Skin analysis system, configuration method and display device

Abstract

The application discloses a skin analysis system, a configuration method and a display device. The configuration method comprises the following steps: obtaining a skin analysis task, determining at least one target diffraction feature corresponding to the skin analysis task, determining m regions at a first surface according to the target diffraction feature, obtaining a detected physiological index and a preset calibration physiological index, determining a loss value based on the detected physiological index and the calibration physiological index, and updating structure information of an optical element, configuration information of a processor, or both the structure information of the optical element and the configuration information of the processor based on the loss value to obtain a configured skin analysis system corresponding to the skin analysis task. The configuration method provided by the application can configure a skin analysis system that takes into account skin detection accuracy and user privacy protection.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of optical detection, and in particular to a skin analysis system, a configuration method and a display device. BACKGROUND

[0002] Skin, as the largest organ of the human body, is a physical barrier to external invasion (resisting bacteria and pollutants) and is also a key tissue for regulating body temperature and excretion. The health status of the skin is easily damaged by internal and external factors such as ultraviolet light, environmental pollution, and aging, leading to problems such as wrinkles, dryness, and decreased barrier function. For example, the water content of the stratum corneum directly affects the appearance, elasticity, and function of the skin. Therefore, accurate analysis of the skin state is crucial for health management.

[0003] The prior art mainly relies on traditional cameras to directly collect visible light images of the user's face, and analyzes the apparent characteristics of the skin through algorithms. Although this type of technology can achieve basic analysis, it has fundamental defects: on the one hand, the original images obtained by the device contain complete facial biometric features, and there is a risk of privacy leakage in the process of data storage or transmission, and it cannot meet the increasingly stringent requirements of personal data protection regulations; on the other hand, due to the limitations of imaging and algorithm analysis capabilities, existing solutions can only obtain limited information from the surface of the skin, and cannot simultaneously extract other deep features. It can be seen that the existing technology is always difficult to balance the two core needs of privacy security and accurate analysis of multi-dimensional skin characteristics, and also restricts the development of the health monitoring field. SUMMARY

[0004] One of the purposes of the present application is to provide a configuration method of a skin analysis system to solve the problem that the skin analysis system configured by the prior art cannot balance the needs of privacy security and accurate analysis of multi-dimensional skin characteristics.

[0005] One of the purposes of the present application is to provide a skin analysis system.

[0006] One of the purposes of the present application is to provide a display device.

[0007] To achieve one of the above-mentioned purposes, an embodiment of the present application provides a configuration method of a skin analysis system, the skin analysis system comprising: an optical element comprising a first surface, configured to receive incident light from a subject, and configured to modulate diffraction feature information; and a processor configured to determine a skin physiological indicator of the subject based on the diffraction feature information; the configuration method comprising: obtaining a skin analysis task; determining at least one target diffraction feature corresponding to the skin analysis task; determining m regions at the first surface based on the target diffraction feature, the m regions being respectively configured to modulate incident light of m target wavebands corresponding to the target diffraction feature, m being an integer greater than or equal to 1; obtaining a detected physiological indicator and a preset calibration physiological indicator, the detected physiological indicator being detected by the skin analysis system based on a preset calibration subject, the calibration physiological indicator corresponding to the calibration subject; determining a loss value based on the detected physiological indicator and the calibration physiological indicator; and updating structure information of the optical element, updating configuration information of the processor, or updating both the structure information of the optical element and the configuration information of the processor based on the loss value, to obtain a configured skin analysis system corresponding to the skin analysis task.

[0008] Optionally, the optical element is configured to generate diffraction feature information including only high-order skin features.

[0009] Optionally, the configuration method comprises: determining m target wavebands corresponding to the target diffraction feature; and configuring the structure of the m regions to modulate only incident light of the corresponding target wavebands.

[0010] Optionally, the target wavebands comprise at least one of: a waveband with a wavelength of 380 nm to 780 nm or other natural light wavebands, a waveband with a wavelength of 780 nm to 2526 nm or other near-infrared light wavebands, a waveband with a wavelength of 590 nm to 610 nm or other orange light wavebands, a waveband with a wavelength of 440 nm to 475 nm or other blue light wavebands, a waveband with a wavelength of 492 nm to 577 nm or other green light wavebands, a waveband with a wavelength of 570 nm to 585 nm or other yellow light wavebands, and a waveband with a wavelength of 625 nm to 740 nm or other red light wavebands.

[0011] Optionally, at least one of the following is included: a distinct spot feature, a pigment deposition feature, a skin texture feature, and a flatness feature corresponding to a natural light waveband, a high-density color spot feature corresponding to an orange light waveband, a hemoglobin feature corresponding to a near-infrared light waveband, a acne feature, an epidermal inflammation feature corresponding to a blue light waveband, an epidermal spot feature, a pimple mark feature, a skin redness feature, a blood vessel dilation feature corresponding to a green light waveband, a sensitive state feature, a microcirculation feature corresponding to a yellow light waveband, a collagen feature, a deep inflammation feature, a structural depression feature, and a pigment depth feature corresponding to a red light waveband.

[0012] Optionally, the first surface is provided with microstructure units to form a metasurface, and the configuration method comprises at least one of the following: updating the phase arrangement of the microstructure units in the m regions on the first surface of the optical element based on the loss value, and updating the material of the microstructure units in the m regions on the first surface of the optical element based on the loss value.

[0013] Optionally, the skin analysis system comprises: a light sensor arranged on the light exit side of the optical element, the light sensor being configured to receive the diffraction feature information and generate a diffraction feature image, and the processor being configured to determine the skin physiological indicators of the subject according to the diffraction feature image.

[0014] Optionally, the optical element is configured according to at least one of the following: the optical element modulates the incident light to generate diffraction feature information for constructing a blurred image of the subject, and the optical element modulates the incident light to generate diffraction feature information of an image containing high-order skin features of the face of the subject.

[0015] Optionally, the processor is configured according to at least one of the following: the processor determines the skin physiological indicators of the subject by implementing a convolutional neural network, and the processor is configured to determine the skin health degree according to the diffraction feature information and a preset database.

[0016] To achieve one of the above purposes, an embodiment of the present application provides a skin analysis system, comprising: an optical element comprising a first surface, configured to receive incident light from a subject and generate diffraction feature information, and a processor configured to determine skin physiological indicators of the subject according to the diffraction feature information, wherein the skin analysis system is configured according to any configuration method of the present application.

[0017] Optionally, the first surface is provided with microstructure units to form a metasurface, the first surface is circular, and the first surface is configured according to at least one of the following: the m regions are arranged along a first diameter of the first surface to form a strip, the m regions are arranged along a first radius of the first surface from the center of the first surface to form one concentric circle and m-1 concentric annuli, and the m regions are arranged around the center of the first surface to form a sector.

