Auxiliary detection system and detection method for local skin quality

Abstract

The present invention relates to the technical field of skin auxiliary detection, and specifically to an auxiliary detection system and method for local skin quality. The present invention first obtains a thermal histogram of a local skin area to be tested, and all thermal intervals of suspected defective skin therein; further, according to the distribution of pixels in the thermal map under each thermal interval of suspected defective skin, all pore areas and all acne areas in the thermal map are determined; then, according to the position distribution and area information of all pore areas, the local skin area to be tested is divided, and the oiliness coefficient of each skin partition is further evaluated in combination with the water loss rate and the distribution of pore acne in each skin partition. The present invention combines the characteristics of oily skin with vigorous oil secretion and the proneness to skin defects such as pore acne, and the characteristics that the temperature of the defective epidermis is slightly higher than the normal healthy epidermis temperature, to quantitatively analyze the oiliness coefficient of the local skin and improve the skin quality detection effect of the local skin.

Description

Technical Field

[0001] The present invention relates to the technical field of skin auxiliary detection, and specifically relates to an auxiliary detection system and method for local skin texture. Background Art

[0002] The skin texture type of each person may vary due to factors such as genes, environment, and lifestyle, and can generally be divided into oily skin, dry skin, and combination skin; skin texture detection can help relevant detection personnel analyze and evaluate the current facial skin condition, so as to formulate a more scientific and personalized skin care plan. At the same time, it can also help solve various skin problems and maintain skin health.

[0003] In the prior art, local skin images can be collected based on image processing technology, and pores in the local skin images can be automatically identified based on machine learning and deep learning models. Then, the oil distribution in the local skin area can be analyzed and evaluated through skin reflectance, pore size, and distribution to assist in skin texture evaluation; however, the training process of the pore recognition model is cumbersome, and the acquisition quality of local skin images and the training effect of the pore recognition model are not good, which will affect the accuracy of the evaluation result of the oil distribution, and further affect the skin texture detection effect of the local skin. Summary of the Invention

[0004] In order to solve the technical problem of the poor skin texture detection effect of the prior art on local skin, the purpose of the present invention is to provide an auxiliary detection system and method for local skin texture, and the specific technical solutions adopted are as follows:

[0005] An auxiliary detection method for local skin texture, the method includes:

[0006] Obtain the thermal histogram of the thermal map of the local skin area to be measured, and screen out all suspected defective skin thermal intervals from all thermal intervals of the thermal histogram;

[0007] According to the distribution and aggregation of pixel points in the thermal map under each suspected defective skin thermal interval, obtain the screening coefficient for each suspected defective skin thermal interval to be a pore thermal interval; screen out the pore thermal intervals from all suspected defective skin thermal intervals according to the screening coefficient, and obtain all pore areas in the thermal map; according to the pore thermal interval and the distribution information of pixel points in the thermal map under the remaining suspected defective skin thermal intervals, obtain all acne areas in the thermal map;

[0008] Divide the local skin area to be measured according to the position distribution and area information of all the pore regions to obtain all skin partitions; obtain the water loss rate within each skin partition; based on the water loss rate within each skin partition, combine the number of pore regions and the number of acne regions within the corresponding regions of each skin partition in the heat map to obtain the oiliness coefficient of each skin partition within the local skin area to be measured.

[0009] Further, the method for obtaining the suspected defective skin heat range includes:

[0010] In the heat histogram, take the product of the normalized result of the frequency corresponding to each heat range and the negatively correlated mapping result of the heat value corresponding to the right endpoint of each heat range as the confidence coefficient that each heat range is the normal skin temperature; screen out the normal skin heat ranges from all heat ranges according to the confidence coefficient;

[0011] Based on the heat histogram, fit a heat distribution curve, and take the heat range corresponding to each maximum point in the heat distribution curve in the heat histogram as a suspected range; in the heat histogram, take all suspected ranges located to the right of the normal skin heat range as the suspected defective skin heat range.

[0012] Further, the method for obtaining the normal skin heat range includes:

[0013] Take the heat range with the largest confidence coefficient in the heat histogram as the normal skin heat range.

[0014] Further, the method for obtaining the screening coefficient includes:

[0015] Take any suspected defective skin heat range as the target range, perform region connectivity detection on all pixel points corresponding to the target range in the heat map, and take each connected domain as a suspected pore; construct a preset neighborhood centered on each suspected pore;

[0016] Within the preset neighborhood of each suspected pore, obtain the screening sub-parameter that the target range is the pore heat range according to the spatial distribution uniformity of the suspected pores and the area similarity between different suspected pores;

[0017] Integrate the screening sub-parameters corresponding to the preset neighborhoods of all the suspected pores to obtain the screening coefficient that the target range is the pore heat range.

