Intelligent detection method for skin care effect based on image features
Patent Information
- Application Number
- CN202511182294.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-22
- Publication Date
- 2026-10-20
AI Technical Summary
[0003]不进行图像增强直接对图像进行检测，会导致部分不太明显的粗大毛孔无法被检测到
[0037]本发明的技术方案的有益效果是：本发明通过获取脸部皮肤图像，根据每个像素点的灰度值获取每个像素点的粗毛孔概率，进而得到疑似粗毛孔像素点，将疑似粗毛孔像素点分为多个类别，根据每个疑似粗毛孔像素点的粗毛孔概率、每个类别以及每个图像块中疑似粗毛孔像素点的数量，获取每个疑似粗毛孔像素点的增强权重，根据增强权重调整增强窗口大小，确保了更可能为粗毛孔的像素点的增强窗口大，从而使得更可能为粗毛孔的像素点的对比度增强程度大，对于干扰区域增强窗口小，防止了干扰区域对皮肤护理效果的检测影响。本发明针对性地增强粗毛孔区域，使得粗毛孔更加清晰明显，可以更好的帮助护理人员更细致的观察皮肤护理效果，使得皮肤护理效果的检测更加准确。
Smart Images
Figure BDA0005561115850000021 
Figure BDA0005561115850000031 
Figure BDA0005561115850000061
Abstract
Description
Technical Field
[0001] This invention relates to the field of image processing technology, and more specifically to an intelligent detection method for skin care effects based on image features. Background Technology
[0002] With the development of economy and science and technology, people are paying more and more attention to their skin care, and there are more and more methods for skin care. Therefore, after using some skin care products, it is necessary to test the effect of skin care. Intelligent detection methods for skin care effect based on image features can help evaluate the effect of skin care products or treatments on the skin and provide personalized suggestions. Among them, large facial pores are a skin care problem that people cannot avoid when caring for their skin. Therefore, it is necessary to test the effect of skin care for large pores.
[0003] Detecting pores directly without image enhancement can result in some less noticeable enlarged pores going undetected. Enhancing the entire image can lead to the detection of many interfering areas when finally detecting enlarged pores. Therefore, image enhancement should be applied only to the less easily detected areas that are targeted at the enlarged pore feature information. Summary of the Invention
[0004] To address the aforementioned problems, this invention provides an intelligent detection method for skin care effects based on image features.
[0005] The intelligent detection method for skin care effects based on image features of the present invention adopts the following technical solution:
[0006] One embodiment of the present invention provides an intelligent detection method for skin care effects based on image features, the method comprising the following steps:
[0007] Acquire RGB images of facial skin, preprocess the RGB images of facial skin to obtain facial skin images;
[0008] The probability of large pores for each pixel is obtained based on the grayscale values of all pixels within a local area of each pixel in the facial skin image; pixels suspected of having large pores are obtained based on the probability of large pores for each pixel.
[0009] The process involves classifying suspected large pore pixels into multiple categories; dividing the facial skin image into multiple image blocks of the same size; obtaining the enhancement weight for each suspected large pore pixel based on its large pore probability, the number of suspected large pore pixels in each category, and the number of suspected large pore pixels in each image block; obtaining the enhancement window size for each suspected large pore pixel based on its enhancement weight; and obtaining the enhancement window for each category based on the enhancement window sizes of all suspected large pore pixels in each category.
[0010] The facial skin image is enhanced according to the enhancement window of each category to obtain the enhanced facial skin image; the skin index corresponding to the facial skin image is obtained from the enhanced facial skin image; and the skin care effect is obtained from the skin index corresponding to the facial skin images before and after the care.
[0011] Preferably, the specific steps for obtaining the probability of coarse pores for each pixel based on the grayscale values of all pixels within a local region of each pixel in the facial skin image are as follows:
[0012]
[0013] Where, p i f represents the probability of coarse pores at the i-th pixel in a facial skin image; i F represents the grayscale value of the i-th pixel in the facial skin image. max is the maximum grayscale value in the facial skin image; n×n is the size of the local region of each pixel in the facial skin image; f i,j is the gray value of the j-th pixel (excluding the center pixel) within the local region of the i-th pixel in the facial skin image; exp() is an exponential function with the natural constant as its base.