[0018] To achieve one of the above purposes, an embodiment of the present application provides a display device, comprising: any skin analysis system of the present application, and a display screen configured to display the skin physiological indicators.

[0019] Optionally, the display device comprises: a mirror configured to present an image of the subject when the display device is implemented.

[0020] Compared with the prior art, the configuration method of the skin analysis system provided in the application can determine the physiological indicators of the skin through the diffraction characteristic information corresponding to multiple wave bands, and can detect the physiological conditions of the skin in multiple dimensions; the skin analysis system is based on diffraction characteristic information for analysis, so that the skin analysis is not limited to clear imaging of the tested object, thereby effectively protecting the privacy of the user; the skin analysis system is configured and trained in an end-to-end manner, so that the optical system containing the optical element and the electrical system containing the processor can be configured and fixed as a whole, the accuracy and reliability of the output results can be improved, and error accumulation can be prevented. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 is a schematic diagram of a skin analysis system in an embodiment of the application.

[0022] Figure 2 is a schematic diagram of a first surface in an embodiment of the application.

[0023] Figure 3 is a schematic diagram of a display device in an embodiment of the application.

[0024] Figure 4 is a schematic diagram of a configuration method of a skin analysis system in an embodiment of the application.

[0025] Figure 5 is a schematic diagram of a tested object and a diffraction characteristic image in an embodiment of the application. DETAILED DESCRIPTION

[0026] The application will be described in detail below with reference to the specific embodiments shown in the drawings. However, these embodiments do not limit the application, and the structural, method, or functional changes made by those of ordinary skill in the art based on these embodiments are also included in the protection scope of the application.

[0027] It should be noted that the term "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article, or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or other elements inherent to such a process, method, article, or device.

[0028] In addition, the terms "first", "second", "third", "fourth", etc. are only for descriptive purposes and cannot be understood as indicating or implying relative importance. There is no necessary relationship between the terms "first", "second", "third", "fourth", etc. For example, an embodiment provided in the application contains "second", which does not mean that the embodiment necessarily contains "first", and so on.

[0029] Skin analysis system

[0030] An embodiment of the present application provides a skin analysis system 100, as shown in the accompanying drawings. Figure 1

[0031] The skin analysis system 100 comprises an optical element 11. The optical element 11 comprises a first surface M1. The optical element 11 is configured to receive incident light from a subject T1 and generate diffraction feature information.

[0032] The first surface M1 can be provided with microstructure units to form a metasurface. The optical element 11 comprising the first surface M1 in the form of a metasurface can be defined as a metasurface optical element.

[0033] The metasurface optical element refers to an artificial layered material with a size smaller than or approximately equal to a wavelength, which can be regarded as a two-dimensional counterpart of metamaterials. The metasurface optical element can control the polarization, phase, amplitude, frequency, propagation mode, etc. of electromagnetic waves through surface sub-wavelength microstructure units (or superstructure units), to realize beam shaping, beam deflection, superlens, superholography, optical rotation, anti-reflection and other characteristics.

[0034] Meanwhile, the metasurface optical element is a sub-wavelength size optical element, which is suitable for current micron-level sensor architecture, and its preparation process is compatible with mature semiconductor sensor technology, and has strong practicability and economy.

[0035] Specifically, the metasurface optical element comprises a substrate and a plurality of microstructure units arranged in an array on the substrate, and each microstructure unit is provided with a nanostructure at a center and / or a vertex position. The microstructure unit is obtained by dividing the metasurface optical element to obtain a structure unit with each nanostructure as the center. Each nanostructure in each period constitutes a microstructure unit. The microstructure unit is a close-packed pattern, for example, can be a regular quadrilateral, a regular hexagon, a sector, etc., each period contains a nanostructure, and the vertex and / or center of the microstructure unit can be provided with a nanostructure. In the case of a regular hexagon, at least one nanostructure is provided at each vertex and center position of the regular hexagon. Similarly, the same applies to the sector and the square.

[0036] The substrate of the metasurface optical element can be selected from, for example, glass materials with similar refractive indexes such as SiO2, BF33, silicon, polymethyl methacrylate, etc.; the nanostructure can be selected from, for example, single crystal silicon (c-Si), polycrystalline silicon (p-Si), amorphous silicon (a-Si), compound semiconductors (such as GaN, GaP, GaAs, SiC, etc.), TiO2, Si3N4, AlSb, AlAs, AlGaAs, AlGaInP, BP, ZnGeP2, other suitable materials, and combinations of the above materials.​

[0037] Specifically, the nanostructure can be configured as a polarization-dependent structure or a polarization-independent structure. Depending on the use scenario, the nanostructure unit can be configured as a polarization-dependent structure or a polarization-independent structure. The polarization-independent structure, for example, a cylindrical shape, a square column shape, a cross column shape, a circular hole square column shape, etc. The polarization-dependent structure, for example, an elliptical column shape, a rectangular column shape, a hexagonal column shape, etc. The nanostructure can be a positive structure or a negative structure. For example, the shape of the nanostructure includes a cylinder, a hollow cylinder, a square prism, a hollow square prism, etc.

[0038] The metasurface optical element can further include a protective layer covering the nanostructure. The material of the protective layer can be any material with low refractive index and absorption coefficient in the visible or near-infrared waveband, for example: silicon dioxide (SiO2), spin-on glass (SOG), or polymers such as polymethyl methacrylate (PMMA), polydimethylsiloxane (PDMS), polymethyl pentene (PMP), and combinations of the above materials. It can also be air (i.e., no protective layer is provided).

[0039] The metasurface can be provided at one or more optical elements 11. For example, a first surface M1 of the optical element 11 is provided with a plurality of microstructure units to form a metasurface; the optical element 11 can also include a second surface, which is also provided with a plurality of microstructure units to form another metasurface. The first surface M1 and the second surface can be two surfaces oppositely arranged on the substrate of the optical element 11.

[0040] The metasurface is used to achieve feature extraction. The optical element 11 provided with multiple metasurfaces has stronger feature extraction capability and obtains more skin feature information than the optical element provided with only one metasurface.

[0041] When the optical element 11 is arranged on one side of the measured object, the first surface M1 of the optical element 11 can be close to the measured object, and the second surface of the optical element 11 can be away from the measured object; at least one of the first surface M1 and the second surface is formed as a metasurface.

[0042] In an embodiment, the optical element 11 can be a refractive optical element. In an embodiment, the optical element 11 can include a refractive optical element. The refractive optical element can be used to replace the metasurface optical element in the above embodiment. The refractive optical element can include but is not limited to a lens or a prism composed of optical glass, optical plastic, optical crystal, etc.