[0018] Further, the method for obtaining the screening sub-parameter includes:

[0019] Within the preset neighborhood of each of the suspected pores, the center of the preset neighborhood corresponding to the suspected pore is taken as the target pore, and the remaining suspected pores are taken as reference pores;

[0020] The sum of the deviations of the distances between each reference pore and the target pore from the average level is accumulated and negatively correlated mapped, and the negatively correlated mapping result is used as the pore uniform distribution parameter; the negatively correlated mapping result of the sum of the area differences between each reference pore and the target pore is used as the pore area similarity parameter;

[0021] The pore uniform distribution parameter and the pore area similarity parameter are fused to obtain a screening sub-parameter with the target interval being the pore heat range.

[0022] Furthermore, the method for obtaining the pore heat range and the pore region includes:

[0023] The heat range with the largest screening coefficient is taken as the pore heat range; region connectivity detection is performed on all pixel points corresponding to the pore heat range in the heat map, and each connected domain is taken as a pore region.

[0024] Furthermore, the method for obtaining the acne region includes:

[0025] In the heat histogram, all suspected intervals located on the right side of the pore heat range are taken as suspected acne heat ranges; under each suspected acne heat range, region connectivity detection is performed on all pixel points corresponding in the heat map, and each connected domain is taken as a suspected acne;

[0026] The suspected acne heat range corresponding to the maximum total area of the suspected acne is taken as the acne heat range; all the suspected acne corresponding to the acne heat range are taken as all acne regions in the heat map.

[0027] Furthermore, the method for obtaining the skin partition includes:

[0028] Based on the area differences and spatial distances between the pore regions, the metric distances between different pore regions are obtained; based on the distance clustering algorithm and the metric distances, all the pore regions are clustered to obtain all clusters; the corresponding region of each cluster in the heat map is taken as a partition region, and the corresponding region of each partition region in the heat map in the local skin region to be measured is taken as a skin partition.

[0029] Furthermore, the method for obtaining the oiliness coefficient includes:

[0030] In the corresponding area of each skin partition in the heat map, the total area of all the pore areas is used as the first oiliness parameter, and the total number of all the acne areas plus a preset normal constant is used as the second oiliness parameter; the negative correlation mapping result of the water loss rate of each skin partition is used as the third oiliness parameter;

[0031] Fuse the first oiliness parameter, the second oiliness parameter and the third oiliness parameter to obtain the oiliness coefficient of the corresponding skin partition.

[0032] The present invention also provides an auxiliary detection system for local skin texture, including a memory, a processor, and a computer program stored in the memory and operable on the processor. When the processor executes the computer program, the steps of the auxiliary detection method for local skin texture are implemented.

[0033] The present invention has the following beneficial effects:

[0034] The present invention obtains the heat histogram of the heat map of the to-be-detected local skin area, and screens out all suspected defective skin heat intervals from all the heat intervals of the heat histogram, so as to prepare for subsequent determination of the corresponding areas of skin defects such as pores and acne based on the epidermal temperature difference characteristics; according to the distribution aggregation of the pixel points in the heat map under each suspected defective skin heat interval, the screening coefficient for each suspected defective skin heat interval to be a pore heat interval is obtained. The screening coefficient reflects the possibility that the pixel points in the heat map under each suspected defective skin heat interval are the corresponding pixel points of the pores. Furthermore, the pore heat intervals can be screened out according to the screening coefficient, and all the pore areas and all the acne areas in the heat map can be accurately obtained; then, based on the similarity and uniform distribution characteristics of the local pore areas, the to-be-detected local skin area is divided to obtain all skin partitions, so as to prepare for subsequent analysis and evaluation of the oiliness degree of the skin of each skin partition; the water loss rate in each skin partition is obtained, and then combined with the number of pore areas and the number of acne areas in the corresponding area of each skin partition in the heat map, the oiliness coefficient of each skin partition in the to-be-detected local skin area is obtained. The present invention combines the characteristics that oily skin is prone to skin defects such as pores and acne due to strong oil secretion, and the characteristic that the temperature of the defective epidermis is slightly higher than that of the normal healthy epidermis, quantitatively analyzes the oiliness coefficient of local skin, and improves the skin texture detection effect of local skin. Description of the Drawings

[0035] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0036] Figure 1 Flow chart of an auxiliary detection method for local skin texture provided by an embodiment of the present invention;

[0037] Figure 2 Thermal histogram provided by an embodiment of the present invention. Detailed implementation manners

[0038] In order to further elaborate on the technical means and effects adopted by the present invention to achieve the intended invention purpose, the following describes in detail an auxiliary detection system and detection method for local skin texture proposed according to the present invention, including its specific implementation manners, structures, features and effects, with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0039] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present invention belongs.

[0040] The following specifically describes the specific solutions of an auxiliary detection system and detection method for local skin texture provided by the present invention with reference to the accompanying drawings.