[0014] Preferably, the specific steps for obtaining suspected large pore pixels based on the probability of large pores in each pixel are as follows:
[0015] The average probability of coarse pores in all pixels of the facial skin image is obtained as the global coarse pore probability threshold. Pixels with a coarse pore probability greater than the global coarse pore probability threshold are identified as suspected coarse pore pixels.
[0016] Preferably, the specific steps for classifying suspected large pore pixels into multiple categories are as follows:
[0017] Connectivity analysis was performed on all suspected coarse pore pixels in the facial skin image to obtain multiple connected components. All suspected coarse pore pixels in each connected component were classified as a category.
[0018] Preferably, the specific steps for obtaining the enhancement weight of each suspected large pore pixel are as follows:
[0019] The ratio of the number of all suspected large pore pixels in each image block to the total number of pixels in the image block is used as the percentage of suspected large pore pixels in each image block.
[0020] The enhancement weight for each suspected large pore pixel is obtained based on the probability of a large pore in each suspected large pore pixel, the number of all suspected large pore pixels in the category to which each suspected large pore pixel belongs, and the proportion of suspected large pore pixels in the image patch to which each suspected large pore pixel belongs.
[0021]
[0022] Among them, W k p' represents the enhancement weight of the k-th suspected large pore pixel; k b represents the probability of a large pore in the k-th suspected large pore pixel; k This represents the number of all suspected large pore pixels contained in the category to which the k-th suspected large pore pixel belongs; b k This represents the percentage of suspected coarse pore pixels in the image block to which the k-th suspected coarse pore pixel belongs; exp() is an exponential function with the natural constant as its base.
[0023] Preferably, the specific steps for obtaining the enhancement window size for each suspected large pore pixel based on the enhancement weight are as follows:
[0024] The enhancement window parameters for each suspected large pore pixel are obtained based on the enhancement weight of each pixel:
[0025] s k =1+[α×W k ]
[0026] Among them, s k W represents the enhancement window parameters for the k-th suspected coarse pore pixel. k α represents the enhancement weight of the kth suspected large pore pixel; α is a hyperparameter; [ ] is the rounding symbol;
[0027] (2s) k +1)×(2s k +1) is the enhancement window size for the kth suspected coarse pore pixel; obtain the enhancement window size for each suspected coarse pore pixel.
[0028] Preferably, the specific steps for obtaining the enhancement window for each category based on the enhancement window size of all suspected large pore pixels in each category are as follows:
[0029] For any category, obtain the mean of the enhancement window size of all suspected coarse pore pixels in that category, round the mean to the nearest integer, and denote the result as the expansion size, denoted by S. Obtain the minimum bounding rectangle of that category, denote the size of the minimum bounding rectangle as u×v, obtain the centroid of the minimum bounding rectangle, and construct a window of size (u+2S)×(v+2S) with the centroid of the minimum bounding rectangle as the center, as the enhancement window of that category.
[0030] Preferably, the specific steps for enhancing the facial skin image according to the enhancement window of each category to obtain the enhanced facial skin image are as follows:
[0031] Histogram equalization is performed on all pixels within the enhancement window of each category to obtain the enhanced image of the enhancement window for each category; the enhancement windows of each category in the facial skin image are replaced with the enhanced images of the enhancement windows of all categories to obtain the facial skin enhancement image.
[0032] Preferably, the specific steps of obtaining skin indicators corresponding to the facial skin image based on the enhanced facial skin image, and obtaining the skin care effect based on the skin indicators corresponding to the facial skin images before and after the care, are as follows:
[0033] Edge detection is performed on the enhanced facial skin image to obtain edge pixels in the enhanced facial skin image. Edge pixels in the facial features area of the enhanced facial skin image are removed, and the remaining edge pixels are used as coarse pore edge pixels.
[0034] The ratio of the number of pixels at the edge of large pores to the total number of pixels in the facial skin image is used as the skin index corresponding to the facial skin image. The skin care effect is obtained by subtracting the skin index corresponding to the facial skin image before treatment from the skin index corresponding to the facial skin image after treatment.
