An intelligent skin management system based on image analysis

Abstract

The present invention relates to the field of image processing, and in particular to an intelligent skin management system based on image analysis, comprising: a data analysis unit, used to determine a pixel value state according to a pixel value contrast and a pixel value concentration, and to determine an image matching method as image parameter direct matching or image cutting indirect matching according to the pixel value state; an image processing unit, used to determine a distribution state of a feature area of ​​a target image according to a discreteness of the feature area of ​​the target image and a regularity of the feature area, and to determine a specific matching method of image cutting indirect matching according to the distribution state of the feature area of ​​the target image; an image matching unit, used to execute instruction information of a data analysis unit or an image processing unit; a pre-storage unit, used to store a plurality of pre-stored images, each pre-stored graphic being provided with a corresponding symptom label; the present invention adjusts an image segmentation method according to an actual image state and an image matching result, thereby improving the recognition accuracy of a skin image.

Description

Technical Field

[0001] The present invention relates to the field of image processing, and in particular to an intelligent skin management system based on image analysis. Background Art

[0002] At present, breakthroughs have been made in information and digital technologies. Neural network recognition technology, featuring image recognition, has shined in many fields. Skin diseases are also one of the application fields of image recognition. In the recognition of skin disease images, the segmentation of skin lesion areas is an extremely critical link. Due to the irregularities, inconsistent colors and unclear edges of the skin lesion areas, the image recognition accuracy is often poor, which in turn leads to low image recognition accuracy. Therefore, how to improve the segmentation and matching efficiency of the skin lesion areas to accurately identify and match the corresponding skin diseases is a problem that needs to be urgently solved by technical personnel in this field.

[0003] Chinese patent publication number CN114693703A discloses a dermoscopic image segmentation model training, dermoscopic image recognition method and device, the method comprising: receiving a dermoscopic image to be recognized; the dermoscopic image to be recognized comprises: a lesion of a skin disease; in response to a dermoscopic image segmentation request triggered by a user, inputting the dermoscopic image to be recognized into a dermoscopic image segmentation model, and determining the lesion area in the dermoscopic image to be recognized; the dermoscopic image segmentation model is obtained by training a first preset model; the first preset model comprises a hole convolution layer and a self-attention module; according to the lesion area in the dermoscopic image to be recognized, generating a segmented image corresponding to the dermoscopic image to be recognized; and outputting a segmented image corresponding to the dermoscopic image to be recognized. The above technical solution has the following problems: the segmentation of the dermoscopic image is performed using a single dermoscopic image segmentation model, and if there is an error in the segmentation result, the accuracy of subsequent image recognition will be affected. Summary of the invention

[0004] To this end, the present invention provides an intelligent skin management system based on image analysis, which adjusts the image segmentation method according to the actual image state and image matching results, thereby overcoming the problem in the prior art that the segmentation of dermoscopic images uses a single dermoscopic image segmentation model to perform segmentation, resulting in poor image recognition accuracy.

[0005] To achieve the above object, the present invention provides an intelligent skin management system based on image analysis, comprising:

[0006] A data analysis unit connected to the data upload end is used to determine the pixel value state according to the pixel value contrast and pixel value concentration of the target image uploaded by the user, and to determine the image matching mode as image parameter direct matching or image cutting indirect matching according to the pixel value state;

[0007] An image processing unit connected to the data analysis unit is used to determine the distribution state of the characteristic regions of the target image according to the discreteness of the characteristic regions of the target image and the regularity of the characteristic regions, and to determine the specific matching mode of the image cutting indirect matching according to the distribution state of the characteristic regions of the target image;

[0008] An image matching unit connected to the data analysis unit and the image processing unit to execute instruction information of the data analysis unit or the image processing unit, wherein different specific matching modes under the image cutting indirect matching correspond to different matching failure conditions;

[0009] A pre-storage unit connected to the image matching unit and used to store a plurality of pre-stored images, each of which is correspondingly provided with a disease label;

[0010] The specific matching method of image cutting indirect matching is to perform cutting matching on the target image according to the edge search method or to perform cutting matching on the target image according to the point-level search method;

[0011] In the case of a matching failure, the image matching unit adjusts the diameter of the minimum matching circle of the target image, or adjusts the matching area of ​​the matching region of the target image.