[0043] In an embodiment, the optical element 11 can be a diffractive optical element. In an embodiment, the optical element 11 can include a diffractive optical element. The diffractive optical element can be used to replace the metasurface optical element in the above embodiments. The diffractive optical element can include, but is not limited to, a binary or multi-step diffractive optical element, a grating, a Dammann grating, a super-structured surface, a hologram, a diffuser, a phase mask, an intensity mask, a spatial light modulator, etc.

[0044] In an embodiment, the optical element 11 can be a scattering medium element. In an embodiment, the optical element 11 can include a scattering medium element. The scattering medium element can be used to replace the metasurface optical element in the above embodiments. The scattering medium element can include, but is not limited to, ground glass, etc.

[0045] In configuring the skin analysis system, at least one of the following can be included: for the plurality of optical elements 11, the combination of different optical elements, the spacing of each optical element, etc. can be configured; for the metasurface optical element, the arrangement period, material, shape, size, position coordinates, etc. of the nanostructure can be configured; for the refractive optical element, the refractive index, curvature radius, etc. can be configured; for the diffractive optical element, the focal length characteristics, phase function of the diffractive surface, radial radius of each annular zone of the diffractive surface, annular zone depth of the diffractive surface, and diffraction efficiency, etc. can be configured.

[0046] Regardless of the optical element, it can include a first surface, and the first surface can be divided into m regions to process incident light of different wavebands.

[0047] The skin analysis system 100 includes a processor 12. The processor 12 is configured to determine the skin physiological indicators of the subject according to the diffraction feature information.

[0048] In an embodiment, the processor 12 directly determines the skin physiological indicators according to the diffraction feature information. In an embodiment, the diffraction feature information is transformed to generate another data information, and the processor 12 determines the skin physiological indicators of the subject according to the data information.

[0049] The skin physiological indicators can be type information of skin abnormalities, and position information of the skin abnormalities. The skin abnormalities can be spots, sensitivity, inflammation, acne marks, acne, wrinkles, etc.

[0050] The skin physiological indicators can also be evaluation information of the skin abnormalities. For example, after identifying the skin abnormalities, the severity of the skin abnormalities is determined and quantified to obtain the evaluation information. The evaluation information can be, for example, ratings such as slight, general, and severe, or scores in numerical form.

[0051] The skin physiological indicator can also be information about the overall health of the skin. For example, based on information characterizing skin abnormalities, the overall health of the skin is determined by comparing with a preset database, thereby determining the overall health of the skin. The preset database can be a public database about skin state such as ISIC 2019, Dermofit Image Library, or a private database established based on screening and classification of a public database.

[0052] The skin physiological indicator can also be other quantitative indicators generated based on high-order skin features. For example, it can be a pigment deposition indicator, a moisture indicator, and an oiliness indicator, etc.

[0053] The high-order skin feature can be a hyperspectral feature of the skin.

[0054] The high-order skin feature can be one of the target diffraction features. The target diffraction feature can include the high-order skin feature.

[0055] When the processor 12 is in operation, the foregoing technical solutions can be implemented by implementing a convolutional neural network. The convolutional neural network is configured to take the output of the optical system including the optical element 11 as input, and to determine skin abnormalities or health as a target for processing, to determine the skin physiological indicator of the subject.

[0056] The skin analysis system 100 is configured according to a configuration method of a skin analysis system. The configuration method can be any of the configuration methods described in the technical solutions hereinafter.

[0057] When the processor 12 is in operation and constructs the convolutional neural network, the model parameters of the convolutional neural network and the configuration of the optical element 11 can be performed simultaneously, so that the electrical system including the convolutional neural network and the optical system can be configured and fixed as a whole. The model parameters of the convolutional neural network can be the weights of each parameter.

[0058] The skin analysis system 100 can also include a light sensor 13. The light sensor 13 can be disposed on the light output side of the optical element 11. The light sensor 13 can be used to receive diffraction feature information and generate a diffraction feature image.

[0059] The light sensor 13 can be a CMOS (Complementary Metal Oxide Semiconductor) photosensitive element (CIS, CMOS Image Sensor), a CCD (Charge Coupled Device) photosensitive element, or an array light detector.

[0060] When the skin analysis system 100 is implemented, the optical element 11 obtains the light signal reflected at the subject T1, different regions of the optical element 11 respectively respond to light signals of different wave bands, and different diffraction characteristic information is correspondingly generated; the diffraction characteristic information is output to the light sensor 13 to form a first distribution, and the actual skin characteristics corresponding to the diffraction characteristic information have a second distribution at the subject T1, and the first distribution and the second distribution can be the same or correspond. In this way, the diffraction characteristic image generated by the light sensor 13 can show the corresponding position of the skin characteristics at the subject T1, so as to further analyze.

[0061] In the embodiment in which the diffraction characteristic information only includes high-order skin characteristics, the diffraction characteristic image can be a high-order characteristic image, which only shows the skin characteristics corresponding to the configuration of the skin analysis system 100, and does not include other private information of the subject T1.

[0062] The high-order characteristic image can be a diffraction characteristic image for showing high-order skin characteristics.

[0063] Specifically, for example, when the skin analysis system 100 is configured to perform a first analysis task on the subject T1, the configured optical element 11 only obtains diffraction characteristic information in a plurality of wave bands corresponding to the first analysis task, so that the formed diffraction characteristic image only contains a plurality of high-order skin characteristics corresponding to the first analysis task, and other information of the subject T1 is missing or ignored to a certain extent; due to this missing or ignoring, other irrelevant information is hidden or blurred, which plays a role in protecting the privacy of the subject T1.

[0064] The diffraction characteristic information can have the form of light information.

[0065] The diffraction characteristic image can have the form of digital information; the diffraction characteristic image contains all the diffraction characteristic information.

[0066] The skin analysis system 100 can include one or more optical elements 11. When the skin analysis system 100 includes a plurality of optical elements 11, the plurality of optical elements 11 can have the same or different parameters, and the plurality of optical elements 11 can be arranged at intervals in the propagation direction of the incident light; the plurality of optical elements 11 and the light sensor 13 together constitute an optical system of the skin analysis system 100.

[0067] The optical system can further include a lens.

[0068] In one embodiment, the skin analysis system 100 includes at least one lens, at least one optical element 11, and a light sensor 13.