[0035] Preferably, the specific steps for preprocessing the RGB image of the facial skin to obtain the facial skin image are as follows:
[0036] The RGB image of the facial skin is denoised, the denoised image is then converted to grayscale, and the resulting grayscale image is used as the facial skin image.
[0037] The beneficial effects of the technical solution of this invention are as follows: This invention acquires facial skin images, obtains the probability of coarse pores for each pixel based on its grayscale value, and then identifies suspected coarse pore pixels. These suspected coarse pore pixels are categorized into multiple classes. Based on the coarse pore probability of each suspected coarse pore pixel, each class, and the number of suspected coarse pore pixels in each image block, an enhancement weight is obtained for each suspected coarse pore pixel. The enhancement window size is adjusted according to the enhancement weight, ensuring that pixels more likely to be coarse pores have larger enhancement windows, resulting in greater contrast enhancement for these pixels. For interference areas, the enhancement window is smaller, preventing interference areas from affecting the detection of skin care effects. This invention specifically enhances areas with coarse pores, making them clearer and more obvious, which can better help caregivers observe the skin care effect more meticulously, making the detection of skin care effects more accurate. Attached Figure Description
[0038] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0039] Figure 1 This is a flowchart illustrating the steps of the intelligent detection method for skin care effects based on image features according to the present invention.
[0040] Figure 2 Image of facial skin;
[0041] Figure 3 This is an edge image. Detailed Implementation
[0042] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the intelligent detection method for skin care effects based on image features proposed in this invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.
[0043] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.
[0044] The specific scheme of the intelligent detection method for skin care effect based on image features provided by the present invention will be described in detail below with reference to the accompanying drawings.
[0045] Please see Figure 1 The diagram illustrates a flowchart of a method for intelligent detection of skin care effects based on image features, according to an embodiment of the present invention. The method includes the following steps:
[0046] S001. Acquire facial skin images.
[0047] An RGB image of facial skin is captured using a camera. To avoid the impact of noise on subsequent facial pore recognition, the RGB image of the facial skin is denoised. The denoised image is then converted to grayscale, and the resulting grayscale image is used as the facial skin image. See one example of a facial skin image from this embodiment of the invention. Figure 2 .
[0048] At this point, the facial skin image has been obtained.
[0049] S002. Obtain the probability of coarse pores for each pixel in the facial skin image, and obtain the pixels suspected of having coarse pores based on the coarse pore probability.
[0050] It should be noted that when capturing facial skin images using a camera, the angle during the capture process and the fact that some pores are not clearly visible against their surroundings can cause some large pores to go undetected during the large pore detection process. For example, regarding... Figure 2 The edge image obtained by edge detection on the facial skin map is shown below. Figure 3 , Figure 3 Only relatively obvious areas of large pores were detected, but less obvious large pores were not detected. Therefore, this embodiment of the invention specifically enhances pixels that may be large pores to ensure that all large pores on the face can be detected subsequently, and then the skin care effect can be detected by comparing the distribution of large pores before and after the treatment.
[0051] It should be further explained that some large pores have relatively low contrast compared to others and are easily overlooked during detection. Therefore, it is necessary to enhance large pores by targeting some external feature information. Large pore areas have more variations in grayscale values than normal areas, possessing richer texture features, and relatively lower grayscale values. Therefore, the large pore area can be quantified by the difference between a pixel and its surrounding pixels, as well as the pixel's own grayscale value, to obtain the probability of a large pore for each pixel.
[0052] In this embodiment of the invention, an n×n window is constructed centered on each pixel in the facial skin image, serving as a local region for each pixel in the facial skin image. This embodiment uses n=5 as an example, but the specific value is not limited. In other embodiments, the implementer can set the size of n according to the actual implementation situation.
