Method for judging body conditions of internal organs of human body and correspondence between body conditions of traditional Chinese medicine based on features extracted from face images by videos

Abstract

The invention relates to the field of face image feature recognition, and discloses and aims to provide a novel method for judging the body condition of the internal organs of a human body and the correspondence between the internal organs and the traditional Chinese medicine constitution by using face image features extracted by a video. According to the method, the health state of an individual is evaluated by analyzing the surface image features, and personalized health management suggestions are provided accordingly. According to the method, key features are extracted from the face image by using an image processing technology, and are combined with a traditional Chinese medicine constitution theory, so that the human health condition is effectively evaluated. In addition, the invention also discusses the application value of the face image characteristics in traditional Chinese medicine diagnosis, and opens up a new path for the combination of traditional medicine and modern science and technology.

Description

Technical Field

[0001] The present invention relates to the field of facial feature recognition. The present invention discloses a new method for using facial features extracted from videos to judge the physical condition of the five internal organs of the human body and the correspondence between the physical condition and the TCM constitution. The method aims to evaluate the health status of an individual by analyzing facial features, and provide personalized health management suggestions accordingly. The present invention uses image processing technology to extract key features from facial images, and combines them with the TCM constitution theory to achieve an effective assessment of the health status of the human body. In addition, the present invention also explores the application value of facial features in TCM diagnosis, opening up a new path for the combination of traditional medicine and modern technology. Background Art

[0002] Data sources: existing books, TCM textbooks and clinical guidelines.

[0003] Related technologies: OpenCV, Dlib, 68 key point model, LAB color space, and NumPy are all mature technical methods. References:

[0004] Classic works of traditional Chinese medicine such as "Huangdi Neijing" and "Treatise on Febrile and Miscellaneous Diseases", as well as modern Chinese medicine textbooks and clinical guidelines.

[0005] The key is how to analyze facial features and match them with human physique, evaluate the individual's health status, and provide personalized health management recommendations based on this. Summary of the invention

[0006] At present, facial feature extraction, as an emerging technology, has certain potential, but it still faces many challenges and shortcomings in practical applications. It is mainly reflected in user-friendliness, high data input requirements, user feedback mechanism, cross-platform compatibility, and environmental adaptability. First, the user interface may not be intuitive enough, and the operation steps are complicated, which makes it difficult for users to use. Secondly, the system may require users to provide high-quality facial images, and have high requirements for light, angle, resolution, etc. Then, the system may lack an effective user feedback mechanism and fail to obtain and process user opinions and suggestions in a timely manner. In addition, the system may perform inconsistently on different platforms (such as Windows, iOS, Android, and Hongmeng). Finally, there is no unified mapping table to correspond facial features to the physical condition of the human body's internal organs and possible Chinese medicine constitution.

[0007] This method extracts facial features through video, providing a non-invasive way of health assessment, avoiding blood drawing or other forms of physical intervention in traditional diagnostic methods. Through instant image analysis, results can be obtained quickly, giving users instant feedback, which helps to adjust their living habits or seek medical help in time; users only need to provide a photo, which is simple to operate and easy to use, especially suitable for the elderly or people who are not familiar with complex technology. With the popularity of smartphones, this function can be integrated into various applications, making it convenient for the public to conduct self-tests at any time. At present, this system has achieved consistent performance on Windows and Android, and iOS and Hongmeng are being implemented. Although this method cannot replace professional medical diagnosis, it can be used as an auxiliary tool to help doctors obtain preliminary health information of patients more quickly. In addition, providing personalized health advice based on individual differences will help promote the development of personalized health management. By continuously recording changes in user facial features, it is also convenient to track health trends in the long term and detect abnormalities in time.

[0008] The system includes: (1) Image reading module: This module converts the input image into a grayscale image through the OpenCV library for 68 facial feature point detection. (2) Face feature point detection module: This module is designed to detect 68 facial feature points in the input image. The transformed image is used to detect faces using Dlib's frontal face detector; the pre-trained "68 key point model" is loaded to detect the largest face, and the feature point coordinates are obtained through the feature point predictor, and finally a feature point list is returned. (3) Reflection area segmentation module: This module is designed to segment the facial reflex areas of the five internal organs and recognize colors. The bounding boxes of different facial areas are calculated based on the feature points, and each area is cropped from the original image. The color classification function is called for each area, and the cropped image list and color information list are returned. The segmentation diagram shows the facial reflex area segmentation of the five internal organs. Figure 3 . (4) Facial glossiness detection module: This module is designed to detect facial glossiness. It extracts feature points of the left and right facial regions, creates masks to obtain images of the left and right facial regions, converts the images to the LAB color space, calculates the average brightness of the L channel, and determines whether there is glossiness based on the average brightness. (5) Lip region segmentation module: This module is designed to segment the lip region and identify the lip color based on 68 facial feature points. The feature points of the lip region are extracted, a mask is created to obtain the image of the lip region, and the average RGB color of the lip region is calculated. According to the predefined color classification, the color classification closest to the average color is found. (6) Mapping module between facial features and the health status of the five internal organs and TCM constitution: This module aims to match facial features with human constitution. Through the mapping table 1 of facial features, the health status of the five internal organs and TCM constitution, facial features are matched with constitution types, and a unified fusion algorithm is used to determine the human constitution, and finally personalized suggestions on food, clothing, housing and transportation are given. See the image feature recognition system architecture in detail Figure 1 . Schematic diagram of facial feature recognition process Figure 2 . Segmentation of facial reflexology zones of the five internal organs Figure 3 . Table 1: Mapping of facial features, health status of the five internal organs and TCM constitution. Mapping table of facial features, health status of internal organs and constitution in traditional Chinese medicine Table 1 DETAILED DESCRIPTION

[0009] Step one: Integrate the entire facial detection system into the company's health management system.

[0010] Step 2: Receive clear facial images uploaded by the front end.

[0011] Step 3: Use the shape_predictor_68_face_landmarks.dat detection model to detect 68 facial landmarks.

[0012] Step 4: Select the facial reflex areas of the internal organs based on the 68 feature points and identify the color features of each area.

[0013] Step 5: Select the facial area based on the 68 feature points and detect the facial glossiness.

[0014] Step 6: Select the lip area based on the 68 feature points and identify the lip color features.

[0015] Step 7: Through the mapping table 1 between facial features and the physical conditions of the five internal organs of the human body and possible TCM constitutions, a unified fusion algorithm is used to determine the human constitution, and ultimately give personalized suggestions on food, clothing, housing and transportation.

[0016] Step 8: Feedback all recognition results to the system front end. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Facial feature recognition system architecture Figure 1 .

[0018] Schematic diagram of facial feature recognition process Figure 2 .

[0019] Segmentation of facial reflexology zones of the five internal organs Figure 3 .

[0020] Table 1: Mapping of facial features, health status of the five internal organs and TCM constitution.

Claims

1. A method based on facial feature extraction corresponding to human physique.

2. Mapping table 1 of facial features, health status of internal organs and TCM constitution.

Images