[0041] Please refer to Figure 1 , which shows a flow chart of an auxiliary detection method for local skin texture provided by an embodiment of the present invention, specifically including:

[0042] Step S1: Obtain the thermal histogram of the thermal map of the local skin area to be measured, and screen out all suspected defective skin thermal intervals from all thermal intervals of the thermal histogram.

[0043] It should be noted that in the embodiments of the present invention, the local skin targeted is facial skin, so the local skin area to be measured is the entire facial area; in other embodiments, the implementer can also set the local skin to be measured as a local skin area of other parts such as the back. The analysis and processing methods for all local skin areas are the same. Here, only the skin texture detection process of facial skin is briefly described.

[0044] Considering that the sebum secretion of the skin is one of the main factors determining the skin type, excessive sebum secretion of the skin usually leads to enlarged and blocked pores, and even causes skin inflammations such as acne. Also considering that due to factors such as excessive local sebaceous gland secretion or inflammatory infection on the skin surface such as pores and acne inflammations, the epidermal temperature is usually slightly higher than that of normal and healthy epidermis. Therefore, in the embodiments of the present invention, first, based on the surface temperature difference characteristics of different skins, a thermal map of the local skin area to be measured is obtained, so as to facilitate subsequent determination of skin defects such as pores and acne, and evaluation of the local sebum distribution in the local skin area to be measured, so as to assist in skin type evaluation.

[0045] In an embodiment of the present invention, a professional skin detector is used to obtain a thermal map of the local skin area to be measured. Since this embodiment evaluates the facial skin type, relevant preparations need to be made before collection, such as keeping the skin surface clear and closing the eyes, etc. Skin defects such as pores and acne are not directly shown in the thermal map, but the pixel values therein reflect the temperature information of the local skin area corresponding to each pixel point. The higher the pixel value, the higher the temperature, so that skin defects such as pores and acne can be indirectly determined.

[0046] It should be noted that obtaining a thermal map of the local skin area to be measured is already a prior art, and implementers can also use other methods to collect the thermal map, which will not be elaborated here.

[0047] Considering that a histogram can help analyze the distribution state of data, a thermal histogram can help more clearly analyze and evaluate the thermal distribution, facilitating subsequent screening of the thermal intervals corresponding to defective skin areas such as pores and acne. Therefore, in an embodiment of the present invention, a thermal histogram of the thermal map is further obtained. Specifically, the thermal value is used as the horizontal axis parameter, and the occurrence frequency of the thermal value is used as the vertical axis parameter. The value range of the thermal value is evenly divided into a preset number, such as 20 thermal intervals, and the occurrence frequency, that is, the frequency, of the thermal value in each thermal interval is obtained, so as to construct a thermal histogram.

[0048] Please refer to Figure 2 , which shows a thermal histogram provided by an embodiment of the present invention. The width of each column corresponds to a thermal interval, and the height of the column corresponds to the frequency of the thermal value in that thermal interval. It should be noted that the construction of the thermal histogram is already a well-known prior art to those skilled in the art, which will not be elaborated here. In other embodiments, implementers can also set the preset number of thermal intervals by themselves or adaptively construct a thermal histogram.

[0049] Considering that under normal circumstances, the area of normal and healthy skin in the local skin area to be measured should always be greater than the area of defective skin such as pores and acne; and considering that the bars in the heat histogram reflect the frequency information of heat values and indirectly reflect the size of the heat value area; based on this, all suspected defective skin heat intervals can be preliminarily screened out from all heat intervals of the heat histogram to prepare for subsequent screening of skin defect areas such as pores and acne.

[0050] Preferably, in an embodiment of the present invention, the method for obtaining the suspected defective skin heat interval includes:

[0051] In the heat histogram, the product of the normalized result of the frequency corresponding to each heat interval and the negatively correlated mapping result of the heat value corresponding to the right endpoint of each heat interval is used as the confidence coefficient that each heat interval is the normal skin temperature; the normal skin heat intervals are screened out from all heat intervals according to the confidence coefficient;

[0052] Based on the heat histogram, a heat distribution curve is fitted, and the heat interval corresponding to each maximum point in the heat distribution curve in the heat histogram is used as a suspected interval; in the heat histogram, all suspected intervals located on the right side of the normal skin heat interval are used as suspected defective skin heat intervals.

[0053] It should be noted that in this example, the heat histogram uses the heat value as the horizontal axis parameter, so the right side of the normal skin heat interval refers to the heat interval greater than the normal skin heat interval. The same applies to the subsequent mentions of the right side in this embodiment and will not be elaborated further.