[0053] Obtain the probability of large pores for each pixel in a facial skin image:
[0054]
[0055] Where, p i f represents the probability of coarse pores at the i-th pixel in a facial skin image; i F represents the grayscale value of the i-th pixel in the facial skin image. max is the maximum grayscale value in the facial skin image; n×n is the size of the local region of each pixel in the facial skin image; f i,j exp() is the grayscale value of the j-th pixel (excluding the center pixel) within the local region of the i-th pixel in the facial skin image; exp() is an exponential function with the natural constant as its base. This represents the degree of difference between the grayscale values of the i-th pixel in a facial skin image and its surrounding pixels. The larger the value, the more variations in grayscale values occur within the local region of the i-th pixel, indicating richer texture features. The i-th pixel more closely resembles the characteristics of a region with large pores. Adding 0.0001 is to prevent the denominator from being 0; when the gray value f of the i-th pixel... i The smaller the value, the greater the difference between the gray values of the i-th pixel and its surrounding pixels, the greater the probability of the i-th pixel having large pores.
[0056] It should be noted that, besides having more distinctive grayscale variations than other pore areas, large pores also exhibit more pronounced distribution characteristics, generally being concentrated in a relatively uniform area. Judging large pores solely based on their probability may not be accurate enough. Therefore, it's necessary to use the probability of large pores to locate pixels that might indicate large pores, allowing for subsequent enhancement of the facial skin image based on the distribution of these suspected large pore pixels.
[0057] In this embodiment of the invention, the average value of the probability of coarse pores of all pixels in the facial skin image is obtained as the global coarse pore probability threshold, and pixels with a coarse pore probability greater than the global coarse pore probability threshold are identified as suspected coarse pore pixels.
[0058] At this point, all pixels suspected of having large pores in the facial skin image have been obtained.
[0059] S003. Based on the distribution of suspected pore pixels in the facial skin image, obtain multiple categories, and obtain an enhancement window for each category based on the suspected pore probability of each suspected pore pixel.
[0060] It should be noted that the reason why large pores are more noticeable than normal fine pores is because large pores occupy a larger area than fine pores. This is reflected in facial skin images as a greater number of clustered low-grayscale pixels. Multiple large pore pixels are concentrated to form a large pore, and the distribution of large pores is usually more dense in the large pore area. Therefore, based on the distribution characteristics of suspected large pore pixels and the probability of suspected large pore pixels, the enhancement weight of each suspected large pore pixel can be obtained. Enhancement windows of different sizes are generated based on the enhancement weights, and the large pore area is enhanced according to the enhancement windows. The distribution of large pores in the enhanced image is used to evaluate the skin care effect.
[0061] In this embodiment of the invention, connected component analysis is performed on all suspected coarse pore pixels in the facial skin image to obtain multiple connected components, and all suspected coarse pore pixels in each connected component are classified as a category.
[0062] It should be noted that when enhancing a certain area, it is not only necessary to consider the probability that the suspected large pore pixel is indeed a large pore pixel, but also to analyze the relationship between the suspected large pore pixel and other suspected large pore pixels in the same category, as well as whether there are other categories of suspected large pore pixels in the area where the suspected large pore pixel is located. Therefore, this embodiment of the invention obtains the enhancement weight of the suspected large pore pixel based on the number of suspected large pore pixels in the category to which the suspected large pore pixel belongs, and the total number of suspected large pore pixels in the area where the suspected large pore pixel is located. This weight is then used to obtain the enhancement window for the suspected large pore pixel, and the pixels within the enhancement window are then enhanced.
[0063] In this embodiment of the invention, the number of all suspected large pore pixels contained in each category is obtained. It should be noted that the more suspected large pore pixels a category contains, the more likely that category is indeed a large pore area.
[0064] The preset number of blocks L is used. This embodiment of the invention uses L=9 as an example for description, but is not specifically limited. In other embodiments, the implementer can set the number of blocks L according to the actual implementation situation. The size of the facial skin image is denoted as M×N, and the facial skin image is divided into L blocks. Image patches of size, in The rounding operator is used to round up. The number of pixels suspected of having large pores in each image block is counted, and the ratio of this ratio to the total number of pixels in the image block is used as the percentage of pixels suspected of having large pores in each image block.
[0065] The enhancement weight for each suspected large pore pixel is obtained based on the probability of a large pore in each suspected large pore pixel, the number of all suspected large pore pixels in the category to which each suspected large pore pixel belongs, and the proportion of suspected large pore pixels in the image patch to which each suspected large pore pixel belongs.
