tongue image segmentation and classification system is implemented.

[0036] Compared with the prior art, the present invention has the following advantages and technical effects:

[0037] The system can automatically complete the segmentation, feature extraction and classification of tongue images, improving the accuracy and efficiency of tongue image classification. The tongue image segmentation and classification system of the present invention can provide users with a comprehensive, accurate and efficient Chinese medicine tongue classification system, while ensuring the security and privacy of user data. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0039] Figure 1 A structural diagram of a tongue image segmentation and classification system according to an embodiment of the present invention;

[0040] Figure 2 Obtaining and encoding tongue category data for the embodiment of the present invention;

[0041] Figure 3 The process of training the model for data in the embodiment of the present invention;

[0042] Figure 4 The embodiment of the present invention performs prediction based on the trained model. DETAILED DESCRIPTION

[0043] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0044] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0045] Embodiment 1

[0046] The present invention provides a tongue image segmentation and classification system, which includes a tongue body segmentation module and a tongue image classification module, and can accurately segment and classify tongue images, thereby assisting TCM diagnosis. The system uses advanced image processing and deep learning technology to achieve automatic segmentation, feature extraction and classification of tongue images, thereby improving the objectivity and repeatability of tongue diagnosis.

[0047] like Figure 1 As shown, this embodiment provides a tongue segmentation and classification system to achieve automatic segmentation and high-precision classification of tongue images. The system specifically includes: a tongue segmentation module, a tongue classification module, a user interface, a privacy protection and data security module, a remote update and performance optimization module, a multi-language support and user feedback mechanism module, an intelligent recommendation and cloud service interface.

[0048] As a specific implementation of this embodiment, the tongue segmentation module: uses an image encoder, a prompt encoder and a mask decoder, combined with zero-sample transfer learning technology, to achieve automatic segmentation of the tongue area in the tongue image. The tongue segmentation module is used to automatically segment the input tongue image to identify and separate the tongue area.

[0049] This embodiment independently developed a tongue segmentation model (segment-tongue) to provide an efficient and accurate solution to these problems. The model has a strong zero-sample transfer learning capability and can effectively perform image segmentation in the absence of a large amount of annotated data. The model combines multiple key technologies, including an image encoder, a prompt encoder, and a mask decoder.

[0050] Tongue segmentation model usage process:

[0051] (1) Use the loaded segment-tongue model to generate an automatic segmentation mask for the input image through SamAutomaticMaskGenerator.

[0052] (2) Find the mask with the largest area among the segmentation masks obtained above, and overlay it on the original image as a transparent layer.

[0053] (3) Find the mask with the largest area and draw the key point coordinates of the mask on the image. And display multiple masks and their key points on the image.

[0054] (4) Find the center point of the mask with the largest area in a set of masks, find the center points of other masks closest to this center point, and then mark these two center points on the image. In this way, we can intuitively identify and locate important objects in the image and their neighboring objects.

[0055] Find the contour of the mask with the largest area in a set of masks, and check whether the center points of other masks are within the contour, as well as the intersection of these masks with the largest area mask, so as to identify important objects in the positioning image and their relationships.

[0056] (5) Find the contour of the mask with the largest area in a set of masks, and check whether the center points of other masks are within the contour, as well as the intersection of these masks with the mask with the largest area, and finally generate a segmented image of only the tongue.

[0057] Furthermore, the tongue segmentation module includes:

[0058] Image encoder: used to extract the texture, color, and shape features of tongue images. The image encoder uses the latest convolutional neural network (CNN) architecture, such as ResNet, Inception, or a custom neural network structure, to extract high-dimensional features of tongue images. The encoder is specially optimized for tongue images and can capture subtle texture and color changes.

[0059] Prompt encoder: It is used to receive and process segmentation instructions and use natural language processing (NLP) technology to convert the user's segmentation instructions into precise instructions for image segmentation. The encoder can understand complex user input and convert it into a format that the model can understand.

[0060] Mask decoder: used to generate a segmentation mask based on the features extracted by the image encoder and the instructions of the prompt encoder, thereby obtaining a segmented tongue image. The mask decoder combines the features extracted by the image encoder and the instructions of the prompt encoder to generate an accurate segmentation mask. The decoder uses U-Net or similar structures to ensure the accuracy of the segmentation edges. Taking the U-Net structure as an example, the specific process includes:

[0061] (1) Input data:

[0062] Output of the image encoder: feature maps extracted by the image encoder. These feature maps contain high-level semantic information of the input image.