[0054] As an example, the calculation formula for the confidence coefficient is: ; i is the serial number of the heat interval in the heat histogram; is the confidence coefficient that the i-th heat interval is the normal skin temperature; is the frequency corresponding to the i-th heat interval; Y is the total number of heat values in the heat map and is also the sum of the frequencies of all heat intervals; is the heat value corresponding to the right endpoint of the i-th heat interval.

[0055] In the above formula, specifically, the frequency of each heat interval is divided by the total frequency for normalization, and then the reciprocal of the heat value corresponding to the right endpoint of each heat interval is used for negative correlation mapping, so that the confidence coefficient of the heat interval with a larger normalized frequency value and a relatively lower heat value is larger, and the possibility that the corresponding skin area is a normal skin area is also larger;

[0056] After obtaining the confidence coefficient, in this example, the heat interval with the largest confidence coefficient in the heat histogram is further used as the normal skin heat interval; then the heat distribution curve corresponding to the heat histogram is fitted, that is, the histogram is curve-fitted; as Figure 2 ,Figure 2 The curve in it is the heat distribution curve, which, like the heat histogram, reflects the distribution information of heat values; then all the maximum values in the heat distribution curve are obtained. The maximum values reflect the mutation of the local heat distribution. Therefore, the heat intervals corresponding to the maximum values are initially taken as the suspected intervals. Furthermore, considering the characteristic that the surface temperature of pores and acne and other defective skins is usually higher than that of normal and healthy skin, all the suspected intervals on the right side of the normal skin heat interval are taken as the suspected defective skin heat intervals.

[0057] It should be noted that in other examples, the implementer can also adopt other normalization methods such as linear normalization, or the negative correlation mapping method that maps the negative correlation to the exponential function. These are all existing technologies and will not be elaborated here; the implementer can also select a preset number of consecutive heat intervals with a relatively large confidence coefficient, such as the first three consecutive heat intervals, as the normal skin heat interval, or directly take all the heat intervals on the right side of the normal heat interval as the suspected intervals for analysis.

[0058] Step S2: According to the distribution and aggregation of the pixel points in the thermal image under each suspected defective skin heat interval, obtain the screening coefficient for each suspected defective skin heat interval to be a pore heat interval; screen out the pore heat intervals from all the suspected defective skin heat intervals according to the screening coefficient, and obtain all the pore regions in the thermal image; according to the pore heat intervals and the distribution information of the pixel points in the thermal image under the remaining suspected defective skin heat intervals, obtain all the acne regions in the thermal image.

[0059] Considering that the pores in the local area are usually evenly distributed and the degree of pore dilation is relatively similar, in the embodiment of the present invention, the screening coefficient for each suspected defective skin heat interval to be a pore heat interval will be obtained according to the distribution and aggregation of the pixel points in the thermal image under each suspected defective skin heat interval. The screening coefficient reflects the possibility that the pixel points in the thermal image under each suspected defective skin heat interval are the pixel points corresponding to the pores, so as to facilitate subsequent screening out of the pore heat intervals, and further facilitate obtaining all the pore regions in the thermal image.

[0060] Preferably, in an embodiment of the present invention, the method for obtaining the screening coefficient includes:

[0061] Taking any suspected defective skin heat interval as the target interval, perform region connectivity detection on all the pixel points corresponding to the target interval in the thermal image, and take each connected domain as a suspected pore; construct a preset neighborhood with each suspected pore as the center;

[0062] In the preset neighborhood of each suspected pore, obtain the screening sub-parameter for the target interval to be a pore heat interval according to the spatial distribution uniformity of the suspected pores and the area similarity between different suspected pores;

[0063] The preset neighborhood corresponding screening sub-parameters of all suspected pores are combined to obtain the screening coefficient whose target interval is the pore thermal interval.

[0064] As an example, in the thermal map, after obtaining all suspected pores under the target interval, a circular preset neighborhood with a radius of 50 is constructed with each suspected pore as the center, and then in the preset neighborhood of each suspected pore, the filter parameters of the target interval as the pore thermal interval are analyzed and obtained; finally, the cumulative value of the filter parameters analyzed and obtained in the preset neighborhood of all suspected pores is used as the screening coefficient of the target interval as the pore thermal interval; by changing the target interval, the screening coefficient of each suspected defective skin thermal interval as the pore thermal interval can be obtained;

[0065] It should be noted that in other examples, the implementer can also set the neighborhood radius according to the actual situation, or directly construct a K neighborhood, that is, a local area containing K suspected pores as a preset neighborhood; regional connectivity detection is an existing technology and will not be described in detail.