[0066]
[0067] Among them, W k p' represents the enhancement weight of the k-th suspected large pore pixel; k b represents the probability of a large pore in the k-th suspected large pore pixel; k This represents the number of all suspected large pore pixels contained in the category to which the k-th suspected large pore pixel belongs; b k This represents the proportion of suspected coarse pore pixels in the image block to which the k-th suspected coarse pore pixel belongs; exp() is an exponential function with the natural constant as the base; when the probability of coarse pores p' k The larger the value, the closer the feature information of the kth suspected coarse pore pixel is to the feature of the coarse pore pixel. At this time, the enhancement weight is greater. By obtaining the location information of the kth suspected coarse pore pixel, the number of all suspected coarse pore pixels contained in the category to which the kth suspected coarse pore pixel belongs is obtained. The more all suspected coarse pore pixels contained in the category, the more suspected coarse pore pixels are around the kth suspected coarse pore pixel, and the more they match the feature of coarse pores. At this time, the enhancement weight of the kth suspected coarse pore pixel is increased. When the proportion of suspected coarse pore pixels in the image block to which the kth suspected coarse pore pixel belongs is larger, the more likely the image block to which the kth suspected coarse pore pixel belongs is a dense pore area. At this time, the enhancement weight of the kth suspected coarse pore pixel is greater.
[0068] The enhancement window parameters for each suspected large pore pixel are obtained based on the enhancement weight of each pixel:
[0069] s k =1+[α×W k ]
[0070] Among them, s k W represents the enhancement window parameters for the k-th suspected coarse pore pixel. kα represents the enhancement weight of the kth suspected coarse pore pixel; α is a hyperparameter. In this embodiment of the invention, α=2 is used as an example for description. The specific value is not limited. Implementers can set the hyperparameter according to the actual implementation situation; [ ] is the rounding symbol; when the enhancement weight of the kth suspected coarse pore pixel is higher, the enhancement window size corresponding to the kth suspected coarse pore pixel should be increased to include more pixels. The contrast is increased by histogram equalization. When the enhancement weight of the kth suspected coarse pore pixel is smaller, it is more likely to be an interference area. Therefore, the enhancement window size should be reduced to prevent over-enhancement and generation of interference points.
[0071] Centered on the kth suspected coarse pore pixel, construct (2s k +1)×(2s k A window of size +1) is used as the enhancement window for the kth suspected large pore pixel. At this time, the enhancement window size for the kth suspected large pore pixel is (2s). k +1)×(2s k +1). Similarly, obtain the enhanced window size for each suspected large pore pixel.
[0072] It should be noted that since suspected large pore pixels of the same category are clustered together, when obtaining the enhancement window of suspected large pore pixels, the generated enhancement window will inevitably have multiple overlapping enhancement windows. However, suspected large pore pixels of the same category should be treated as a single region for enhancement. Therefore, in this embodiment of the invention, the enhancement window of each category is obtained based on the enhancement window size of all suspected large pore pixels in each category.
[0073] In this embodiment of the invention, for any category, the mean value of the enhancement window size of all suspected coarse pore pixels in that category is obtained. The mean value is rounded to the nearest integer, and the result is recorded as the expansion size, denoted by S. The minimum bounding rectangle of that category is obtained, and its size is denoted as u×v. The centroid of the minimum bounding rectangle is obtained. A window of size (u+2S)×(v+2S) is constructed with the centroid of the minimum bounding rectangle as the center, which serves as the enhancement window for that category.
[0074] Similarly, obtain the enhanced window for each category.
[0075] S004. Enhance the facial skin image according to the enhancement window of each category to obtain the enhanced facial skin image.
[0076] Histogram equalization is performed on all pixels within the enhancement window of each category to obtain the enhanced image for each category's enhancement window. The enhanced images of all categories' enhancement windows are then used to replace the enhancement windows of each category in the facial skin image to obtain the facial skin enhanced image. It should be noted that when enhancement windows of different categories overlap, the average grayscale value of the overlapping pixels in the enhanced images of the different categories' enhancement windows is used as the enhanced grayscale value of the overlapping pixels. During the acquisition of the facial skin enhanced image, the enhanced grayscale value of the overlapping pixels replaces the original grayscale value.