[0069] In one embodiment, the lens, optical element 11, and light sensor 13 can be arranged sequentially in the direction of incident light propagation. In another embodiment, in the direction of incident light propagation, a portion of the lens, optical element 11, another portion of the lens, and light sensor 13 can be arranged sequentially, or a portion of the optical element 11, lens, another portion of the optical element 11, and light sensor 13 can be arranged sequentially, or a portion of the lens, a portion of the optical element 11, another portion of the lens, another portion of the optical element 11, and light sensor 13 can be arranged sequentially.

[0070] The processor 12 can be used to determine the skin physiological parameters of the test subject T1 based on the diffraction feature image.

[0071] Processor 12 can be coupled to optical sensor 13 to receive the diffraction feature image. In this embodiment, processor 12 can determine the skin physiological parameters based on the diffraction feature image containing diffraction feature information.

[0072] The processor 12 can also perform operations such as spectral data normalization to perform preprocessing on diffraction feature images.

[0073] Combination Figure 2 As shown, the first surface M1 is provided with microstructure units to form a metasurface.

[0074] The first surface M1 can be circular.

[0075] The circular first surface M1 can achieve uniform modulation of the incident light based on its rotational symmetry. Its boundary is less prone to diffraction anomalies compared to the corner areas of a rectangular structure, which can ensure the consistency of the light field distribution. Furthermore, based on the full-angle phase response, it can improve the detection accuracy to a certain extent.

[0076] The first surface M1 is determined to have m regions according to the configuration method provided in this application.

[0077] In one embodiment, such as Figure 2 As shown in (a), the m regions are arranged in a strip shape along the first diameter L1 of the first surface M1.

[0078] When the first surface M1 is positioned such that the first diameter L1 extends vertically, m regions are arranged vertically, and m regions can extend horizontally. When the first surface M1 is positioned such that the first diameter L1 extends horizontally, m regions are arranged horizontally, and m regions can extend vertically.

[0079] The m strip-shaped regions can have the same or different lengths. The m strip-shaped regions can have the same or different widths. The m strip-shaped regions can have the same or different areas.

[0080] The nanostructures in the microstructure units at m regions have the same or different phase arrangement, orientation (or rotation angle), material, shape, size, and position coordinates.

[0081] In one embodiment, such as Figure 2 As shown in (b), the m regions are arranged from the center O1 of the first surface M1 along the first radius R1 of the first surface to form a concentric circle and m-1 concentric rings.

[0082] The first radius R1 can be the radius extending in any direction from the center of the circle at the first surface M1 of the circle. The circles and annexes formed in this way have the same center and are arranged radially outward from the center. Among the m regions, the region closest to the center is the concentric circle, and the other regions are annular.

[0083] The m regions arranged in this way can have the same or different dimensions along the first radius R1; wherein, for the concentric circles, the dimension along the first radius R1 is the radius of the circle, and for the concentric annulus, the dimension along the first radius R1 is the width of the annulus. The m regions arranged in this way can have the same or different areas.

[0084] The nanostructures in the microstructure units at m regions have the same or different phase arrangement, orientation (or rotation angle), material, shape, size, and position coordinates.

[0085] In one embodiment, such as Figure 2 As shown in (c), the m regions are arranged around the center O1 of the first surface M1 to form a fan shape.

[0086] The m regions of a sector can have the same radius. The m regions of a sector can have the same or different central angles. The m regions of a sector can have the same or different areas.

[0087] The nanostructures in the microstructure units at m regions have the same or different phase arrangement, orientation (or rotation angle), material, shape, size, and position coordinates.

[0088] The microstructural units in m regions can be used to process optical signals in different wavelength bands, producing different responses, thereby achieving efficient and wide-band spectral dispersion.

[0089] In one embodiment, the m regions include a region dedicated to processing natural light wavelengths; specifically, the microstructure units in this region are configured to process only optical signals with wavelengths from 380 nm to 780 nm.

[0090] In one embodiment, the m regions include a region dedicated to processing near-infrared light; specifically, the microstructure units in this region are configured to process only optical signals with wavelengths from 780 nm to 2526 nm.

[0091] In one embodiment, the m regions include a region dedicated to processing the orange light band; specifically, the microstructure units in this region are configured to process only optical signals with wavelengths from 590 nm to 610 nm.

[0092] In one embodiment, the m regions include a region dedicated to processing the blue light band; specifically, the microstructure units in this region are configured to process only optical signals with wavelengths from 440 nm to 475 nm.

[0093] In one embodiment, the m regions include a region dedicated to processing the green light band; specifically, the microstructure units in this region are configured to process only optical signals with wavelengths from 492 nm to 577 nm.

[0094] In one embodiment, the m regions include a region dedicated to processing the yellow light band; specifically, the microstructure units in this region are configured to process only optical signals with wavelengths from 570 nm to 585 nm.

[0095] In one embodiment, the m regions include a region dedicated to processing the red light band; specifically, the microstructure units in this region are configured to process only optical signals with wavelengths from 625 nm to 740 nm.

[0096] The above-mentioned band-segmentation process involves the configuration of microstructural units in m regions. The actual light source used to illuminate the object under test and generate the incident light can be any visible light source that covers the desired target band.

[0097] In one embodiment, m = 6. In other embodiments, m can be any integer greater than or equal to 1.

[0098] In one embodiment, the configuration of the m regions in the configured optical element 11 is related to the skin analysis task currently being performed by the skin analysis system 100. Specifically, this may involve determining the target diffraction characteristics based on the skin analysis task, determining the corresponding target wavelength based on the target diffraction characteristics, and determining the number of regions based on the number of target wavelengths. In one embodiment, the number of target wavelengths is equal to the number of regions at the first surface M1.

[0099] Display devices

[0100] One embodiment of this application provides a display device 1000, such as... Figure 3 As shown.

[0101] The display device 1000 includes a skin analysis system. The skin analysis system can be the skin analysis system described in any of the technical solutions of this application, or it can be configured according to the configuration method of any of the technical solutions of this application.

[0102] Specifically, the display device 1000 includes an optical element 11. The optical element 11 includes a first surface M1. The optical element 11 is used to receive incident light from the object under test and generate diffraction feature information.

[0103] The first surface M1 can be configured with microstructure units to form a metasurface.

[0104] Specifically, the display device 1000 includes a processor 12. The processor 12 is used to determine the skin physiological parameters of the subject based on diffraction feature information.

[0105] At least the optical element 11 and the processor 12 are configured according to the configuration method of a skin analysis system provided in this application. The configuration method can be any of the configuration methods described in the following technical solutions of this application.

[0106] The optical element 11 and its first surface M1 can be used to form a detection module of the display device 1000. The detection module can be disposed on the top of the display device 1000 to realize skin detection and analysis of the test object without affecting the normal display of other parts.