[0066] Wherein, in a preferred embodiment of the present invention, the method for obtaining the screening sub-parameters includes:

[0067] In the preset neighborhood of each suspected pore, the suspected pore corresponding to the center of the preset neighborhood is taken as the target pore, and the remaining suspected pores are taken as reference pores; the deviations of the distances between each reference pore and the target pore relative to the average level are accumulated and negatively correlated, and the negative correlation mapping result is taken as the pore uniform distribution parameter; the negative correlation mapping result of the accumulated sum of the area differences between each reference pore and the target pore is taken as the pore area similarity parameter; the pore uniform distribution parameter and the pore area similarity parameter are integrated to obtain the screening sub-parameter whose target interval is the pore thermal interval;

[0068] Take any preset neighborhood of suspected pores in the target interval as an example, the calculation formula of the filter parameter is:

[0069] ; where k is the symbol of the reference pore; v is the symbol of the target pore; is the preset neighborhood of the target pore, and the target interval is the screening sub-parameter of the pore thermal interval; K is the total number of reference pores; is the Euclidean distance between the kth reference pore and the target pore; is the average Euclidean distance between all reference pores and target pores; is a very small positive parameter. In this example, it is 0.001 to prevent the denominator from being 0; is the pore uniform distribution parameter; is the area of ​​the kth reference pore; is the area of ​​the target pores; is the pore area similarity parameter;

[0070] In the above formula, the absolute value of the difference is specifically used to evaluate the deviation of the distance between each reference pore and the target pore from the average level. The smaller the deviation, the more similar the distance between each reference pore and the target pore, which to a certain extent reflects the more uniform distribution of suspected pores in the preset neighborhood. Then, after adding a very small positive parameter to the sum of the deviations, the reciprocal operation is performed to adjust the logic so that the smaller the sum of the deviations, the larger the pore uniform distribution parameter; at the same time, the absolute value of the difference is used to evaluate the area difference between the reference pore and the target pore, and then, after adding a very small positive parameter to the cumulative value of the area difference, the reciprocal operation is performed to adjust the logic so that the smaller the cumulative value of the area difference, the larger the pore area similarity parameter; finally, the pore uniform distribution parameter and the pore area similarity parameter are multiplied and fused to obtain the screening sub-parameter whose target interval is the pore heat map interval.

[0071] It should be noted that the area of a suspected pore can be evaluated by the total number of its internal pixel points; in other examples, the implementer can also adopt other negative correlation mapping means, such as taking the sum of accumulations as x in the exponential function exp(-x) with the natural constant e as the base, or combining the pore uniform distribution parameter and the pore area similarity parameter through basic mathematical means such as addition or weighted fusion, which are all common technical means and will not be elaborated here.

[0072] In another embodiment of the present invention, the implementer can also evaluate the pore uniform distribution parameter based on the centroid method. First, obtain the centroid and the center of the preset neighborhood, calculate the Euclidean distance between the corresponding positions of the center and the centroid, and further perform a negative correlation mapping on the Euclidean distance, such as performing a reciprocal operation after adding a very small positive parameter, and taking the result of the negative correlation mapping as the pore uniform distribution parameter. Further, the pore uniform distribution parameter and the pore area similarity parameter are fused to obtain the screening sub-parameter.

[0073] After obtaining the screening coefficient, the pore heat map interval can be screened out from all suspected defective skin heat map intervals according to the screening coefficient, and further all pore regions in the heat map can be obtained.

[0074] Preferably, in an embodiment of the present invention, the heat map interval with the largest screening coefficient is used as the pore heat map interval; perform region connectivity detection on all pixel points corresponding to the pore heat map interval in the heat map, and take each connected domain as a pore region, so as to obtain all pore regions in the heat map.

[0075] Considering that the surface temperature of the skin corresponding to acne is usually higher than the surface temperature of the skin corresponding to the pore region, and considering that the area of the region corresponding to acne in the heat map is also relatively large; therefore, in the embodiment of the present invention, all acne regions in the heat map will be obtained according to the pore heat map interval and the distribution information of pixel points in the heat map under the remaining suspected defective skin heat map intervals.

[0076] Preferably, in one embodiment of the present invention, considering that the skin surface temperature corresponding to acne is relatively higher than the skin surface temperature corresponding to pores, all suspected intervals located on the right side of the pores are taken as suspected acne thermal intervals; and considering that the skin temperature corresponding to acne is not only higher but also relatively larger in area, the distribution information of the pixel points in the heat map under each suspected acne thermal interval is further analyzed and evaluated to obtain the acne area; the method for obtaining the acne area includes:

[0077] In the thermal histogram, all suspected intervals located to the right of the pore thermal interval are regarded as suspected acne thermal intervals; in each suspected acne thermal interval, all corresponding pixels in the thermal map are subjected to regional connectivity detection, and each connected domain is regarded as a suspected acne;

[0078] The suspected acne thermal interval corresponding to the maximum total area of ​​suspected acne is used as the acne thermal interval; all suspected acne corresponding to the acne thermal interval are used as all acne areas in the heat map.

[0079] So far, all pore areas and all acne areas in the thermal map of the local skin area to be tested have been obtained.