[0077] At this point, the enhanced facial skin image has been obtained.
[0078] S005. Obtain skin care effects based on enhanced facial skin images.
[0079] Edge detection is performed on the enhanced facial skin image to obtain edge pixels. Machine learning is then used to obtain edge pixels of the facial features region. These edge pixels are then removed, leaving the remaining edge pixels as coarse pore edge pixels. It should be noted that using machine learning to obtain edge pixels of the facial features region in the enhanced facial skin image is a known technique and will not be described in detail in this embodiment.
[0080] The ratio of the number of pixels at the edge of large pores to the total number of pixels in the facial skin image is used as the skin index corresponding to the facial skin image. Skin indices corresponding to facial skin images at different treatment times are obtained. The skin index corresponding to the facial skin image before treatment is subtracted from the skin index corresponding to the facial skin image after treatment. The larger the result, the better the skin treatment effect.
[0081] It should be noted that by enhancing facial skin images based on suspected large pore pixels, the areas with large pores become clearer and more obvious, which can better help caregivers observe the skin care effects more closely.
[0082] The above steps complete the intelligent detection of skin care effects.
[0083] This invention acquires facial skin images, calculates the probability of coarse pores for each pixel based on its grayscale value, and then identifies suspected coarse pore pixels. These suspected coarse pore pixels are categorized into multiple classes. An enhancement weight is calculated for each suspected coarse pore pixel based on its coarse pore probability, the class of pixels, and the number of suspected coarse pore pixels in each image block. The enhancement window size is adjusted according to this weight, ensuring that pixels more likely to have coarse pores have larger enhancement windows, resulting in greater contrast enhancement for these pixels. Conversely, smaller enhancement windows are used for interference areas to prevent interference from affecting the detection of skincare effects. This invention specifically enhances areas with coarse pores, making them clearer and more visible, allowing skincare professionals to observe skincare effects more closely and making the detection of skincare effects more accurate.
[0084] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A method for intelligent detection of skin care effects based on image features, characterized in that, The method includes the following steps: Acquire RGB images of facial skin, preprocess the RGB images of facial skin to obtain facial skin images; The probability of large pores for each pixel is obtained based on the grayscale values of all pixels within a local area of each pixel in the facial skin image; pixels suspected of having large pores are obtained based on the probability of large pores for each pixel. The process involves classifying suspected large pore pixels into multiple categories; dividing the facial skin image into multiple image blocks of the same size; obtaining the enhancement weight for each suspected large pore pixel based on its large pore probability, the number of suspected large pore pixels in each category, and the number of suspected large pore pixels in each image block; obtaining the enhancement window size for each suspected large pore pixel based on its enhancement weight; and obtaining the enhancement window for each category based on the enhancement window sizes of all suspected large pore pixels in each category. The facial skin image is enhanced according to the enhancement window of each category to obtain the enhanced facial skin image; the skin index corresponding to the facial skin image is obtained from the enhanced facial skin image; and the skin care effect is obtained from the skin index corresponding to the facial skin images before and after the care.
2. The intelligent detection method for skin care effect based on image features according to claim 1, characterized in that, The specific steps for obtaining the probability of coarse pores for each pixel based on the grayscale values of all pixels within a local region of each pixel in a facial skin image are as follows: Where, p i f represents the probability of coarse pores at the i-th pixel in a facial skin image; i F represents the grayscale value of the i-th pixel in the facial skin image. max is the maximum grayscale value in the facial skin image; n×n is the size of the local region of each pixel in the facial skin image; f i,j is the gray value of the j-th pixel (excluding the center pixel) within the local region of the i-th pixel in the facial skin image; exp() is an exponential function with the natural constant as its base.
3. The intelligent detection method for skin care effect based on image features according to claim 1, characterized in that, The specific steps for obtaining suspected large pore pixels based on the probability of large pores in each pixel are as follows: The average probability of coarse pores in all pixels of the facial skin image is obtained as the global coarse pore probability threshold. Pixels with a coarse pore probability greater than the global coarse pore probability threshold are identified as suspected coarse pore pixels.
