Color style makeup diagnosis analysis method and system

Abstract

The invention provides a color style makeup diagnosis analysis method and system, and relates to the technical field of computer vision, and the method comprises the steps: obtaining a non-makeup face image and makeup preference information uploaded by a user; performing chromatic aberration balance on the face image to obtain an image after chromatic aberration balance, and converting the image after chromatic aberration balance into an HSV image; extracting color information of each area in the HSV image, and integrating the color information into a complete color map through a color block splicing technology; analyzing and diagnosing the color map and the makeup preference information by using a preset makeup color diagnosis model to obtain a diagnosis result; and automatically generating a color analysis report according to the diagnosis result. According to the method, the face image is subjected to chromatic aberration balance and converted into the HSV image, so that the influence of external illumination and equipment difference on the color can be effectively eliminated, and the color information extraction precision is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer vision, in particular to a color style makeup diagnosis analysis method and system. BACKGROUND

[0002] In the modern beauty industry, the demand for personalized makeup is growing. Consumers want to be able to choose the most suitable makeup color according to their own color characteristics.

[0003] The invention patent with application number 202010298899.5 discloses a makeup evaluation method, device, cosmetic mirror and storage medium. The method includes: obtaining a face image of a user, detecting the face image through a trained makeup evaluation model to obtain a user makeup score and output, wherein the makeup evaluation model is a lightweight deep learning model after model compression, which is widely applicable and fast in calculation, can ensure the timeliness of makeup evaluation while ensuring the accuracy of makeup evaluation, and is especially suitable for cosmetic mirrors with limited device cost and general processor performance, thereby providing timely and relatively accurate makeup scores for users when they are making up, facilitating users to adjust the makeup according to the makeup feedback, and effectively improving the user experience. However, the makeup evaluation model is highly dependent on the quality of the input image. If the user provides an image that is blurred, insufficient in light, or has reflection, etc., the model may not be able to accurately evaluate the makeup. SUMMARY

[0004] In order to overcome the shortcomings of the prior art, the purpose of the present application is to provide a color style makeup diagnosis analysis method and system, which can effectively eliminate the influence of external light and device differences on color by balancing the color difference of the face image and converting it to an HSV image, thereby improving the accuracy of color information extraction.

[0005] To achieve the above purpose, the present application provides the following scheme:

[0006] A color style makeup diagnosis analysis method, comprising:

[0007] Obtaining an unmade-up face image and makeup preference information uploaded by a user;

[0008] Balancing the color difference of the face image to obtain a color difference balanced image, and converting the color difference balanced image into an HSV image;

[0009] Extracting color information of each region in the HSV image, and integrating the color information into a complete color atlas through color block splicing technology;

[0010] The preset makeup color diagnosis model is used to analyze and diagnose the color atlas and the makeup preference information, and a diagnosis result is obtained; the diagnosis result includes color preference and optimal makeup color of the user;

[0011] A color analysis report is automatically generated according to the diagnosis result.

[0012] Preferably, the makeup preference information includes preferred color, makeup style, scene preference, skin type and skin color, past makeup habit and seasonal preference of the user.

[0013] Preferably, the face image is subjected to color difference balancing to obtain a color difference balanced image, and the color difference balanced image is converted into an HSV image, including:

[0014] The pixel mean values of the face image in RGB three channels are calculated; wherein the pixel mean value calculation formula is: wherein D represents the pixel mean value, represents the pixel mean value of the face image in the red channel, represents the pixel mean value of the face image in the green channel, represents the pixel mean value of the face image in the blue channel;

[0015] The balancing coefficients of each color channel are calculated according to the pixel mean values;

[0016] The face image is subjected to color difference balancing by using the balancing coefficients to obtain a color difference balanced image;

[0017] Each color channel of the color difference balanced image is subjected to normalization processing to obtain a normalization result;

[0018] The extreme value in the normalization result is selected;

[0019] A color space conversion formula is constructed according to the normalization result and the extreme value in the normalization result;

[0020] The color difference balanced image is converted into an HSV image by using the color space conversion formula.

[0021] Preferably, the extreme value in the normalization result is selected, including:

[0022] The formula The extreme value in the normalization result is selected; wherein C max represents the maximum value in the normalization result, and C min represents the minimum value in the normalization result.