[0080] Step S3, according to the location distribution and area information of all pore areas, divide the local skin area to be tested, and obtain all skin partitions; obtain the water loss rate in each skin partition; according to the water loss rate in each skin partition, combined with the number of pore areas and the number of acne areas in the corresponding area of ​​each skin partition in the thermal map, obtain the oiliness coefficient of each skin partition in the local skin area to be tested.

[0081] Considering that the pores in a local skin area are usually evenly distributed and have similar pore expansion, the pores in different local areas may be different. For example, the pores in a skin area with strong oil secretion are relatively large, and are more likely to present an oily skin quality, while the skin area with smaller pores is more likely to present a dry skin quality; therefore, the embodiment of the present invention further divides the local skin area to be tested according to the position distribution and area information of all pore areas, obtains all skin partitions, and then facilitates the subsequent analysis and evaluation of the skin oiliness of each skin partition, in preparation for the subsequent skin quality evaluation.

[0082] Preferably, in one embodiment of the present invention, considering that clustering can cluster similar features into one cluster, the pore areas are clustered based on the area characteristics of the pore areas in the local skin area to be measured and the distribution positions in the heat map, so that the skin areas that may be of similar skin quality are divided into a skin partition; the method for obtaining the skin partition includes:

[0083] Based on the area difference and spatial distance between pore regions, obtain the metric distance between different pore regions; based on the distance clustering algorithm and the metric distance, cluster all pore regions to obtain all clusters; take the corresponding region of each cluster in the heat map as a partition region, and take the corresponding region of each partition region in the heat map in the local skin region to be measured as a skin partition.

[0084] As an example, specifically, construct an image coordinate system with the center of the heat map as the coordinate origin, obtain the position coordinates of the center corresponding to each pore region, and then, between any two different pore regions, take the absolute value of the area difference as the first parameter and the Euclidean distance between the position coordinates as the second parameter. Furthermore, fuse the first parameter and the second parameter to obtain the Euclidean norm between different pore regions, and take the Euclidean norm as the metric distance between the corresponding different pore regions; then, based on the preset K value and the K-means algorithm, perform distance clustering on all pore regions in the heat map, where the preset K value is obtained based on the elbow method; then, take the corresponding local region of each cluster in the heat map as a partition region, and then through mask mapping, the corresponding skin partition of each partition region in the local skin region to be measured can be obtained.

[0085] It should be noted that the K-means algorithm, the elbow method, and mask mapping are all existing technologies; in other examples, the implementer can also use other clustering algorithms such as the DBSCAN algorithm to cluster all pore regions, which will not be elaborated here.

[0086] It should be noted that the skin partitions clustered and divided based on the relevant features of the pore regions may not cover all the local skin regions to be measured. Therefore, the other skin regions that cannot participate in the clustering and division are regarded as a whole skin partition.

[0087] Considering that the moisture loss situation on the skin surface can also help evaluate the skin texture. For example, the skin surface of oily skin has strong oil secretion and can form an oil film protection layer to help lock in moisture, while dry skin is more likely to appear dry and tight, and its surface moisture is more likely to be lost. Therefore, in an embodiment of the present invention, a relevant transepidermal water loss measuring instrument is further used to obtain the moisture loss rate at the center of each skin partition, so as to prepare for subsequent analysis of the oiliness coefficient of each skin partition.

[0088] In another embodiment of the present invention, the implementer can also directly divide the local skin region to be measured into a preset number of skin patches, such as 50, before dividing the skin partitions to be measured, use the transepidermal water loss measuring instrument to obtain the moisture loss rate of each skin patch, and then, after obtaining all the skin partitions, take the average value of the moisture loss rates of all the skin patches in the skin partition as the moisture loss rate of the skin partition.

[0089] It should be noted that obtaining the water loss rate in each skin area is already an existing technology well-known to those skilled in the art. Implementers can also use other relevant measurement methods to obtain it, so it will not be elaborated here.

[0090] After obtaining the water loss rate at each skin area, the oiliness coefficient of each skin area in the local skin area to be measured can be further obtained by combining the number of pore areas and the number of acne areas in the corresponding area of each skin area in the heat map, so as to prepare for subsequent evaluation of skin texture.

[0091] Preferably, in an embodiment of the present invention, considering that the more pores and the larger the area in each skin area, the slower the water loss rate, and the more acne, it indicates that the possibility of strong oil secretion in this skin area is greater, and the corresponding oiliness coefficient is also greater. Therefore, the method for obtaining the oiliness coefficient includes:

[0092] In the corresponding area of each skin area in the heat map, the total area of all pore areas is used as the first oiliness parameter, and the total number of all acne areas plus a preset normal constant is used as the second oiliness parameter; the negative correlation mapping result of the water loss rate of each skin area is used as the third oiliness parameter; the first oiliness parameter, the second oiliness parameter, and the third oiliness parameter are fused to obtain the oiliness coefficient of the corresponding skin area.