[0063] Outputs of the hint encoder: Instructions generated by the hint encoder, which can be text, labels, or other forms of metadata, to guide the segmentation process.

[0064] (2) Feature Fusion:

[0065] Feature fusion layer: fuses the feature maps from the image encoder with the instructions from the hint encoder. This can be done in many ways, for example:

[0066] Direct weighting: The output of the hint encoder is weighted and added to the feature map of the image encoder.

[0067] Attention Mechanism: Use the attention mechanism to dynamically adjust the importance of different regions in the feature map.

[0068] Convolutional layer: The output of the hint encoder is converted into a tensor of the same dimension as the image encoder feature map through a convolutional layer and then fused.

[0069] (3) U-Net structure:

[0070] U-Net is a commonly used segmentation network structure with a symmetrical encoder and decoder structure, which transfers the encoder's feature map to the decoder through skip connections.

[0071] Encoder part

[0072] Convolutional layers: Multiple convolutional layers are used to extract features, and each layer is usually followed by an activation function (such as ReLU) and batch normalization.

[0073] Pooling layer: Use pooling layers (such as max pooling) to reduce the spatial dimension of the feature map while increasing the receptive field.

[0074] Decoder part

[0075] Upsampling: Use transposed convolution or upsampling layers to increase the spatial dimension of feature maps.

[0076] Skip connection: Connect the feature map of the corresponding layer in the encoder with the feature map of the decoder to retain more detailed information.

[0077] Convolutional layers: Convolutional layers are used again to refine the feature maps, and each layer is usually followed by an activation function and batch normalization.

[0078] (4) Output layer:

[0079] 1x1 convolution: At the end of the decoder, a 1x1 convolution is used to reduce the number of channels of the feature map to the required number of categories (usually 1 channel for a binary classification problem).

[0080] Activation function: Use sigmoid or softmax activation function to generate segmentation mask. For binary classification problems, sigmoid is a common choice.

[0081] (5) Post-processing:

[0082] Thresholding: Thresholding is applied to the output segmentation mask to produce a binary image (0 and 1).

[0083] Morphological operations: Morphological operations (such as dilation and erosion) may be required to improve the segmentation results.

[0084] (6) Training:

[0085] Loss function: Use a loss function suitable for the segmentation task, such as cross entropy loss, Dice loss, or a combination of them.

[0086] Optimizer: Choose a suitable optimizer, such as Adam or SGD, to train the network.

[0087] (7) Evaluation:

[0088] Evaluation metrics: Use metrics such as IoU (Intersection over Union) and Dice coefficient to evaluate the quality of the segmentation mask.

[0089] Through the above steps, the mask decoder can combine the features extracted by the image encoder and the instructions of the prompt encoder to generate an accurate segmentation mask, thereby achieving accurate segmentation of the tongue image.

[0090] Automatic Mask Generator: It is used to generate automatic segmentation masks of the tongue through zero-shot transfer learning in the absence of a large amount of annotated data. The automatic mask generator applies the generative adversarial network (GAN) technology to automatically generate segmentation masks in the case of limited data annotation. The generator is able to learn from unannotated samples and generate high-quality segmentation masks.

[0091] Furthermore, the implementation process of the tongue segmentation module includes: the user selects the option to upload a tongue image through the system's user interface, and uploads a tongue image by dragging and dropping or browsing. After the image is uploaded, the system automatically transfers the image to the image encoder of the tongue segmentation module. The image encoder uses a pre-trained deep learning model, such as ResNet-50, to extract the features of the tongue image, including texture, color, and shape. The prompt encoder receives the segmentation instructions input by the user, for example, the user may specify the tongue area that needs to be focused on. The mask decoder combines the features extracted by the image encoder and the instructions of the prompt encoder to generate a preliminary segmentation mask. The automatic mask generator further optimizes the segmentation mask to ensure that the edges are clear and the tongue area is accurately segmented. The system outputs the segmented tongue image to the tongue classification module for further processing.