[0023] Preferably, the color space conversion formula is:

[0024]

[0025] V = C max

[0026] Wherein, H represents the hue component, S represents the saturation component, and V represents the lightness component.

[0027] Preferably, color information of each region in the HSV image is extracted, and the color information is integrated into a complete color atlas by color block splicing technology, including:

[0028] The HSV image is segmented into different facial regions using facial recognition technology; the facial regions include the eye region, the lip region, and the face region;

[0029] Color information is extracted from the HSV image of each facial region; the color information includes hue and saturation;

[0030] According to the extracted color information, a corresponding color block is generated for each facial region;

[0031] All color blocks are spliced into a complete color atlas according to a preset order of the face of each facial region.

[0032] Preferably, the method for constructing the makeup color diagnosis model comprises:

[0033] An initial convolutional neural network is obtained; the convolutional neural network comprises an input layer, a convolutional layer, a fully connected layer, a fusion layer, and an output layer connected in sequence; wherein the input layer is used to obtain the color atlas and the makeup preference information; the convolutional layer is used to extract spatial features of the color atlas to obtain color spectrum features; the fully connected layer is used to combine the color spectrum features and the makeup preference information to obtain combined features; the fusion layer is used to fuse the combined features to obtain fusion features; and the output layer is used to perform diagnostic analysis according to the fusion features to obtain a diagnosis result;

[0034] Obtain sample atlas and sample preference information;

[0035] The sample atlas and the sample preference information are input into the convolutional neural network for training to obtain the makeup color diagnosis model.

[0036] A color style makeup diagnosis and analysis system comprises:

[0037] A data acquisition module is configured to acquire an unmade-up face image and makeup preference information uploaded by a user;

[0038] An image processing module is configured to balance color difference of the face image to obtain a color-difference-balanced image, and convert the color-difference-balanced image into an HSV image;

[0039] a graph construction module, configured to extract color information of each region in the HSV image, and integrate the color information into a complete color graph by using a color block splicing technology;

[0040] an analysis module, configured to analyze and diagnose the color graph and the makeup preference information by using a preset makeup color diagnosis model, and obtain a diagnosis result; the diagnosis result includes color preference of the user and optimal makeup color;

[0041] a report generation module, configured to automatically generate a color analysis report according to the diagnosis result.

[0042] According to the specific embodiments provided by the present application, the following technical effects are achieved:

[0043] The present application provides a color style makeup diagnosis and analysis method and system, which comprises the following steps: obtaining an unmade-up face image and makeup preference information uploaded by a user; performing color difference balancing on the face image to obtain a color difference balanced image, and converting the color difference balanced image into an HSV image; extracting color information of each region in the HSV image, and integrating the color information into a complete color graph by using a color block splicing technology; analyzing and diagnosing the color graph and the makeup preference information by using a preset makeup color diagnosis model, and obtaining a diagnosis result; the diagnosis result includes color preference of the user and optimal makeup color; and automatically generating a color analysis report according to the diagnosis result. By performing color difference balancing on the face image and converting the face image into an HSV image, the present application can effectively eliminate the influence of external light and equipment difference on color, thereby improving the accuracy of color information extraction; and by using a preset makeup color diagnosis model, the system can quickly analyze the color graph and the makeup preference information of the user, provide accurate color diagnosis results, and save time and effort of the user. BRIEF DESCRIPTION OF DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0045] Figure 1 a method flowchart provided for the embodiments of the present application;

[0046] Figure 2 a system structure schematic diagram provided for the embodiments of the present application. DETAILED DESCRIPTION

[0047] With reference to the accompanying drawings, the technical solutions in the embodiments of the present application will be described clearly and completely. Obviously, the described embodiments are only a part of the embodiments of the present application, but not all the embodiments. Based on the embodiments of the present application, all the other embodiments obtained by a person of ordinary skill in the art without creative effort belong to the scope of the present application.

[0048] The present application aims to provide a color style makeup diagnostic analysis method and system, which can effectively eliminate the influence of external light and equipment difference on color by balancing color difference and converting a face image into an HSV image, thereby improving the accuracy of color information extraction.