[0093] As an example, the calculation formula for the oiliness coefficient is: ; where u is the serial number of the skin area; is the oiliness coefficient of the u-th skin area; is the total area of the pore areas in the corresponding area of the u-th skin area in the heat map, and is also the first oiliness parameter; is the total number of acne areas in the corresponding area of the u-th skin area in the heat map; a is a preset normal constant, which is taken as 1 in this example to avoid the calculation result being meaningless due to the number of acne areas being 0; is the second oiliness parameter of the u-th skin area; is the water loss rate of the u-th skin area; is the third oiliness parameter of the u-th skin area; is the linear normalization function.

[0094] In the above formula, since the water loss rate of the skin area cannot be 0, the water loss rate of the skin area is specifically inverted to adjust the logic of negative correlation mapping, so that the smaller the water loss rate, the larger the corresponding oiliness coefficient; finally, the first oiliness parameter, the second oiliness parameter, and the third oiliness parameter are multiplied and normalized, so that the larger each parameter is, the larger the obtained oiliness coefficient is.

[0095] In other examples, the implementer may also adopt other negatively correlated mapping means, or may also adopt basic mathematical operations such as addition or weighted summation to combine the three parameters. All of these are existing technologies and will not be elaborated here.

[0096] In an embodiment of the present invention, after obtaining the oiliness coefficient of each skin sub-region in the local skin region to be measured, the implementer can assist relevant testers in evaluating the skin type of the local skin region to be measured based on the specific manifestations of oily skin type, dry skin type, and combination skin type; for example, for oily skin type, the oiliness coefficient of the skin sub-region corresponding to the T-zone in the facial region is large, and the oiliness coefficients of the other skin sub-regions are also relatively large, but slightly lower than that of the T-zone; the oiliness coefficient corresponding to the entire facial region of dry skin type will be small; for combination skin type, the oiliness coefficient of the skin sub-region corresponding to the T-zone is large, but the oiliness coefficients of the other skin sub-regions will be relatively small; based on this feature and other detection results of the local skin region to be measured, relevant testers can be assisted in evaluating the skin type.

[0097] It should be noted that the embodiments of the present invention only provide reference for relevant skin type analysis and do not directly diagnose and evaluate the skin type.

[0098] In an embodiment of the present invention, after obtaining the skin type of the local skin region to be measured, the implementer can further assign relevant skin type labels to the heat map of the local skin region to be measured, and then obtain a large number of labeled heat maps. All the labeled heat maps are divided into a training set, a validation set, and a test set according to the preset ratio of 60%, 20%, and 20%; then select EfficientNet as the convolutional neural network architecture, select the cross-entropy loss function as the loss function, and select the Adam optimizer to update the weights and biases of the model, so as to train the neural network model. Then, the newly obtained skin heat map is input into the trained neural network model, and the model can directly output the reference information for skin type assisted detection, improving the efficiency of skin type assisted detection.

[0099] It should be noted that the training and application of the convolutional neural network model are both existing technologies and will not be elaborated.

[0100] The present invention also proposes an auxiliary detection system for local skin type, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of an auxiliary detection method for local skin type are implemented.

[0101] In summary, the present invention first obtains the thermal histogram of the local skin area to be measured and all suspected defective skin thermal intervals therein; further determines all pore areas and all acne areas in the thermal image according to the distribution and aggregation of pixel points in the thermal image under each suspected defective skin thermal interval; then divides the local skin area to be measured according to the position distribution and area information of all pore areas, and further combines the water loss rate and the distribution of pores and acne in each skin partition to evaluate the oiliness coefficient of each skin partition. The present invention combines the characteristics of oily skin, such as strong oil secretion and easy appearance of skin defects such as pores and acne, and the characteristic that the epidermal temperature of the defect is slightly higher than that of the normal healthy epidermis, quantitatively analyzes the oiliness coefficient of the local skin, and improves the skin quality detection effect of the local skin.