[0092] As a specific implementation of this embodiment, the tongue image classification module is used to classify the segmented tongue images output by the tongue segmentation module to determine the type of tongue image. The tongue image classification module uses a deep learning network and a classifier, combined with an attention mechanism, to achieve feature extraction and classification of the segmented tongue images.

[0093] For tongue image classification, this embodiment develops a tongue image classification model (SE_tongue_net). The SE block adaptively recalibrates the feature response of the channel dimension by learning the importance of different channels. This mechanism makes it possible to better identify and utilize the key features that distinguish different tongue image categories. The tongue_net model adopts a multi-path structure to improve the performance of the model by stacking multiple residual blocks in parallel in the same network level. This design increases the width and complexity of the model, allowing the model to capture richer feature representations, thereby improving the accuracy of classification. In addition, the design of the SE_ResNeXt model is highly modular and can be adjusted and expanded according to the needs of specific tasks. For example, the model can be optimized by increasing or decreasing the number of SE blocks and adjusting the depth and width of the residual blocks.

[0094] Tongue classification model usage process:

[0095] (1) Figure 2 As shown, obtain the tongue category data and encode it:

[0096] 15 categories in total: {'pale white tongue with gray-black fur':'100001000','pale white tongue with white fur':'100000010','pale white tongue with yellow fur':'100000001','pale red tongue with gray-black fur':'000011000','pale red tongue with white fur':'000010010','pale red tongue with yellow fur':'000010001','red tongue with gray-black fur':'000101000','red tongue with white fur':'000011000' Coating':'000100010','Red tongue with yellow coating':'000100001','Red tongue with gray and black coating':'001001000','Red tongue with white coating':'001000010','Red tongue Yellow coating':'001000001','Purple tongue with gray and black coating':'010001000','Purple tongue with white coating':'010000010','Purple tongue with yellow coating':'010000001'}

[0097] (2) Figure 3 As shown, the model is trained using the data.

[0098] (3) Figure 4 As shown, use the trained model to make predictions.

[0099] Furthermore, the tongue image classification module includes:

[0100] Feature Extractor: Multi-scale feature extraction technology is used to extract key features from the segmented tongue image, including but not limited to color, texture, shape, and tongue coating features. The extractor can capture features of different scales and directions to provide richer information.

[0101] Deep learning network: used to classify tongue images based on the extracted tongue features. The network includes multiple convolutional layers, pooling layers, and fully connected layers. The deep learning network builds a multi-layer convolutional neural network, including batch normalization layers and ReLU activation functions to improve the generalization ability of the model. The network design takes into account the balance between computational efficiency and classification performance.

[0102] Classifier: It is used to receive the output of the deep learning network and determine the classification result of the tongue image. The classifier performs classification based on predefined tongue image categories. The classifier uses an ensemble learning method, such as random forest or gradient boosting machine (GBM), combined with the output of the deep learning network to improve the accuracy and robustness of the classification.

[0103] Attention mechanism: By introducing channel attention and spatial attention mechanisms, the model can focus on the most important features for classification. This mechanism helps improve the model's ability to identify key features.

[0104] Furthermore, the implementation process of the tongue classification module includes: the tongue classification module receives the segmented tongue image from the tongue segmentation module. The feature extractor extracts key features from the segmented image, such as tongue color, tongue coating, tongue shape, etc. The deep learning network analyzes the extracted features, and the network structure may include multiple convolutional layers, pooling layers, fully connected layers, and batch normalization layers. The classifier receives the output of the deep learning network and classifies it through the softmax function to determine the type of tongue image. The attention mechanism plays a role in the classification process to ensure that the model focuses on the most discernible features. The classification results are displayed to the user together with the segmented tongue image, including the type of tongue image and related health advice.

[0105] As a specific implementation of this embodiment, the user interface provides functions such as an image uploading unit, a result display unit, and a data management unit, so that the user can conveniently use the system to perform tongue diagnosis.

[0106] Furthermore, the user interface includes:

[0107] Image upload unit: Design an intuitive image upload unit interface that supports multiple file formats and image sources, such as direct shooting or selection from the album.