[0049] To make the above-mentioned purposes, features and advantages of the present application more apparent and easy to understand, the present application will be described in further detail below with reference to the accompanying drawings and specific embodiments.

[0050] Figure 1 As shown in the method flowchart provided by the embodiments of the present application, Figure 1 As shown in the method flowchart provided by the embodiments of the present application,

[0051] Step 100: obtaining an unmade-up face image uploaded by a user and makeup preference information;

[0052] Step 200: balancing color difference of the face image to obtain a color-difference-balanced image, and converting the color-difference-balanced image into an HSV image;

[0053] Step 300: extracting color information of each region in the HSV image, and integrating the color information into a complete color atlas through color block splicing technology;

[0054] Step 400: analyzing and diagnosing the color atlas and the makeup preference information by using a preset makeup color diagnostic model to obtain a diagnostic result; the diagnostic result includes color preference of the user and optimal makeup color;

[0055] Step 500: automatically generating a color analysis report according to the diagnostic result.

[0056] Specifically, the present embodiment first designs a user interface, first, a friendly user interface (UI) is designed, so that the user can conveniently upload an unmade-up face image and input makeup preference information. The interface includes:

[0057] 1. Image uploading function: a button or a drag-and-drop area is provided to allow the user to select and upload his unmade-up face image. Multiple image formats (such as JPEG, PNG, etc.) are supported.

[0058] 2. Makeup preference information input: Set multiple input fields or selection boxes for users to fill in or select their makeup preference information, including color preference, makeup style, suitable occasion, etc.

[0059] Further, the embodiment performs the following processing after the image upload and processing:

[0060] 1. Image verification: Check the format and size of the uploaded image to ensure that it meets the system requirements. If not, prompt the user to upload again.

[0061] 2. Image preprocessing: Perform preliminary processing on the uploaded image, such as adjusting image size, removing background noise, etc., to ensure the accuracy of subsequent analysis.

[0062] Further, the embodiment performs the following operations after the user inputs makeup preference information on the interface:

[0063] 1. Data collection: Store the user's inputted makeup preference information (such as color preference, makeup style, occasion, etc.) in a structured manner in the database for subsequent analysis.

[0064] 2. Information verification: Perform basic verification on the user's inputted preference information to ensure the integrity and validity of the information. For example, check if the required items are selected.

[0065] Further, the embodiment finally performs data integration and transmission, integrating the user's uploaded makeup-free face image and makeup preference information into a data structure for subsequent analysis and processing; for example, the embodiment creates an object or dictionary containing the user's image and makeup preference information for easy transmission and use in subsequent steps, and then transmits the integrated data to the makeup color diagnosis model for subsequent color analysis and diagnosis.

[0066] Through the above steps, the present application can effectively obtain the user's uploaded makeup-free face image and makeup preference information, providing basic data support for subsequent makeup color diagnosis.

[0067] Preferably, the makeup preference information includes: user's preferred color, makeup style, scene preference, skin type and skin color, past makeup habits and seasonal preference.

[0068] Specifically, the user preferred color of the embodiment is the color that the user tends to use in daily makeup, such as preferring warm color, cold color or neutral color, etc.; the makeup style is the makeup style that the user likes, for example, natural, mature, simple, retro, fashion, etc., which can affect the selection of color; the occasion preference is the occasion information of the makeup, such as daily, work, dinner party, party, wedding, etc., different occasions may need different makeup colors. Skin type and skin color are the skin type (such as oily, dry, mixed, sensitive) and specific skin color information of the user, which help to adjust the color recommendation to suit individual characteristics. Past makeup habits are the makeup schemes or products that the user often uses in the past, which help to analyze the habitual selection of the user for certain colors or combinations; seasonal preference is the user's makeup color preference for different seasons of the four seasons, for example, bright color in summer or deep color in winter, etc. Demonstratively, the makeup preference information of the embodiment also includes special requirements or taboos, i.e. special requirements or avoidance of certain colors by the user. The preference information of the embodiment can better recommend suitable makeup colors for the user in combination with the unmade-up face image of the user and the color diagnosis model, and provide personalized color analysis report.