[0102] It should be noted that the above order of the embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0103] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

Claims

1. An auxiliary detection method for local skin quality, characterized in that: The method comprises: Obtaining a thermal histogram of a thermal map of a local skin area to be tested, and screening out all thermal intervals of suspected defective skin from all thermal intervals of the thermal histogram; According to the distribution and aggregation of pixel points in the heat map under each suspected blemish skin thermal interval, the screening coefficient of each suspected blemish skin thermal interval as a pore thermal interval is obtained; according to the screening coefficient, the pore thermal interval is screened out from all suspected blemish skin thermal intervals, and all pore areas in the heat map are obtained; according to the pore thermal interval and the distribution information of pixel points in the heat map under the other suspected blemish skin thermal intervals, all acne areas in the heat map are obtained; According to the position distribution and area information of all the pore areas, the local skin area to be tested is divided to obtain all skin partitions; the water loss rate in each skin partition is obtained; according to the water loss rate in each skin partition, combined with the number of the pore areas and the number of the acne areas in the corresponding area of ​​each skin partition in the thermal map, the oiliness coefficient of each skin partition in the local skin area to be tested is obtained; The method for obtaining the screening coefficient includes: Taking any suspected blemish skin thermal interval as the target interval, performing regional connectivity detection on all pixel points corresponding to the target interval in the thermal map, and taking each connected domain as a suspected pore; constructing a preset neighborhood with each of the suspected pores as the center; In a preset neighborhood of each of the suspected pores, according to the uniformity of the spatial distribution of the suspected pores and the area similarity between different suspected pores, a screening sub-parameter whose target interval is the pore thermal interval is obtained; The preset neighborhoods of all the suspected pores correspond to the screening sub-parameters to obtain the screening coefficient whose target interval is the pore thermal interval; The method for obtaining the screening sub-parameters includes: In a preset neighborhood of each of the suspected pores, the center of the preset neighborhood corresponding to the suspected pore is used as the target pore, and the remaining suspected pores are used as reference pores; The deviations of the distances between each reference pore and the target pore relative to the average level are accumulated and negatively correlated, and the negative correlation mapping result is used as the pore uniform distribution parameter; the negative correlation mapping result of the accumulated sum of the area differences between each reference pore and the target pore is used as the pore area similarity parameter; The pore uniform distribution parameter and the pore area similarity parameter are integrated to obtain a screening sub-parameter whose target interval is the pore thermal interval.

2. The auxiliary detection method of local skin quality according to claim 1, characterized in that: The method for obtaining the thermal interval of the suspected defective skin includes: In the thermal histogram, the product of the normalized result of the frequency corresponding to each thermal interval and the negative correlation mapping result of the thermal value corresponding to the right end point of each thermal interval is used as the confidence coefficient of each thermal interval being a normal skin temperature; and the normal skin thermal interval is selected from all thermal intervals according to the confidence coefficient; A thermal distribution curve is fitted based on the thermal histogram, and the thermal interval corresponding to each maximum point in the thermal distribution curve in the thermal histogram is taken as a suspected interval; in the thermal histogram, all suspected intervals located to the right of the normal skin thermal interval are taken as suspected defective skin thermal intervals.

3. The auxiliary detection method of local skin quality according to claim 2, characterized in that: The method for obtaining the normal skin thermal range includes: The thermal interval with the largest confidence coefficient in the thermal histogram is taken as the normal skin thermal interval.

4. The auxiliary detection method of local skin quality according to claim 1, characterized in that: The method for obtaining the pore thermal interval and the pore area includes: The thermal interval with the largest screening coefficient is taken as the pore thermal interval; the pore thermal interval in the thermal map is subjected to regional connectivity detection corresponding to all pixel points, and each connected domain is taken as a pore area.

5. The auxiliary detection method of local skin quality according to claim 2, characterized in that: The method for obtaining the acne area comprises: In the thermal histogram, all suspected intervals located to the right of the pore thermal interval are regarded as suspected acne thermal intervals; in each suspected acne thermal interval, all corresponding pixel points in the thermal map are subjected to regional connectivity detection, and each connected domain is regarded as a suspected acne; The suspected acne thermal interval corresponding to the maximum total area of ​​the suspected acne is used as the acne thermal interval; all the suspected acnes corresponding to the acne thermal interval are used as all acne areas in the heat map.

6. The auxiliary detection method of local skin quality according to claim 1, characterized in that: The method for obtaining skin partitions comprises: Based on the area difference and spatial distance between the pore areas, the metric distance between different pore areas is obtained; based on the distance clustering algorithm and the metric distance, all the pore areas are clustered to obtain all clusters; the corresponding area of ​​each cluster in the heat map is taken as a divided area, and the corresponding area of ​​each divided area in the heat map in the local skin area to be measured is taken as a skin partition.

7. The auxiliary detection method of local skin quality according to claim 1, characterized in that: The method for obtaining the oiliness coefficient includes: In the area corresponding to each skin partition in the heat map, the total area of ​​all the pore areas is used as the first oiliness parameter, and the total number of all the acne areas plus a preset normal number is used as the second oiliness parameter; the negative correlation mapping result of the water loss rate of each skin partition is used as the third oiliness parameter; The first oiliness parameter, the second oiliness parameter and the third oiliness parameter are integrated to obtain the oiliness coefficient of the corresponding skin partition.

8. An auxiliary detection system for local skin quality, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of the auxiliary detection method for local skin quality as described in any one of claims 1 to 7 are implemented.

Images