[0012] Furthermore, the data analysis unit detects the pixel value of each image point in the target image and determines the image matching method according to the pixel value state, including direct matching of image parameters and indirect matching of image cutting.

[0013] Further, the data analysis unit determines the pixel value state according to the pixel value contrast and the pixel value concentration in the target image, including:

[0014] In a first pixel value state, a pixel value contrast in the target image is greater than a preset pixel value contrast and a pixel value concentration is greater than a preset pixel value concentration;

[0015] In the second pixel value state, the pixel value contrast in the target image is less than or equal to the preset pixel value contrast or the pixel value concentration is less than or equal to the preset pixel value concentration.

[0016] Furthermore, in the direct matching of image parameters, the image matching unit calculates the similarity between the target image and each pre-stored image respectively, and records the corresponding pre-stored image with the greatest similarity as a successfully matched image.

[0017] Furthermore, the image processing unit determines a specific matching method of image cutting indirect matching according to the distribution state of the characteristic area of ​​the target image, including cutting matching for the target image according to an edge search method or cutting matching for the target image according to a point-level search method.

[0018] Furthermore, the image processing unit determines the distribution state of the feature area of ​​the target image according to the discreteness of the feature area of ​​the target image and the regularity of the feature area, including:

[0019] In a first distribution state, the discreteness of the characteristic region of the target image is less than a preset discreteness and the regularity of the characteristic region of the target image is greater than a preset regularity;

[0020] In the second distribution state, the discreteness of the characteristic region of the target image is greater than or equal to a preset discreteness or the regularity of the characteristic region of the target image is less than or equal to a preset regularity.

[0021] Further, the image matching unit performs cutting matching on the target image according to the edge search method, wherein the image matching unit performs reference difference value matching on the feature area of ​​the target image and the feature area of ​​each pre-stored image;

[0022] In the case of a matching failure, the diameter of the minimum matching circle of the target image is reduced.

[0023] Further, the image matching unit performs cutting matching on the target image according to the point-level search method, wherein the image matching unit records the sub-quadrant with the largest area of ​​the corresponding characteristic sub-region in the target image as the matching region, and the image matching unit performs reference difference value matching between the matching region and the sub-quadrant with the largest area of ​​the characteristic sub-region corresponding to the characteristic region of each pre-stored image;

[0024] In the case of a matching failure, the matching area of ​​the matching region of the target image is increased and adjusted.

[0025] Furthermore, the image matching unit selects a sub-quadrant adjacent to the matching area to supplement the matching area in the increase adjustment of the matching area of ​​the target image;

[0026] If there are two selectable adjacent sub-quadrants in the matching region, the image matching unit adds the sub-quadrant with a larger number of corresponding feature sub-regions to the matching region.

[0027] Furthermore, the image matching unit is set with a maximum matching number, and when the number of matching failures is equal to the maximum matching number, the image matching unit determines that the image matching has failed.

[0028] Compared with the prior art, the beneficial effect of the present invention lies in that, in the technical scheme of the present invention, the data analysis unit detects the pixel value of each image point in the target image and determines the image matching method according to the pixel value state, wherein the pixel value state reflects the contrast degree between the pixel points in the target image and whether the pixel values ​​are clustered, and then different image matching methods are selected for the target image, so that the image matching method is more in line with the actual application scenario, avoiding the poor image recognition targeting and poor practicality caused by a single image recognition matching mode, and further improving the skin image recognition matching efficiency of the present invention.

[0029] Furthermore, the image processing unit in the present invention determines the specific matching method of image cutting indirect matching according to the distribution state of the characteristic area of ​​the target image, and characterizes the characteristics of the skin lesion area of ​​the target image by the discrete degree and regularity of the characteristic area of ​​the target image, so that the selection of the specific matching method of image cutting indirect matching is more in line with the specific actual situation of the image, making the image processing of the present invention more targeted, and further improving the skin image recognition and matching efficiency of the present invention.