[0069] Preferably, the color difference balance is performed on the face image to obtain a color difference balanced image, and the color difference balanced image is converted into an HSV image, comprising:

[0070] The pixel mean values of the face image in the RGB three channels are calculated; wherein the pixel mean value calculation formula is: wherein D represents the pixel mean value, represents the pixel mean value of the face image in the red channel, represents the pixel mean value of the face image in the green channel, represents the pixel mean value of the face image in the blue channel;

[0071] The balance coefficient of each color channel is calculated according to the pixel mean value;

[0072] The color difference balance is performed on the face image by using the balance coefficient to obtain a color difference balanced image;

[0073] The normalization processing is performed on each color channel of the color difference balanced image to obtain a normalization result;

[0074] The extreme value in the normalization result is selected;

[0075] The color space conversion formula is constructed according to the normalization result and the extreme value of the normalization result;

[0076] The color difference balanced image is converted into an HSV image by using the color space conversion formula.

[0077] Preferably, the extreme value in the normalized result is selected, including:

[0078] The formula is used The extreme value in the normalized result is selected; wherein, C max represents the maximum value in the normalized result, C min represents the minimum value in the normalized result.

[0079] Specifically, when the external light condition changes, the color of the object surface is unchanged, but the color of the object collected by the imaging device will deviate from the true color of the object due to the change of the light. The color difference balancing algorithm of the embodiment can compensate for the defect of the imaging device and correct the color difference of the image.

[0080] Preferably, the color space conversion formula is:

[0081]

[0082] V=C max

[0083] Wherein, H represents the hue component, S represents the saturation component, and V represents the brightness component.

[0084] Specifically, in the image enhancement process, the image needs to be subjected to illumination compensation or color correction processing. This process needs to separate the illumination component and the reflection component from the image. The face image is generally an RGB image, and the components cannot be directly separated and compensated. Therefore, in the process of related enhancement processing of the degraded image of a complex scene, the image needs to be preprocessed by color space conversion to convert it to the HSV space.

[0085] Preferably, the color information of each region in the HSV image is extracted, and the color information is integrated into a complete color atlas by color block splicing technology, including:

[0086] The HSV image is segmented into different face regions by using face recognition technology; the face regions include eyes, lips, and face;

[0087] Color information is extracted from the HSV image of each face region; the color information includes hue and saturation;

[0088] According to the extracted color information, a corresponding color block is generated for each face region;

[0089] All color blocks are spliced into a complete color atlas according to the preset order of the face regions of the face.

[0090] Specifically, the color atlas construction process of the embodiment includes:

[0091] 1. Face region segmentation, first, this embodiment uses face recognition technology to process the HSV image to identify and segment different face regions. The specific steps include:

[0092] (1) Face detection, apply a pre-trained face detection model (such as Haar cascade classifier or deep learning model) to locate the position of the face in the image.

[0093] Region division, within the detected face region, further use key point detection algorithm (such as Dlib or MediaPipe) to identify face feature points, divide eye, lip and face regions. The coordinate information of each region will be used for subsequent color extraction.

[0094] 2. Color information extraction after completing the segmentation of face region, this embodiment will extract color information from each region of HSV image. The specific steps are as follows:

[0095] (1) Region extraction, according to the region coordinates divided before, extract the sub image of each face region from the HSV image.

[0096] (2) Color analysis, color analysis is carried out for each sub image, and its hue (Hue) and saturation (Saturation) are calculated. The representative color information can be obtained by calculating the average or median of HSV value of each region.

[0097] 3. Color block generation according to the extracted color information, generate corresponding color block for each face region. The specific steps include:

[0098] (1) Color block creation, use the extracted hue and saturation value to generate the corresponding color block. The color of the color block can be realized by mapping the hue value to the RGB color space.

[0099] (2) Color block size adjustment, set uniform size for each color block to ensure neatness and beauty when splicing.

[0100] 4. Color block splicing, finally, this embodiment splices all generated color blocks into a complete color atlas according to the preset order of face region. The specific steps are as follows:

[0101] (1) Splicing order, according to the preset order of face region (such as eye, lip, face), arrange the color blocks in turn.