[0108] Result display unit: Displays the segmentation and classification results in a graphical way, including the segmented tongue image, detailed explanation of the classification results and health advice. The result page also provides the function of saving and sharing.

[0109] Data Management Unit: Provides a user-friendly data management unit interface that allows users to view and manage their health records and diagnostic data, including history and trend analysis.

[0110] Furthermore, the user interface implementation process includes: After the user uploads the tongue image through the user interface, the system provides real-time feedback, including the upload progress and estimated processing time. The segmentation and classification results are displayed to the user through an intuitive graphical interface, and the user can view the segmented tongue image and detailed classification results. The user can view and manage his or her health records, including historical tongue images, classification results, and health recommendations through the data management unit function. The user can also choose to save the results locally or share them with medical professionals via email.

[0111] As a specific implementation of this embodiment, the privacy protection and data security module: adopts data encryption and access control technology to ensure the security and privacy of user data.

[0112] Furthermore, the privacy protection and data security module includes:

[0113] Data encryption unit: Uses end-to-end encryption algorithms, such as AES or RSA, to ensure the security of user data during transmission and storage.

[0114] Access Control Unit: Implement role-based access control policies to ensure that only authorized users can access sensitive data. At the same time, multi-factor authentication (MFA) is provided to enhance account security.

[0115] Furthermore, the implementation process of the privacy protection and data security module includes: when users register system accounts, the system enforces a strong password policy and provides a multi-factor authentication (MFA) option. All user data, including tongue images and health records, are encrypted during transmission and storage. The system conducts regular security audits to ensure that there is no risk of unauthorized data access and leakage.

[0116] As a specific implementation of this embodiment, the remote update and performance optimization module: automatically updates the system through the Internet to obtain the latest functions and improvements, and uses hardware acceleration to improve system performance.

[0117] Algorithm optimization: Through continuous model training and algorithm optimization, we ensure that the system's performance is always at its best. The optimization process includes hyperparameter adjustment and network structure improvement.

[0118] Hardware acceleration: Use dedicated hardware resources such as GPUs or TPUs to accelerate image processing and deep learning calculations to achieve real-time or near real-time analysis.

[0119] Furthermore, the implementation process of the remote update and performance optimization module includes: the system regularly checks for updates to algorithms and models, automatically downloads and installs the latest updates to improve the accuracy of segmentation and classification. The system uses GPU to accelerate deep learning calculations to ensure fast response even when processing high-resolution images.

[0120] As a specific implementation of this embodiment, the multi-language support and user feedback mechanism module provides a multi-language user interface and allows users to provide feedback so that the system can learn and improve itself.

[0121] Furthermore, the multi-language support and user feedback mechanism modules include:

[0122] Multilingual user interface: Provides user interfaces in multiple languages, including but not limited to Chinese, English, Spanish, etc., to meet the needs of users of different languages.

[0123] User Feedback: Build a user feedback system to allow users to provide feedback on segmentation and classification results. This feedback will be used for further training and improvement of the model.

[0124] Furthermore, the implementation process of the multi-language support and user feedback mechanism module includes: the system provides user interfaces in multiple languages ​​such as Chinese, English, and Spanish, and users can select their preferred language in the settings. After using the system, users can provide opinions and suggestions through the user feedback function, and the system will collect these feedbacks and use them for future system improvements.

[0125] As a specific implementation method of this embodiment, intelligent recommendation and cloud service interface: provide personalized health advice and conditioning plans based on the user's tongue image classification results, and provide a cloud service interface for use by third-party services and applications.

[0126] Furthermore, the intelligent recommendation and cloud service interface includes:

[0127] Intelligent recommendation: Based on the user's tongue image classification results, combined with traditional Chinese medicine theory and big data analysis, provide personalized health advice and conditioning plans.

[0128] Cloud Service Interface: Provides a cloud service interface that allows third-party services and applications to access and use the segmentation and classification capabilities of the system to achieve data sharing and service expansion.

[0129] Furthermore, the implementation process of intelligent recommendation and cloud service interface includes: based on the user's tongue classification results, the system provides personalized health suggestions, such as diet adjustment, lifestyle improvement, etc. The system connects with the third-party health management system through the cloud service interface, allowing the user's health data to be seamlessly synchronized across multiple platforms.