[0030] Furthermore, in the present invention, under the condition of matching failure, the image matching unit adjusts the diameter of the minimum matching circle of the target image, or adjusts the matching area of ​​the matching region of the target image. The image matching unit makes corresponding adjustments to the matching process according to the matching results, so that the adjustment of the image recognition matching process is more in line with the actual working process, thereby improving the skin image recognition matching efficiency in the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 This is a unit connection diagram of the intelligent skin management system based on image analysis of the present invention;

[0032] Figure 2 A flow chart of the present invention for determining an image matching method according to a pixel value state;

[0033] Figure 3 The present invention is a flow chart of determining a specific matching method for indirect matching of image cutting according to the distribution state of the characteristic regions of the target image. DETAILED DESCRIPTION

[0034] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0035] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.

[0036] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside" and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the drawings. This is merely for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present invention.

[0037] In addition, it should be noted that in the description of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0038] See also Figures 1 to 3 As shown, the present invention provides an intelligent skin management system based on image analysis, comprising:

[0039] A data analysis unit connected to the data upload end is used to determine the pixel value state according to the pixel value contrast and pixel value concentration of the target image uploaded by the user, and to determine the image matching mode as image parameter direct matching or image cutting indirect matching according to the pixel value state;

[0040] An image processing unit connected to the data analysis unit is used to determine the distribution state of the characteristic regions of the target image according to the discreteness of the characteristic regions of the target image and the regularity of the characteristic regions, and to determine the specific matching mode of the image cutting indirect matching according to the distribution state of the characteristic regions of the target image;

[0041] An image matching unit connected to the data analysis unit and the image processing unit to execute instruction information of the data analysis unit or the image processing unit, wherein different specific matching modes under the image cutting indirect matching correspond to different matching failure conditions;

[0042] A pre-storage unit connected to the image matching unit and used to store a plurality of pre-stored images, each of which is correspondingly provided with a disease label;

[0043] The specific matching method of image cutting indirect matching is to perform cutting matching on the target image according to the edge search method or to perform cutting matching on the target image according to the point-level search method;

[0044] In the case of a matching failure, the image matching unit adjusts the diameter of the minimum matching circle of the target image, or adjusts the matching area of ​​the matching region of the target image.

[0045] In the present invention, the target image uploaded by the user through the data upload end is a dermoscopic image of the user's skin lesion area. The dermoscopic image refers to an image obtained by observing the skin lesion area using a dermatoscope, which can reveal the color and structure of the skin lesions that are not observable to the naked eye. The dermoscopic image and the acquisition method thereof are common knowledge already mastered by those skilled in the art and will not be elaborated here.

[0046] Specifically, the data analysis unit detects the pixel value of each image point in the target image and determines the image matching method according to the pixel value state, including direct matching of image parameters and indirect matching of image cutting.

[0047] Specifically, the data analysis unit determines the pixel value state according to the pixel value contrast and the pixel value concentration in the target image, including:

[0048] In a first pixel value state, a pixel value contrast in the target image is greater than a preset pixel value contrast and a pixel value concentration is greater than a preset pixel value concentration;

[0049] In the second pixel value state, the pixel value contrast in the target image is less than or equal to the preset pixel value contrast or the pixel value concentration is less than or equal to the preset pixel value concentration.

[0050] If the pixel value state is in the first pixel value state, the data analysis unit determines to adopt image parameter direct matching;

[0051] If the pixel value state is in the second pixel value state, the data analysis unit determines to adopt image cutting indirect matching.

[0052] The data analysis unit performs preprocessing on the target image, including converting the target image from a color image to a grayscale image, and extracting the pixel value of each pixel point, the pixel value reflects the brightness of the corresponding pixel point, the image point is the pixel point, and the pixel value concentration degree is confirmed by calculating the average value P0 of the sum of the pixel values ​​of each image point, the pixel value concentration degree is P, and setting,

[0053]