[0102] (2) Image synthesis, use image processing library (such as OpenCV or PIL) to synthesize each color block into a complete color atlas. You can choose horizontal or vertical splicing to ensure good visual effect of the final atlas.

[0103] (3) Output the results, and the generated complete color atlas will be used as input for subsequent analysis, providing makeup color diagnosis model.

[0104] Through the above steps, the embodiment can effectively extract color information in the HSV image and integrate it into a complete color atlas, providing personalized makeup color suggestions for users.

[0105] Specifically, when splicing color blocks to form a complete color atlas, the preset order of the face area generally follows the natural face structure and the common order of makeup. The splicing order of the embodiment is as follows:

[0106] The eye area, including the color blocks of the left eye and the right eye, is located at the top or the upper left corner of the color atlas. The color information of the eye area is crucial for the selection of eye shadow and eyeliner.

[0107] The lip area, following the eye area, is located in the middle part or below of the color atlas. The color information of the lip area is very important for the selection of lipstick and lip gloss.

[0108] The face area, including the color blocks of the entire face, is usually located at the bottom or the right side of the color atlas. The color information of the face area helps to select foundation, blush, and highlight products.

[0109] Further, the implementation process of the color block splicing technology of the embodiment includes the following steps:

[0110] 1. Color block generation: Extract color information from the HSV image of each face area, calculate the dominant color and saturation of each area; generate a corresponding color block for each face area based on the extracted color information, usually using uniform size and shape (such as square or rectangle).

[0111] 2. Determine the splicing order: Determine the splicing order of the color blocks (such as eye area, lip area, face area) according to the face structure and makeup habits, and ensure that the spliced color atlas conforms to the natural visual flow.

[0112] 3. Splicing operation: Use image processing libraries (such as OpenCV or PIL) to splice the generated color blocks in the specified order to form a complete color atlas; during splicing, ensure that the spacing and alignment between color blocks are consistent to maintain the overall visual order and beauty.

[0113] 4. Output results: Save the spliced color atlas as an image file or display it on the user interface for subsequent analysis and use.

[0114] Through the above steps, the color block splicing technology of the embodiment can effectively integrate the color information of different face areas into a complete color atlas, facilitating subsequent makeup color diagnosis and suggestions.

[0115] Exemplarily, the splicing example sequence of the present embodiment is as follows:

[0116] 1, top: eye color blocks (left eye color block, right eye color block); 2, middle: lip color block; 3, bottom: face color block. This sequence of the present embodiment not only conforms to the natural structure of the face, but also helps the user to more intuitively understand the color information of different regions, so as to make a more suitable makeup selection.

[0117] Preferably, the method for constructing the makeup color diagnosis model comprises:

[0118] obtaining an initial convolutional neural network; the convolutional neural network comprises an input layer, a convolutional layer, a fully connected layer, a fusion layer and an output layer connected in turn; wherein the input layer is used to obtain the color atlas and the makeup preference information; the convolutional layer is used to extract spatial features of the color atlas to obtain color spectrum feature; the fully connected layer is used to combine the color spectrum feature and the makeup preference information to obtain a combination feature; the fusion layer is used to fuse the combination feature to obtain a fusion feature; and the output layer is used to diagnose and analyze according to the fusion feature to obtain a diagnosis result;

[0119] obtaining sample atlas and sample preference information;

[0120] inputting the sample atlas and the sample preference information into the convolutional neural network for training to obtain the makeup color diagnosis model.

[0121] Specifically, the model construction process of the present embodiment comprises:

[0122] 1, collect a large amount of user data, including their color atlas and related makeup preference information. The sample atlas includes color block combination of different facial regions, and the sample preference information should include the user's preference for color and style.

[0123] 2, convert the color atlas into a format suitable for input into the neural network (such as normalization processing and size standardization). Encode the makeup preference information so that it can be used for combination with the atlas feature (for example, use label encoding or embedding technology).

[0124] 3, input layer design, the present embodiment defines a double input structure, one input color atlas, and the other input makeup preference information.

[0125] 3, convolutional layer design, in the color atlas branch, a plurality of convolutional layers and pooling layers are applied to gradually extract and reduce spatial features. Different size convolution kernels are used to extract various atlas details and patterns.