[0130] As a specific implementation of this embodiment, data synchronization and backup include:

[0131] Data synchronization: Through cloud services, users can have a consistent experience on different devices and synchronize data in real time. The synchronization process ensures the consistency and integrity of the data.

[0132] Data backup and recovery: Regularly back up user data and provide recovery services when data is lost or damaged. The backup strategy includes local backup and cloud backup to provide flexible recovery options.

[0133] As a specific implementation of this embodiment, auditing and compliance include:

[0134] Data access log: Records access logs of user data, including time, operation, and operator information, for easy auditing and monitoring. Logs are stored on a secure server with access restrictions.

[0135] Compliance checks: Ensure that the system complies with relevant data protection regulations and industry standards, such as GDPR or HIPAA. Compliance checks include regular internal audits and external certifications.

[0136] The data synchronization step between the electronic devices and the server in this system ensures that the user's health records and diagnostic data are consistent across multiple devices.

[0137] Embodiment 2

[0138] This embodiment also provides a computer terminal device, including:

[0139] one or more processors;

[0140] A memory, coupled to the processor, for storing one or more programs;

[0141] When the one or more programs are executed by the one or more processors, the one or more processors implement a tongue image segmentation and classification system.

[0142] This embodiment further provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, a tongue image segmentation and classification system is implemented.

[0143] The above are only preferred specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A tongue image segmentation and classification system, characterized in that: include: Tongue segmentation module, tongue image classification module, user interface, privacy protection and data security module, and intelligent recommendation and cloud service interface; Wherein, the user interface is used to provide an interactive interface for the user and manage the user's tongue health data; The tongue segmentation module is used to automatically segment the tongue image inputted in the interactive interface to obtain the tongue area; The tongue image classification module is used to classify the tongue area to obtain the tongue image type; The privacy protection and data security module is used to encrypt the user's tongue health data and transmission process; The intelligent recommendation and cloud service interface provides personalized suggestions based on the tongue image type and provides users with a tongue image health data transmission interface between platforms.

2. The tongue image segmentation and classification system according to claim 1, characterized in that: The user interface includes: an image uploading unit, a result display unit and a data management unit; Wherein, the image uploading unit is used to support uploading tongue image pictures in different file formats and from different sources; The result display unit is used to display the results of tongue image segmentation and classification in a graphical manner; The data management unit is used for users to view and manage tongue health data.

3. The tongue image segmentation and classification system according to claim 2, characterized in that: The tongue segmentation module includes: an image encoder, a prompt encoder, a mask decoder and an automatic mask generator; Wherein, the image encoder is used to extract high-dimensional features of the tongue image; The prompt encoder is used to receive and process the segmentation instruction obtained based on the user interface; The mask decoder generates a segmentation mask based on the high-dimensional feature and the segmentation instruction, and obtains the tongue body region based on the segmentation mask; The automatic mask generator is used to enlarge the number of segmentation masks.

4. The tongue image segmentation and classification system according to claim 3, characterized in that: The tongue image classification module includes: a feature extractor, a deep learning network and a classifier; Wherein, the feature extractor is used to extract key features of the tongue area; The deep learning network is used to analyze the key features to obtain feature analysis results; The classifier is used to receive the output of the deep learning network and determine the tongue image type.

5. The tongue image segmentation and classification system according to claim 4, characterized in that: The privacy protection and data security module includes: a data encryption unit and an access control unit; Wherein, the data encryption unit is used to encrypt the tongue health data and the transmission and storage process of the tongue health data through an end-to-end encryption algorithm; The access control unit performs user access control based on the role-based access control policy.

6. The tongue image segmentation and classification system according to claim 5, characterized in that: The tongue image segmentation and classification system also includes a remote update and performance optimization module and a multi-language support and user feedback mechanism module; Wherein, the remote update and performance optimization module is used to optimize the classifier and update the hardware performance; The multi-language support and user feedback mechanism module is used to provide users with a multi-language selection mechanism and a feedback mechanism.

7. A computer terminal device, characterized in that: include: one or more processors; A memory, coupled to the processor, for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the tongue image segmentation and classification system as described in any one of claims 1-6.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the tongue image segmentation and classification system according to any one of claims 1 to 6 is implemented.

Images