[0054] Wherein, Pi is the pixel value of the i-th pixel point, i=1,2,3,...,n, n is the total number of pixels, P reflects the degree of difference in pixel values ​​between pixels with high pixel values ​​and pixels with low pixel values ​​in the target image, the pixel value contrast is confirmed by recording the image points whose corresponding pixel difference values ​​are greater than the preset pixel difference values ​​as valid image points, pixel value contrast=number of valid image points / n, pixel difference value corresponding to the i-th image point=|Pi-P0|, the preset pixel difference value is 30% of P0, the pixel value contrast reflects the difficulty of identifying the edge of the lesion area in the target image, the smaller the pixel value contrast, the more image points that affect the image segmentation, the preset pixel value contrast and the preset pixel value aggregation value can be set by the user according to the actual application scenario, the higher the user's demand for image segmentation accuracy, the larger the preset pixel value contrast and the preset pixel value aggregation value, and the preset pixel value contrast and the preset pixel value aggregation value are provided, the preset pixel value aggregation value is 50% of P0, and the preset pixel value contrast is 70% of n.

[0055] Specifically, in the direct matching of image parameters, the image matching unit calculates the similarity between the target image and each pre-stored image respectively, and records the corresponding pre-stored image with the greatest similarity as a successfully matched image.

[0056] Among them, two existing similarity confirmation methods are provided, one is a histogram algorithm, which is a commonly used image similarity comparison method. The histogram of each channel (such as red, green, and blue) can be calculated separately. For the histograms of two pictures, different distance or similarity measurement methods can be used for comparison, including Euclidean distance, Manhattan distance, or Bhattacharyya distance. According to these measurement results, the similarity score between the two pictures is calculated. The other is a cosine similarity algorithm, which converts the image into a feature vector, and then uses cosine similarity to compare the similarity of these feature vectors, preferably a cosine similarity algorithm; the above are all contents that are easy to understand for those skilled in the art, and will not be repeated here.

[0057] Specifically, the image processing unit determines a specific matching method of image cutting indirect matching according to the distribution state of the feature area of ​​the target image, including cutting matching for the target image according to an edge search method or cutting matching for the target image according to a point-level search method.

[0058] Specifically, the image processing unit determines the distribution state of the feature area of ​​the target image according to the discreteness of the feature area of ​​the target image and the regularity of the feature area, including:

[0059] In a first distribution state, the discreteness of the characteristic region of the target image is less than a preset discreteness and the regularity of the characteristic region of the target image is greater than a preset regularity;

[0060] In the second distribution state, the discreteness of the characteristic region of the target image is greater than or equal to a preset discreteness or the regularity of the characteristic region of the target image is less than or equal to a preset regularity.

[0061] Wherein, when the distribution state of the characteristic region of the target image is the first distribution state, the image processing unit determines to perform cutting matching on the target image according to the edge search method;

[0062] When the distribution state of the feature area of ​​the target image is the second distribution state, the image processing unit determines to perform cutting matching on the target image according to the point-level search method.

[0063] Specifically, the image processing unit in the present invention establishes a number of feature sub-regions for the target image. The method for establishing the feature sub-regions is to establish the feature sub-regions for the image points in order from large to small pixel values. For the i-th image point, the pixel values ​​of all the image points connected to the image point are detected. If the pixel values ​​of the image points connected to the i-th image point are all greater than or equal to 80% of the pixel value Pi of the image point, then it is determined that the connected image points are all recorded as feature sub-regions and continue to generate feature sub-regions. If there is at least one image point among the image points connected to the i-th image point whose pixel value is less than 80% of the pixel value Pi of the image point, then it is determined that the feature sub-region generation fails, and the image point does not correspond to an existing feature region.

[0064] When the generation of the feature sub-region is completed, a minimum circle including all the feature sub-regions is established as the feature region. The discreteness of the feature region = the area of ​​the feature region - (the sum of the areas of each feature sub-region), the regularity of the feature region = Mmax-Mmin, and a two-dimensional coordinate system is established with the center of the circle of the feature region as the midpoint. The two-dimensional coordinate system divides the feature region into four sub-quadrants. Max is the maximum number of feature sub-regions in a single sub-quadrant, and Min is the minimum number of feature sub-regions in a single sub-quadrant. The discreteness and regularity are used to reflect the distribution density and shape regularity of the lesion area respectively. The values ​​of the preset discreteness and the preset regularity can be set by the user according to the actual application scenario. The smaller the value of the preset discreteness and the larger the value of the preset regularity, the greater the user's demand for image recognition accuracy. A preset discreteness and preset regularity are provided. The preset discreteness is 70% of the area of ​​the feature region, and the value of the preset regularity is 50% of Mmax.