[0126] 4. Full connection layer design, linear combination between the extracted features from convolution and the encoded makeup preferences. Activation functions (e.g. ReLU) are used to provide non-linear capabilities to the network.

[0127] 5. Fusion layer design, fusion of the combined color atlas features and preference information. This embodiment uses simple concatenation or complex fusion strategies using attention mechanisms.

[0128] 6. Output layer design, application of activation functions and output structures to output color diagnosis results according to task requirements. Softmax is used for classification results, and linear activation is used for regression problems.

[0129] Further, the model training process of the embodiment includes:

[0130] 1. Training data input, input sample atlas and sample preference information into the prepared network structure.

[0131] 2. Loss function and optimization, define the loss function (such as cross-entropy loss for classification, MSE for regression), evaluate the error between the model prediction and the true label through this function. Use optimization algorithms (e.g. Adam or SGD) for gradient descent, update network weights to minimize loss.

[0132] 3. Iterative training, repeat the training process through multiple rounds of epochs to gradually improve the prediction accuracy of the model.

[0133] 4. Diagnosis analysis, the network captures the complex relationship between color atlas and preference information by learning fusion features. The diagnosis model uses these features to make the final color suggestion, where the color atlas provides the visual basis and the preference information ensures the individual matching.

[0134] 5. Output result generation, output the user's color preference and the optimal makeup scheme, these results are directly output or further processed to generate user reports.

[0135] Through the above steps, the makeup color diagnosis model of the embodiment can effectively combine the image features and preference information uploaded by the user to provide accurate color analysis and suggestions for the user.

[0136] Further, the diagnosis process of the diagnosis model of the embodiment which captures the complex relationship between color atlas and preference information by learning fusion features includes:

[0137] 1. Data preparation

[0138] a. Color atlas collection and standardization: Extract the standardized color atlas from the user image, which represents the color characteristics of each facial region (such as eyes, lips, face). Then convert the color atlas data into a format suitable for model input (such as matrix form), and standardize each color value (usually between 0 and 1).

[0139] b. Encoding preference information: Use label encoding or one-hot encoding to structure the user's makeup preference into a format that the model can process. The preference information in this embodiment is complex (e.g. includes multiple features: style, color preference), and an embedding layer is used to reduce the dimensionality while retaining useful information.

[0140] 2. Model structure

[0141] a. Color atlas feature extraction (convolution branch) includes:

[0142] (1) Input layer, receiving normalized data of color atlas.

[0143] (2) Convolution layer, using a multi-layer convolutional neural network (CNN) to extract local features of color, including multiple convolution kernels of different sizes to capture features of different scales.

[0144] (3) Pooling layer: Use max-pooling or average-pooling layer to gradually reduce the size of the feature map and improve position invariance.

[0145] b. Preference information extraction (embedding branch), including:

[0146] a. Fully connected layer, which condenses preference information through one or two fully connected networks into a feature representation. At the same time, apply a nonlinear activation function (such as ReLU) to enhance the expression ability.

[0147] 3. Fusion layer - feature fusion, combining convolution features and preference features through vector concatenation or weighted fusion methods. This embodiment uses attention mechanism to enhance the weight of important features.

[0148] 4. Output layer and analysis: Process the synthesized features after fusion through the last one or several fully connected networks to get the final prediction.

[0149] This embodiment outputs specific results for the task. For classification tasks, use the Softmax layer to output the probability of each color scheme. Regression tasks get continuous output values through linear activation, representing accurate color suggestions.

[0150] 5. Model training and evaluation

[0151] a. Training-data split, divide the dataset into training, validation, and test parts; loss function: choose the corresponding loss function (such as cross-entropy for classification, mean square error for regression); use Adam or RMSprop to optimize the loss.

[0152] b. Evaluation and tuning, adjust hyperparameters using the validation set to avoid overfitting; observe the model performance on the test set and adjust the model structure / strategy to improve accuracy.

[0153] This embodiment can accurately combine atlas features and personal preferences to achieve precise makeup color diagnosis and provide personalized recommendations. The success of the model lies in its meticulous handling and structuring of data, as well as the reasonable strategy selection during the fusion stage, which ensures the optimal matching of the visual basis of the color atlas and the individualization of preference information, thereby reducing the defect of insufficient information disclosure.