[0107] Furthermore, the light sensor associated with the optical element 11 may also be included in the detection module. Based on the requirement of receiving incident light, the optical element 11 is at least partially exposed outside the housing of the display device 1000. In other words, at least some components of the optical system of the skin analysis system are exposed outside the housing of the display device 1000.

[0108] The electrical components of the skin analysis system can be housed within the casing of the display device 1000. For example, the processor 12 is housed within the casing, and the user cannot directly observe the processor 12 from the outside when using the display device 1000.

[0109] The processor 12 can be coupled to the detection module. In one embodiment, the processor 12 is coupled to the optical sensor.

[0110] Display device 1000 includes display screen 101. Display screen 101 is used to display skin physiological indicators. The skin physiological indicators are generated by the skin analysis system.

[0111] When the skin physiological indicators are information about the type of skin abnormality, the display screen 101 shows what kind of skin abnormality the subject currently has.

[0112] When the skin physiological indicators are evaluation information of skin abnormalities, the display screen 101 shows the score or rating of the severity of the skin abnormality of the current test subject.

[0113] When the skin physiological indicator is the overall health of the skin, the display screen 101 shows the score or rating of the current skin health of the subject.

[0114] The display device 1000 includes a reflector 102. The reflector 102 is used to present an image of the object under test when the display device 1000 is implemented.

[0115] When the display device 1000 includes a display screen 101, the display screen 101 may also be made of a specularly reflective material or covered with a cover plate made of a specularly reflective material to achieve an integrated appearance with the reflector 102.

[0116] In another embodiment, the display screen 101 can be integrated with the reflector 102, which is equivalent to displaying skin physiological indicators on the reflector 102.

[0117] In one embodiment, the display device 1000 includes the aforementioned light sensor, which is used to generate a diffraction feature image. The display screen 101 or the reflector 102 can also be used to output the diffraction feature image. Specifically, the display position of the diffraction feature image can correspond to the distribution area of ​​the tested object on the display screen 101 or the reflector 102, thereby achieving an augmented reality (AR) effect.

[0118] The display device 1000 may also have other additional functions. These additional functions can be configured not only by further configuration of the electrical or optical systems, but also by further configuration of its own structure. For example, based on the inclusion of a reflector 102, the display device 1000 may have the appearance of a cabinet, with a cavity for placing items, thus constituting a smart mirror cabinet.

[0119] Configuration method of skin analysis system

[0120] One embodiment of this application provides a configuration method for a skin analysis system, such as... Figure 4 As shown.

[0121] Combination Figure 1 The skin analysis system 100 includes an optical element 11. The optical element 11 includes a first surface M1. The optical element 11 is used to receive incident light from the test object T1 and generate diffraction feature information.

[0122] The first surface M1 can be configured with microstructure units to form a metasurface.

[0123] The optical element 11 can be configured based on any of the aforementioned technical solutions.

[0124] The test object T1 can be a human face, a human body, or the face or limbs of another animal.

[0125] The diffraction feature information may be high-order skin feature information; the diffraction feature information may include high-order skin feature information.

[0126] Higher-order skin features can be the hyperspectral characteristics of the skin. Higher-order skin feature information can be the hyperspectral characteristic information of the skin.

[0127] Before executing the configuration method described below, the optical element 11 can determine the initial first surface M1 based on the initialization parameters, or determine the initial configuration of the first surface M1 and then make fine adjustments through the configuration method described below.

[0128] The skin analysis system 100 includes a processor 12. The processor 12 is used to determine the skin physiological parameters of the subject based on diffraction feature information.

[0129] The skin physiological indicators can be information on the type of skin abnormality, evaluation information on skin abnormality, information on the overall health of the skin, or other quantitative indicators generated based on higher-order skin characteristics.

[0130] When the processor 12 is in operation, it can determine the skin physiological indicators of the test subject by implementing a convolutional neural network.

[0131] Before executing the configuration method described below, the processor 12 may determine the initial configuration of the processor 12 or the initial convolutional neural network based on the initialization parameters, and then make fine adjustments through the configuration method described below.

[0132] The configuration of the first surface M1 (e.g., the configuration of the microstructure units at the first surface M1) and the configuration information of the processor 12 can be uniformly fixed based on end-to-end training.

[0133] The processor 12 can be configured based on any of the aforementioned technical solutions.

[0134] Combination Figure 4 As shown, the configuration method of the skin analysis system includes at least one of the following steps.

[0135] Step S1: Obtain the skin analysis task and determine at least one corresponding target diffraction feature.

[0136] Step S2: Determine m regions on the first surface based on the target diffraction characteristics.

[0137] Each of the m regions is used to modulate the incident light in one of the m target wavelength bands corresponding to the target diffraction characteristics.

[0138] m is an integer greater than or equal to 1.

[0139] Step S3: Obtain the detection physiological indicators and the preset calibration physiological indicators.

[0140] The physiological indicators detected are obtained based on the skin analysis system's detection of preset calibration subjects.

[0141] The calibrated physiological indicators correspond to the calibrated objects.

[0142] Step S4: Determine the loss value based on the detected physiological indicators and the calibrated physiological indicators.

[0143] Step S5: Based on the loss value, update the structural information of the optical element, update the configuration information of the processor, or update both the structural information of the optical element and the configuration information of the processor to obtain the configured skin analysis system for the corresponding skin analysis task.

[0144] Thus, on the one hand, by determining m regions corresponding to the target diffraction features for the skin analysis task, and processing the target wavelength corresponding to the target diffraction features in each region, the configured skin feature analysis system can perform hyperspectral skin feature analysis and achieve accurate analysis of skin features; and on the other hand, since different regions process the incident light of the corresponding wavelength, the output of the optical element usually does not contain all the light signals required to fully image the test object, which not only achieves privacy protection from the beginning of the output of the optical element, but also reduces the data transmission pressure inside the skin analysis system and prevents error accumulation.

[0145] On the other hand, the configuration method constitutes an end-to-end training approach, where the structural information of the optical element and the configuration information of the processor can be uniformly trained and determined, maintaining the integrity of the system, and in particular, realizing the unification of the optical system and the electrical system; the loss value used to adjust the configuration is determined based on the output of the processor, which to some extent reduces the amount of computation and the complexity of the configuration process.

[0146] There is a correspondence between skin analysis tasks and target diffraction characteristics. For example:

[0147] When the skin analysis task is to assess and manage "acne-prone skin", the target diffraction features include epidermal inflammation features, acne scar features, and skin redness features.

[0148] When the skin analysis task is to develop an anti-aging plan for "early aging / mature skin", the target diffraction features include structural depression features (or wrinkle features) and microcirculation features.