[0065] Specifically, the image matching unit performs cutting matching on the target image according to the edge search method, wherein the image matching unit performs reference difference value matching on the feature area of ​​the target image and the feature area of ​​each pre-stored image;

[0066] In the case of a matching failure, the diameter of the minimum matching circle of the target image is reduced.

[0067] The matching failure condition is that the reference difference values ​​between the feature area of ​​the target image and the feature areas of each pre-stored image are both less than a preset reference difference value or the number of pre-stored images whose reference difference values ​​between the feature area of ​​the target image and the pre-stored image are greater than 1;

[0068] Under the condition of successful matching, the target image is marked as the same type of disease as the successfully matched pre-stored image and sent to the doctor for final diagnosis. The condition of successful matching is that the reference difference value between the feature area of ​​the target image and the feature area of ​​the pre-stored image is greater than or equal to the preset reference difference value;

[0069] The single diameter reduction adjustment is 5% of the initial diameter.

[0070] The reference difference value is confirmed as follows: reference difference value = (|H1-H0|) / H1×α1+(|J1-J0|) / J1×α2+(|K1-K0|) / K1×α3, wherein H1 is the average pixel value of the feature area in the target image, H0 is the average pixel value of the feature area of ​​the pre-stored image for comparison, J1 is the area of ​​the feature area in the target image, J0 is the area of ​​the feature area of ​​the pre-stored image for comparison, K1 is the discreteness of the feature area in the target image, K0 is the discreteness of the feature area of ​​the pre-stored image for comparison, α1 is the first weight coefficient, α2 is the second weight coefficient, and α3 is the third weight coefficient. Users can determine the influence of the average pixel value, the area of ​​the feature area, and the discreteness on the matching accuracy through the big data model based on historical data to set the value of each weight coefficient by themselves. A value of a weight coefficient is provided: α1=0.3, α2=0.3, α3=0.3, and the preset reference difference value=0.2.

[0071] Specifically, the image matching unit performs cutting matching on the target image according to the point-level search method, wherein the image matching unit records the sub-quadrant with the largest area of ​​the corresponding characteristic sub-region in the target image as the matching region, and the image matching unit matches the matching region with the sub-quadrant with the largest area of ​​the characteristic sub-region corresponding to the characteristic region of each pre-stored image by reference difference value;

[0072] In the case of a matching failure, the matching area of ​​the matching region of the target image is increased and adjusted.

[0073] Specifically, the image matching unit selects a sub-quadrant adjacent to the matching area to supplement the matching area in the increase adjustment of the matching area of ​​the target image;

[0074] If there are two selectable adjacent sub-quadrants in the matching region, the image matching unit adds the sub-quadrant with a larger number of corresponding feature sub-regions to the matching region.

[0075] Specifically, the image matching unit is set with a maximum matching number, and when the number of matching failures is equal to the maximum matching number, the image matching unit determines that the image matching fails. In the point-level search mode, the maximum matching number is 4, and in the edge search mode, the maximum matching number is 6.