[0154] Optionally, the diagnostic results of the embodiment include:

[0155] 1. User color preference analysis

[0156] a. Color preference category: based on user input preference information and model analysis, provide the user's possible color category (such as warm tone, cold tone, neutral color, etc.).

[0157] b. Hue and saturation suggestions: specifically suggest the hue (Hue) and saturation (Saturation) that the user should use in the makeup, helping the user choose the appropriate makeup color.

[0158] 2. Optimal makeup color recommendation

[0159] Facial region color suggestions:

[0160] a. Eye area: recommend eye shadow and eyeliner colors suitable for the user, which may include specific color numbers or color systems.

[0161] b. Lip area: suggest appropriate lipstick and lip gloss colors, providing specific color numbers or color systems.

[0162] c. Face: recommend appropriate blush and foundation tones to help users choose products that match their skin tone.

[0163] 3. Personalized makeup style suggestions

[0164] Makeup style matching: based on user preferences and color analysis, suggest appropriate makeup styles (such as natural, heavy makeup, retro, etc.), helping users choose the right makeup for different occasions.

[0165] 4. Color matching scheme

[0166] a. Overall color matching: Provide color matching solutions for overall makeup, ensure color harmony and unity in each facial area.

[0167] b. Color combination suggestions: Suggest how to combine different colors to achieve the best visual effect, which may include color gradient, contrast color use and other techniques.

[0168] Further, the embodiment also includes data storage and analysis steps, that is, storing user's color preference data in the database. Through big data analysis, continuously optimize the color diagnosis algorithm, improve the accuracy and individuality of diagnosis. This embodiment provides accurate color style makeup diagnosis analysis for customers through advanced image processing and machine learning technology. The system not only meets the individual needs of customers, but also continuously improves service quality through big data analysis. In the future, with the continuous progress of technology, this embodiment will bring more innovation and possibility to the beauty industry.

[0169] Corresponding to the above method, the embodiment also provides a color style makeup diagnosis analysis system, as shown in Figure 2 , comprising:

[0170] A data acquisition module is configured to acquire an unmade-up face image and makeup preference information uploaded by a user.

[0171] An image processing module is configured to balance color difference of the face image to obtain a color difference balanced image, and convert the color difference balanced image into an HSV image.

[0172] A graph construction module is configured to extract color information of each region in the HSV image, and integrate the color information into a complete color graph through color block splicing technology.

[0173] An analysis module is configured to analyze and diagnose the color graph and the makeup preference information by using a preset makeup color diagnosis model to obtain a diagnosis result; the diagnosis result includes color preference of the user and optimal makeup color.

[0174] A report generation module is configured to automatically generate a color analysis report according to the diagnosis result.

[0175] The beneficial effects of the present application are as follows:

[0176] (1) The present application acquires an unmade-up face image and makeup preference information uploaded by a user, which can provide personalized makeup suggestions to meet the unique needs and aesthetic preferences of users.

[0177] (2) The present application balances color difference of the face image and converts it into an HSV image, which can effectively eliminate the influence of external light and equipment difference on color, thereby improving the accuracy of color information extraction.

[0178] (3) The application uses color block splicing technology to integrate color information of different areas, creating a complete color atlas, making the makeup color recommendation more comprehensive and detailed.

[0179] (4) The application uses a preset makeup color diagnosis model, and the system can quickly analyze the color atlas and makeup preference information of the user, provide accurate color diagnosis results, and save the time and effort of the user.

[0180] The embodiments in the specification are described in a progressive manner, and each embodiment focuses on the difference from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method part.

[0181] The principles and implementation modes of the application are described by applying specific examples in this paper, and the above embodiment description is only used to help understand the method and core idea of the application; at the same time, for the general technical personnel in the art, according to the idea of the application, the specific implementation mode and application range will be changed. In summary, the content of the specification should not be understood as a limitation of the application.