[0149] When the skin analysis task is to target whitening and spot-fading care for "pigmentation / uneven skin tone", the target diffraction features correspond to epidermal pigmentation features and pigmentation depth features.

[0150] Assuming there are n target diffraction features, step S2 can be specifically as follows: first, determine the m target bands corresponding to the n target diffraction features, determine the m regions corresponding to the m target bands on the first surface, and then configure the initial structural information at the m regions according to the m target bands.

[0151] Specifically, the initial structural information of microstructure units in m regions is configured according to m target bands.

[0152] The value of n is determined based on the number of target diffraction features corresponding to the skin analysis task, the value of m is determined based on the number of target bands corresponding to the target diffraction features, and m is also the number of regions divided on the first surface.

[0153] The values ​​of m and n can be equal, in which case the number of target diffraction features, target bands, and regions are equal. The values ​​of m and n can also be unequal, in which case there may be a single target diffraction feature corresponding to multiple target bands, or multiple target diffraction features corresponding to a single target band.

[0154] Based on this, when performing steps S1 and S2, several target diffraction features corresponding to the skin analysis task are first determined. On the one hand, the target band corresponding to each target diffraction feature is determined. On the other hand, the first surface is divided into m regions, and the microstructure units in the m regions are configured to process the m target bands respectively, thereby realizing the extraction of hyperspectral diffraction features.

[0155] When the skin analysis system includes multiple first surfaces for extracting target diffraction features, preferably, the multiple first surfaces are divided into the same regions, and the corresponding regions on the multiple first surfaces are processed with optical signals of the same target wavelength.

[0156] When constructing the correspondence between the target diffraction characteristics and multiple regions on the first surface, it can be done by configuring the microstructure units in the region so that the region only processes optical signals with the corresponding target wavelength; or by making the microstructure units in the region have higher transmittance for optical signals with the target wavelength.

[0157] The calibration object corresponds to the calibration physiological indicator. The calibration object or its information can be a pre-defined, known test object or its information, and the calibration physiological indicator is a known skin physiological indicator of the known test object. In step S3, the skin analysis system performs skin analysis on the information of the calibration object to obtain the detection physiological indicator as the detection quantity, so that it can be compared with the standard quantity of the calibration physiological indicator in step S4.

[0158] The loss value can characterize the difference between the detected physiological index and the calibrated physiological index. Specifically, it can be calculated and determined using mean squared error or other loss functions.

[0159] The goal of the configuration method is to make the detected physiological indicators approach the calibrated physiological indicators. Therefore, step S5 can be: updating the structural information of the optical element and / or the configuration information of the processor until the loss value is less than the preset value or the loss value converges. Then, the structural information of the optical element at this time is fixed as the final structural information, and the configuration information of the processor at this time is fixed as the final configuration information, thereby finally determining the configured skin analysis system.

[0160] The structural information of the adjusted optical element can be the structural information of m regions.

[0161] Microstructure units can be set in m regions. In this case, the structural information of the optical element being adjusted can be at least one of the phase arrangement, orientation (or rotation angle), material, shape, size, and position coordinates of the nanostructures in the microstructure units in the m regions.

[0162] Taking phase alignment as an example, after the skin analysis system determines the final phase alignment information based on the loss value, since each of the m regions has a corresponding center wavelength, the shape, orientation (or rotation angle) and size of the nanostructure of the microstructure unit in the region can be determined accordingly, and a phase map of the microstructure unit that can cover a range of 2π under the current center wavelength can be obtained. The final optical element is then determined by combining the previously determined phase alignment information.

[0163] When the skin analysis system includes multiple optical elements, the structural information may also include the combination of the optical elements, the spacing between each optical element, etc.

[0164] The processor configuration information that is adjusted can be the processor's internal parameters, instructions, weights, etc.

[0165] The optical element can be configured to generate diffraction feature information that includes only high-order skin features.

[0166] Advanced skin features can be the hyperspectral features of the skin.

[0167] Higher-order skin features can be one type of target diffraction feature. The target diffraction feature can include higher-order skin features.

[0168] In one embodiment, the optical elements configured based on the initial structural information are configured to generate diffraction feature information that includes only high-order skin features. In another embodiment, the optical elements in the configured skin analysis system are configured to generate diffraction feature information that includes only high-order skin features. In yet another embodiment, the optical elements in the skin analysis system consistently generate diffraction feature information that includes only high-order skin features.

[0169] With this configuration, on the one hand, the skin analysis system can extract only high-level skin features without imaging, thus protecting privacy information such as facial features. Imaging refers to the process of obtaining all information about the test object and generating a display of its complete details. However, the optical element provided in this application only outputs diffraction feature information in the corresponding wavelength band of the test object through modulation. Even if a diffraction feature image is subsequently generated based on this information, it does not include the complete details of the test object, thus differing from the imaging process in existing technologies.

[0170] On the other hand, each of the m regions corresponds to a feature extraction process under different wavelengths. Skin physiological indicators are determined based on a comprehensive assessment of high-order skin features across multiple wavelengths, which is equivalent to making a comprehensive judgment based on richer skin condition information, thus obtaining more accurate analysis results.

[0171] In one embodiment, the configuration method further includes: determining m target bands corresponding to the target diffraction features; and configuring the structure of the m regions to modulate the incident light of the corresponding target bands only.

[0172] In one embodiment, the optical elements are configured to process optical signals of different wavelengths in different regions before end-to-end training is performed.

[0173] In one embodiment, after end-to-end training configuration, different regions of the determined optical element process optical signals of different wavelengths.

[0174] The two embodiments can be combined, and in other embodiments, the optical element can always be configured to process optical signals of different wavelengths for different regions thereon.

[0175] In one embodiment, the target band includes a band with wavelengths from 380 nm to 780 nm.

[0176] In one embodiment, the target wavelength band includes the natural light wavelength band.

[0177] For example, if the target diffraction features of several targets include the natural light band among the target bands, then at least one region on the first surface can be configured to modulate the incident light under the natural light band only, so as to accurately obtain the target diffraction features corresponding to the natural light band.

[0178] In one embodiment, the target band includes a band with wavelengths from 780 nm to 2526 nm.

[0179] In one embodiment, the target wavelength band includes the near-infrared band.

[0180] For example, if the target diffraction features of several targets include the near-infrared light band, then at least one region on the first surface can be configured to modulate the incident light in the near-infrared light band only, so as to accurately obtain the target diffraction features of the corresponding near-infrared light band.

[0181] In one embodiment, the target band includes a band with wavelengths from 590 nm to 610 nm.

[0182] In one embodiment, the target wavelength includes the orange light band.

