Search

CN-121998925-A - Multi-model hair detection and hair volume density calculation method based on image input

CN121998925ACN 121998925 ACN121998925 ACN 121998925ACN-121998925-A

Abstract

The invention provides a hair detection and hair volume density calculation method based on image input and multi-model analysis. The method comprises the steps of collecting head images by using common imaging equipment, preprocessing the images, inputting the images into a plurality of analysis models for independent analysis, optimizing the hair detection and hair volume density calculation precision through a self-grinding fusion algorithm and high-order multi-task analysis, outputting hair volume estimation, zoned hair density, scalp coverage rate, health score, thermodynamic diagram and trend curve, and generating personalized nursing suggestions. Furthermore, the method periodically acquires images and related data, stores the images and the related data as historical data, and periodically analyzes the historical data to generate a hair health change trend, and the system can access an additional analysis model or external sensor data through an expansion interface to realize long-term tracking and comprehensive evaluation, so that the evaluation precision and the use value of a user are improved.

Inventors

  • Request for anonymity

Assignees

  • 陈晨

Dates

Publication Date
20260508
Application Date
20260119

Claims (10)

  1. 1. The main claim 1 is a hair detection and hair volume density calculation method based on image input and multi-model analysis, comprising the steps of: s1, acquiring a head image by using common imaging equipment; S2, preprocessing the image, including denoising, enhancement and standardization; S3, inputting the image into a plurality of artificial intelligence models for analysis, wherein the artificial intelligence models comprise a hair and scalp segmentation model, a hair quantity and density estimation model, a hair health score and a trend prediction model; S4, carrying out weighted integration on the output results of the multiple models through a self-grinding fusion algorithm, and optimizing the hair detection and hair volume density calculation accuracy by utilizing high-order multitask analysis; S5, outputting multi-dimensional evaluation including hair volume estimation, zoned hair density, scalp coverage rate, health score, thermodynamic diagram and trend curve, and generating personalized nursing advice; s6, periodically acquiring the images and related data, storing the images and the related data as historical data, periodically analyzing the historical data to generate a hair health change trend, and accessing an additional analysis model or external sensor data through an expansion interface to realize long-term tracking and comprehensive evaluation.
  2. 2. The method according to the dependent claim 1, wherein said preprocessing of step S2 comprises image color normalization, denoising and contrast enhancement.
  3. 3. The method according to the dependent claim 1, wherein the segmentation model of step S3 employs a general convolutional neural network structure for identifying hair regions and scalp regions.
  4. 4. The method according to the dependent claim 1, wherein the number and density estimation model of S3 steps improves the detection accuracy by means of a high-order feature extraction and attention mechanism.
  5. 5. The method according to the dependent claim 1, wherein the self-grinding fusion algorithm of the step S4 performs a weighting process on the uncertainty of each model output, so as to realize the optimal integration of multiple model results.
  6. 6. The method according to the main claim 1, wherein said personalized care advice of step S5 is generated from user history image data and multi-dimensional assessment results.
  7. 7. The method according to the dependent claim 1, wherein the periodic analysis of step S6 comprises a statistical calculation of historical data of hair number, density and health score and generating a trend curve.
  8. 8. The method according to the main claim 1, wherein said expansion interface is used for accessing additional analysis models, including but not limited to scalp environment analysis models or wearable sensor data interfaces.
  9. 9. The method according to the dependent claim 1, wherein the personalized care advice generated by the long-term tracking is presented in association with the historical trend results to enhance the user experience and the accuracy of the assessment.
  10. 10. The method according to the dependent claim 1, wherein the history data store comprises data version management, ensuring that the periodic analysis can compare data at different points in time.

Description

Multi-model hair detection and hair volume density calculation method based on image input Technical Field The invention relates to the technical field of hair health detection and image analysis, in particular to a hair detection and hair volume density calculation method based on image input and multi-model analysis, the method can utilize common imaging equipment to collect images, complete hair segmentation, hair volume estimation, density calculation and trend analysis through a self-grinding multi-model fusion algorithm, and generate a visual report and personalized nursing suggestions. Background The existing hair detection technology relies on polarized light devices, skin mirrors or high-magnification microscopic imagers, and is high in cost, large in size, complex to operate and generally used only in professional institutions or medical scenes. The image-based hair analysis method is dependent on a single model, cannot realize multi-dimensional analysis of hair quantity, density, health score and trend prediction at the same time, and lacks continuity and expansibility. With the development of mobile devices and AI technology, an average user can easily acquire head images, but there is still a lack of a long-term traceable hair health assessment scheme without professional equipment. Disclosure of Invention Aiming at the problems of high cost, complex operation, single function and the like in the prior art, the invention provides a method for realizing hair health assessment based on common images without professional equipment, and the detection precision and reliability are improved through a self-grinding fusion algorithm and high-order multi-task analysis. Technical proposal The invention provides a hair detection and hair volume density calculation method based on image input and multi-model analysis, which comprises the following steps: S1, image acquisition And equipment such as a smart phone, a common camera and the like is used for shooting head images, so that different illumination conditions are supported. S2, image preprocessing Denoising, enhancing and standardizing the acquired image, and optionally carrying out illumination correction and contrast adjustment. S3, multimodal analysis Inputting an image into a plurality of artificial intelligence model analyses, comprising: A hair and scalp segmentation model for identifying hair regions and scalp regions; A hair number and density estimation model for calculating local and global hair number and distribution density; a hair health score and trend prediction model for assessing hair health and predicting future trend of change. S4, self-grinding fusion algorithm and high-order multitasking analysis And carrying out weighted integration on the output results of the models, processing uncertainty and prediction deviation, and generating a final hair health evaluation result. The high-order multitasking analysis can optimize the quantity, the density and the health score simultaneously, so that the overall evaluation accuracy is improved. S5, outputting and visualizing results And outputting multi-dimensional evaluation contents including hair volume estimation, zoned hair density, scalp coverage, health score, thermodynamic diagram and trend curve. And generating personalized care suggestions and long-term trend analysis reports according to the historical image data. The results may be presented at a mobile or web-side interface and may be stored for long-term tracking and periodic comparative analysis. S6 The system acquires the head image and related data of the user according to a preset period (such as day, week and month) and stores the head image and related data as historical data. The historical data can be recorded through version management, so that the data at different time points can be compared. The periodic analysis module performs statistical processing on the quantity, the density and the health score of the hair in the historical data to generate a hair health trend curve. The system can be accessed into other analysis models or external sensor data through an expansion interface to enhance the long-term tracking function. And generating personalized nursing suggestions according to the historical trend and the current detection result, and displaying the nursing suggestions in a correlated manner with the trend result so as to improve the user experience and the reference value. Advantageous effects The invention has the advantages of no need of special equipment, low cost and simple operation, and is suitable for family self-test, medical auxiliary diagnosis and hair care institutions. The multi-model fusion and the high-order multi-task analysis improve the hair detection precision and reliability. Hair number, density, health score, trend prediction, and personalized care advice may be provided simultaneously. The historical images of the user can be tracked for a long time, and short-term, medium-term and lon