Claims

1. A color style makeup diagnostic analysis method, characterized by, The method comprises the following steps: obtaining an uploaded face image without makeup and makeup preference information of a user; performing color difference balancing on the face image to obtain a color difference balanced image, and converting the color difference balanced image into an HSV image; extracting color information of each region in the HSV image, and integrating the color information into a complete color atlas through color block splicing technology; analyzing and diagnosing the color atlas and the makeup preference information by using a preset makeup color diagnosis model to obtain a diagnosis result; the diagnosis result comprises color preference and optimal makeup color of the user; generating a color analysis report automatically according to the diagnosis result.

2. The color style makeup look diagnostic analysis method of claim 1, wherein, The makeup preference information comprises preferred color, makeup style, scene preference, skin type and color, past makeup habit and seasonal preference of the user.

3. The color style makeup look diagnostic analysis method of claim 1, wherein, The color difference balancing on the face image to obtain a color difference balanced image, and converting the color difference balanced image into an HSV image comprises the following steps: Calculate the pixel mean value of the face image in the RGB three channels; wherein, the pixel mean value calculation formula is: Wherein, D represents the pixel mean value, R represents the pixel mean value of the face image in the red channel, G represents the pixel mean value of the face image in the green channel, B represents the pixel mean value of the face image in the blue channel; calculating a balancing coefficient of each color channel according to the pixel mean value; performing color difference balancing on the face image by using the balancing coefficient to obtain a color difference balanced image; performing normalization processing on each color channel of the color difference balanced image to obtain a normalization result; selecting an extreme value in the normalization result; constructing a color space conversion formula according to the normalization result and the extreme value of the normalization result; converting the color difference balanced image into an HSV image by using the color space conversion formula.

4. The color style makeup look diagnostic analysis method of claim 3, wherein, The selecting of the extreme value in the normalization result comprises the following steps: The formula is The extreme value in the normalized result is selected; wherein, C max represents the maximum value in the normalized result, C min represents the minimum value in the normalized result.

5. The color style makeup look diagnostic analysis method of claim 4, wherein, The color space conversion formula is as follows: V = C max wherein, H represents a hue component, S represents a saturation component, and V represents a brightness component.

6. The color style makeup diagnostic analysis method according to claim 1, characterized by, The extracting of color information of each region in the HSV image, and the integrating of the color information into a complete color atlas through color block splicing technology comprises the following steps: segmenting the HSV image into different facial regions by using facial recognition technology; the facial regions comprise eye region, lip region and face region; extracting color information from the HSV image of each facial region; the color information comprises hue and saturation; generating a corresponding color block for each facial region according to the extracted color information; splicing all the color blocks into a complete color atlas according to a preset order of the face of the facial region.

7. The color style makeup look diagnostic analysis method of claim 1, wherein, The method for constructing the makeup color diagnosis model comprises the following steps: obtaining an initial convolutional neural network; the convolutional neural network comprises an input layer, a convolutional layer, a full connection layer, a fusion layer and an output layer connected in sequence; wherein, the input layer is used for obtaining the color atlas and the makeup preference information; the convolutional layer is used for performing spatial feature extraction on the color atlas to obtain a color spectrum feature; the full connection layer is used for combining the color spectrum feature and the makeup preference information to obtain a combined feature; the fusion layer is used for fusing the combined feature to obtain a fused feature; and the output layer is used for diagnosing and analyzing according to the fused feature to obtain a diagnosis result; obtaining sample atlas and sample preference information; inputting the sample atlas and the sample preference information into the convolutional neural network for training to obtain the makeup color diagnosis model.

8. A color style makeup diagnostic analysis system, characterized by, The method comprises the following steps: The data acquisition module is configured to acquire an uploaded face image without makeup and makeup preference information of a user; The image processing module is configured to perform color difference balancing on the face image to obtain a color-difference-balanced image, and convert the color-difference-balanced image into an HSV image; The atlas construction module is configured to extract color information of each region in the HSV image, and integrate the color information into a complete color atlas through a color block splicing technique; The analysis module is configured to analyze and diagnose the color atlas and the makeup preference information by using a preset makeup color diagnosis model to obtain a diagnosis result; the diagnosis result includes color preference and optimal makeup color of the user; The report generation module is configured to automatically generate a color analysis report according to the diagnosis result.

Images