[0183] For example, if the target diffraction features of several targets include an orange light band, then at least one region on the first surface can be configured to modulate the incident light under the orange light band only, so as to accurately obtain the target diffraction features corresponding to the orange light band.

[0184] In one embodiment, the target band includes a band with wavelengths from 440 nm to 475 nm.

[0185] In one embodiment, the target wavelength includes the blue light wavelength.

[0186] For example, if the target diffraction features of several targets include a blue light band, then at least one region on the first surface can be configured to modulate the incident light in the blue light band only, so as to accurately obtain the target diffraction features corresponding to the blue light band.

[0187] In one embodiment, the target band includes a band with wavelengths from 492 nm to 577 nm.

[0188] In one embodiment, the target wavelength includes the green light wavelength.

[0189] For example, if the target diffraction features of several targets include a green light band, then at least one region on the first surface can be configured to modulate the incident light under the green light band only, so as to accurately obtain the target diffraction features corresponding to the green light band.

[0190] In one embodiment, the target band includes a band with wavelengths from 570 nm to 585 nm.

[0191] In one embodiment, the target wavelength includes the yellow light wavelength.

[0192] For example, if the target diffraction features of several targets include a yellow light band, then at least one region on the first surface can be configured to modulate the incident light under the yellow light band only, so as to accurately obtain the target diffraction features corresponding to the yellow light band.

[0193] In one embodiment, the target band includes a band with wavelengths from 625 nm to 740 nm.

[0194] In one embodiment, the target wavelength includes the red light wavelength.

[0195] For example, if the target diffraction features of several targets include a red light band, then at least one region on the first surface can be configured to modulate the incident light in the red light band only, so as to accurately obtain the target diffraction features corresponding to the red light band.

[0196] In one embodiment, obvious spot features correspond to the natural light wavelength. In one embodiment, pigmentation features correspond to the natural light wavelength. In one embodiment, skin texture features correspond to the natural light wavelength. In one embodiment, smoothness features correspond to the natural light wavelength.

[0197] For example, if at least one of the target diffraction features corresponding to a skin analysis task includes obvious spot features, pigmentation features, skin texture features, and smoothness features, then the natural light band is determined as one of the target bands, and the structure at the first surface is configured to modulate only the incident light of the natural light band. The natural light band can be in the 380nm to 780nm band.

[0198] In one embodiment, the high-density color spot features correspond to the orange light band.

[0199] For example, if at least one target diffraction feature corresponding to a skin analysis task includes a high-density color spot feature, then the orange light band is determined as one of the target bands, and the structure at the first surface is configured to modulate only the incident light of the orange light band. The orange light band can be in the 590nm to 610nm band.

[0200] The high-density pigmentation includes melasma, freckles, and freckle-like moles.

[0201] In one embodiment, hemoglobin characteristics correspond to the near-infrared light band.

[0202] For example, if at least one target diffraction feature corresponding to a skin analysis task includes hemoglobin features, then the near-infrared light band is determined as one of the target bands, and the structure at the first surface is configured to modulate only the incident light in the near-infrared light band. The near-infrared light band can be in the 780nm to 2526nm band.

[0203] The hemoglobin characteristics correspond to capillary dilation, sensitive skin areas, etc.

[0204] In one embodiment, acne features correspond to the blue light band. In another embodiment, epidermal inflammation features correspond to the blue light band.

[0205] For example, if at least one target diffraction feature corresponding to a skin analysis task includes at least one of acne features or epidermal inflammation features, then the blue light band is determined as one of the target bands, and the structure at the first surface is configured to modulate only the incident light of the blue light band. The blue light band can be in the 440nm to 475nm band.

[0206] Blue light can also correspond to other types of inflammation.

[0207] In one embodiment, epidermal pigmentation features correspond to the green light band. In one embodiment, acne scar features correspond to the green light band. In one embodiment, skin redness features correspond to the green light band. In one embodiment, vasodilation features correspond to the green light band.

[0208] For example, if at least one of the target diffraction features corresponding to a skin analysis task includes at least one of epidermal pigmentation features, acne scar features, skin redness features, and vascular dilation features, then the green light band is determined as one of the target bands, and the structure at the first surface is configured to modulate only the incident light of the green light band. The green light band can be in the 492nm to 577nm band.

[0209] In one embodiment, the sensitive state characteristics correspond to the yellow light band. In another embodiment, the microcirculation characteristics correspond to the yellow light band.

[0210] For example, if at least one of the target diffraction features corresponding to a skin analysis task includes at least one of a sensitive state feature and a microcirculation feature, then the yellow light band is determined as one of the target bands, and the structure at the first surface is configured to modulate only the incident light of the yellow light band. The yellow light band can be in the 570nm to 585nm band.

[0211] Microcirculation characteristics can also specifically include fine lines, skin laxity, etc.

[0212] In one embodiment, collagen features correspond to the red light band. In one embodiment, deep inflammation features correspond to the red light band. In one embodiment, structural depression features correspond to the red light band. In one embodiment, pigmentation depth features correspond to the red light band.

[0213] For example, if at least one of the target diffraction features corresponding to a skin analysis task includes at least one of collagen features, deep inflammation features, structural depression features, and pigmentation depth features, then the red light band is determined as one of the target bands, and the structure at the first surface is configured to modulate only the incident light of the red light band. The red light band can be in the 625nm to 740nm band.

[0214] In one scenario, the first surface is provided with microstructure units to form a metasurface.

[0215] In one embodiment, the configuration method of this application includes the step of: updating the phase arrangement of microstructure units in m regions at the first surface of the optical element based on the loss value.

[0216] This step can be included in step S5, and in particular in the step of "updating the structural information of the optical element based on the loss value".

[0217] After determining the phase alignment, the shape, rotation angle, and size of the nanostructure can be adjusted by combining the center wavelength of the region. After determining the phase map of the region, the structural information of the optical element can be determined by combining the phase map and the phase alignment.

[0218] The phase arrangement includes at least one of the following: the arrangement of nanostructures within microstructure units, or the arrangement of microstructure units at metasurfaces.

[0219] In one embodiment, the configuration method of this application includes the step of: updating the material of microstructure units in m regions at the first surface of the optical element based on the loss value.

[0220] This step can be included in step S5, and in particular in the step of "updating the structural information of the optical element based on the loss value".

[0221] Compared to the phase arrangement of microstructural units, the material configuration of the microstructural units or nanostructures can be used to improve the information diversity of feature extraction. Specifically, microstructural units in different regions can be prepared based on different materials. In one embodiment, microstructural units or nanostructures in one of m regions can be prepared from TiO2, which can be used for the extraction of visible light-related features, such as the extraction of obvious spot features, pigmentation features, skin texture features, and smoothness features. In another embodiment, microstructural units or nanostructures in one of m regions can be prepared from amorphous silicon, which can be used for the extraction of near-infrared light-related features, such as the extraction of hemoglobin features.