[0076] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

[0077] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An intelligent skin management system based on image analysis, characterized in that: include: A data analysis unit connected to the data upload end is used to determine the pixel value state according to the pixel value contrast and pixel value concentration of the target image uploaded by the user, and to determine the image matching mode as image parameter direct matching or image cutting indirect matching according to the pixel value state; An image processing unit connected to the data analysis unit is used to determine the distribution state of the characteristic regions of the target image according to the discreteness of the characteristic regions of the target image and the regularity of the characteristic regions, and to determine the specific matching mode of the image cutting indirect matching according to the distribution state of the characteristic regions of the target image; An image matching unit connected to the data analysis unit and the image processing unit to execute instruction information of the data analysis unit or the image processing unit, wherein different specific matching modes under the image cutting indirect matching correspond to different matching failure conditions; A pre-storage unit connected to the image matching unit and used to store a plurality of pre-stored images, each of which is correspondingly provided with a disease label; The specific matching method of image cutting indirect matching is to perform cutting matching on the target image according to the edge search method or to perform cutting matching on the target image according to the point-level search method; In the cutting matching of the target image according to the edge search method, under the matching failure condition, the image matching unit adjusts the diameter of the minimum matching circle of the target image; In the cutting and matching of the target image according to the point-level search method, under the matching failure condition, the matching area of ​​the matching area of ​​the target image is adjusted; The method for confirming the pixel value concentration is to calculate the average value P0 of the sum of the pixel values ​​of each image point, and the pixel value concentration is P. Set, Where Pi is the pixel value of the i-th pixel, i = 1, 2, 3, ..., n, and n is the total number of pixels; The discreteness of the feature area = the area of ​​the feature area - (the sum of the areas of each feature sub-area), the regularity of the feature area = Mmax-Mmin, and a two-dimensional coordinate system is established with the center of the feature area as the midpoint. The two-dimensional coordinate system divides the feature area into 4 sub-quadrants. Max is the maximum number of feature sub-areas in a single sub-quadrant, and Min is the minimum number of feature sub-areas in a single sub-quadrant.

2. The intelligent skin management system based on image analysis according to claim 1, characterized in that: The data analysis unit detects the pixel value of each image point in the target image and determines the image matching method according to the pixel value state, including direct matching of image parameters and indirect matching of image cutting.

3. The intelligent skin management system based on image analysis according to claim 2, characterized in that: The data analysis unit determines the pixel value state according to the pixel value contrast and the pixel value concentration in the target image, including: In a first pixel value state, a pixel value contrast in the target image is greater than a preset pixel value contrast and a pixel value concentration is greater than a preset pixel value concentration; In the second pixel value state, the pixel value contrast in the target image is less than or equal to the preset pixel value contrast or the pixel value concentration is less than or equal to the preset pixel value concentration.

4. The intelligent skin management system based on image analysis according to claim 3, characterized in that: In the direct matching of image parameters, the image matching unit calculates the similarity between the target image and each pre-stored image respectively, and records the corresponding pre-stored image with the greatest similarity as the successfully matched image.

5. The intelligent skin management system based on image analysis according to claim 4, characterized in that: The image processing unit determines the distribution state of the feature area of ​​the target image according to the discreteness of the feature area of ​​the target image and the regularity of the feature area, including: In a first distribution state, the discreteness of the characteristic region of the target image is less than a preset discreteness and the regularity of the characteristic region of the target image is greater than a preset regularity; In the second distribution state, the discreteness of the characteristic region of the target image is greater than or equal to a preset discreteness or the regularity of the characteristic region of the target image is less than or equal to a preset regularity.

6. The intelligent skin management system based on image analysis according to claim 5, characterized in that: The image matching unit performs cutting matching on the target image according to the edge search method, wherein the image matching unit performs reference difference value matching on the feature area of ​​the target image and the feature area of ​​each pre-stored image; In the case of a matching failure, the diameter of the minimum matching circle of the target image is reduced.

7. The intelligent skin management system based on image analysis according to claim 6, characterized in that: The image matching unit performs cutting matching on the target image according to the point-level search method, wherein the image matching unit records the sub-quadrant with the largest area of ​​the corresponding characteristic sub-region in the target image as the matching region, and the image matching unit matches the matching region with the sub-quadrant with the largest area of ​​the characteristic sub-region corresponding to the characteristic region of each pre-stored image by reference difference value; In the case of a matching failure, the matching area of ​​the matching region of the target image is increased and adjusted.

8. The intelligent skin management system based on image analysis according to claim 7, characterized in that: The image matching unit selects a sub-quadrant adjacent to the matching area to supplement the matching area in the increase adjustment of the matching area of ​​the target image; If there are two selectable adjacent sub-quadrants in the matching region, the image matching unit adds the sub-quadrant with a larger number of corresponding feature sub-regions to the matching region.

9. The intelligent skin management system based on image analysis according to claim 8, characterized in that: The image matching unit is set with a maximum matching number. When the number of matching failures is equal to the maximum matching number, the image matching unit determines that the image matching fails.

Images