4. The intelligent detection method for skin care effect based on image features according to claim 1, characterized in that, The specific steps involved in classifying suspected large pore pixels into multiple categories are as follows: Connectivity analysis was performed on all suspected coarse pore pixels in the facial skin image to obtain multiple connected components. All suspected coarse pore pixels in each connected component were classified as a category.
5. The intelligent detection method for skin care effect based on image features according to claim 1, characterized in that, The specific steps for obtaining the enhancement weight of each suspected large pore pixel are as follows: The ratio of the number of all suspected large pore pixels in each image block to the total number of pixels in the image block is used as the percentage of suspected large pore pixels in each image block. The enhancement weight for each suspected large pore pixel is obtained based on the probability of a large pore in each suspected large pore pixel, the number of all suspected large pore pixels in the category to which each suspected large pore pixel belongs, and the proportion of suspected large pore pixels in the image patch to which each suspected large pore pixel belongs. Among them, W k p' represents the enhancement weight of the k-th suspected large pore pixel; k h represents the probability of a large pore in the k-th suspected large pore pixel; k This represents the number of all suspected large pore pixels contained in the category to which the k-th suspected large pore pixel belongs; b k This represents the percentage of suspected coarse pore pixels in the image block to which the k-th suspected coarse pore pixel belongs; exp() is an exponential function with the natural constant as its base.
6. The intelligent detection method for skin care effect based on image features according to claim 1, characterized in that, The specific steps for obtaining the enhancement window size for each suspected large pore pixel based on its enhancement weight are as follows: The enhancement window parameters for each suspected large pore pixel are obtained based on the enhancement weight of each pixel: s k =1+[α×W k ] Among them, s k W represents the enhancement window parameters for the k-th suspected coarse pore pixel. k This represents the enhancement weight of the kth suspected large pore pixel; α is a hyperparameter; [] is the rounding symbol; (2s) k +1)×(2s k +1) is the enhancement window size for the kth suspected coarse pore pixel; obtain the enhancement window size for each suspected coarse pore pixel.
7. The intelligent detection method for skin care effect based on image features according to claim 1, characterized in that, The specific steps for obtaining the enhancement window for each category based on the enhancement window size of all suspected large pore pixels in each category are as follows: For any category, obtain the mean of the enhancement window size of all suspected coarse pore pixels in that category, round the mean to the nearest integer, and denote the result as the expansion size, denoted by S. Obtain the minimum bounding rectangle of that category, denote the size of the minimum bounding rectangle as u×v, obtain the centroid of the minimum bounding rectangle, and construct a window of size (u+2S)×(v+2S) with the centroid of the minimum bounding rectangle as the center, as the enhancement window of that category.
8. The intelligent detection method for skin care effect based on image features according to claim 1, characterized in that, The specific steps involved in enhancing the facial skin image according to the enhancement window for each category to obtain an enhanced facial skin image are as follows: Histogram equalization is performed on all pixels within the enhancement window of each category to obtain the enhanced image of the enhancement window for each category; the enhancement windows of each category in the facial skin image are replaced with the enhanced images of the enhancement windows of all categories to obtain the facial skin enhancement image.
9. The intelligent detection method for skin care effect based on image features according to claim 1, characterized in that, The specific steps for obtaining skin indicators corresponding to the facial skin image based on the enhanced facial skin image, and obtaining the skin care effect based on the skin indicators corresponding to the facial skin images before and after the care, are as follows: Edge detection is performed on the enhanced facial skin image to obtain edge pixels in the enhanced facial skin image. Edge pixels in the facial features area of the enhanced facial skin image are removed, and the remaining edge pixels are used as coarse pore edge pixels. The ratio of the number of pixels at the edge of large pores to the total number of pixels in the facial skin image is used as the skin index corresponding to the facial skin image. The skin care effect is obtained by subtracting the skin index corresponding to the facial skin image before treatment from the skin index corresponding to the facial skin image after treatment.
10. The intelligent detection method for skin care effect based on image features according to claim 1, characterized in that, The specific steps involved in preprocessing the RGB image of the facial skin to obtain the facial skin image are as follows: The RGB image of the facial skin is denoised, the denoised image is then converted to grayscale, and the resulting grayscale image is used as the facial skin image.