[0222] Combination Figure 1As shown, the skin analysis system 100 may further include a light sensor 13. The light sensor 13 may be disposed on the light-emitting side of the optical element 11. The light sensor 13 may be used to receive diffraction feature information and generate a diffraction feature image.

[0223] The optical sensor 13 obtains diffraction feature information from the optical element 11, and the resulting diffraction feature image does not include features other than diffraction features, thus enabling privacy protection.

[0224] The processor 12 can be used to determine the skin physiological parameters of the test subject T1 based on the diffraction feature image.

[0225] The optical sensor 13 can be configured based on any of the aforementioned technical solutions.

[0226] In one embodiment, the optical element may be configured to modulate incident light to generate diffraction feature information for constituting a blurred image of the object under test.

[0227] In this embodiment, the diffraction feature information output by the optical element is used to generate a blurred image. Blurring refers to the blurred appearance of the object under test in the image.

[0228] The blurred image can be the diffraction feature image. The object under test appears blurred in the diffraction feature image. The optical sensor generates a blurred and encrypted diffraction feature image based on the diffraction feature information.

[0229] In one scenario, the blurring can be the effect of generating a diffraction feature image based on diffraction feature information. In other words, since the diffraction feature information corresponds to the target band, the various diffraction feature information corresponding to multiple target bands does not provide detailed imaging of the tested object, but there is a mapping relationship in the distribution. Therefore, the generated diffraction feature image presents a blurred appearance of the tested object, but this does not affect the privacy protection characteristics of the skin analysis system.

[0230] In another scenario, the blurring can be a configuration of the first surface. In one feasible embodiment, microstructural units in all regions of the first surface can be configured to have a blurred, encrypted structural configuration, thereby producing a blurred image with better encryption. In another feasible embodiment, a separate region on the first surface can be configured for blurred imaging, making the diffraction feature image more referential while ensuring the privacy of the object under test.

[0231] When the loss value converges or is less than a preset value, the diffraction feature image can be actually output to present the target diffraction features and / or the distribution of the target diffraction features.

[0232] In one embodiment, optical elements configured based on the initial structural information are configured to generate diffraction feature information for constructing a blurred image of the test object. In another embodiment, optical elements in a configured skin analysis system are configured to generate diffraction feature information for constructing a blurred image of the test object.

[0233] In one embodiment, the optical element may be configured to modulate incident light to generate diffraction feature information of an image containing high-order skin features of the face of the subject.

[0234] An image containing high-order skin features of the subject's face can be the diffraction feature image. The optical sensor generates a diffraction feature image containing high-order skin features of the face based on diffraction feature information.

[0235] Specifically, different regions of the optical element correspond to different wavebands. The diffraction feature image generated based on diffraction feature information under multiple wavebands has hyperspectral characteristics, and thus can present high-order skin features of the face.

[0236] In one embodiment, the processor may be configured to determine the skin physiological parameters of the subject by implementing a convolutional neural network.

[0237] Based on this, the processor's configuration information may include the model parameters and weights of the convolutional neural network. The model parameters and weights of the convolutional neural network participate in the configuration method and are ultimately determined together with the structural information of the optical element.

[0238] In one embodiment, the processor may be configured to determine the degree of skin health based on diffraction feature information and a preset database.

[0239] In one scenario, the skin health status can be the skin physiological indicator, or the skin health status can be included within the skin physiological indicator. In this case, the skin health status indicated by the diffraction feature information can be determined by matching it against the database based on the defined diffraction feature information, and this determination can be used as the output skin physiological indicator.

[0240] In another scenario, the skin health level is determined based on skin physiological indicators and serves as one of the processor's outputs. In this case, the determined skin physiological indicators can be matched against a database to determine and output the skin health level indicated by those indicators.

[0241] In one embodiment, to further enhance security, the aforementioned display device or skin analysis system does not exchange data with the Internet. In this case, the display device or skin analysis system also includes a memory for constructing the database as a local database.

[0242] In one embodiment, after determining diffraction feature information, skin physiological indicators, or skin health status, skin care suggestions (e.g., "replenish moisture if skin is dehydrated") or health reports (e.g., "UV damage detected") can be given based on the comparison results of the database.

[0243] In one scenario, the skin analysis task involves the assessment and management of acne-prone skin. Based on this, target diffraction features can be identified, including epidermal inflammation features, acne scar features, and skin redness features. On one hand, these features are determined to correspond to the blue and green light bands, respectively. On the other hand, a first surface can be divided into at least two regions (preferably, the first surface is divided into m regions, where m is a multiple of 2). Microstructural units in different regions are used to modulate the blue and green light bands in the incident light and generate diffraction feature information respectively. The processor, based on the diffraction feature information or the diffraction feature images generated therefrom, determines skin physiological indicators representing inflammation, acne scars, or redness by implementing a convolutional neural network.

[0244] When blue light is shone on the skin's surface, its short wavelength allows it to clearly highlight blemishes in the epidermis. More importantly, inflamed, active pimples are typically red or purple; under blue light, the color of these inflamed areas contrasts more strongly with the surrounding normal skin, making them clearly identifiable. This helps in assessing the current level of inflammation and activity of the pimples.

[0245] The green light exhibits superior performance in detecting red and brown features. For red acne scars (post-inflammatory erythema), which are essentially caused by dilated capillaries, green light is strongly absorbed by hemoglobin, allowing for very precise delineation of the extent and severity of red acne scars. For brown acne scars (post-inflammatory hyperpigmentation), which are essentially caused by melanin deposition, green light is also sensitive to melanin, clearly identifying these areas. Green light also provides excellent assessment results for overall facial redness and for sensitive, red skin accompanying acne.

[0246] In one scenario, the skin analysis task involves developing an anti-aging plan for "early aging / mature skin." Based on this, target diffraction features can be identified, including structural depression features (or wrinkle features) and microcirculation features. On one hand, these features are determined to correspond to the red and yellow light bands, respectively. On the other hand, the first surface can be divided into at least two regions (preferably, the first surface is divided into m regions, where m is a multiple of 2). Microstructural units in different regions are used to modulate the red and yellow light bands in the incident light and generate diffraction feature information respectively. Based on the diffraction feature information or the diffraction feature images generated therefrom, the processor, through the implementation of a convolutional neural network, determines skin physiological indicators representing wrinkle conditions, skin texture, or microcirculation.

[0247] The red light can penetrate into the dermis, and its reflection can indirectly reflect the skin's firmness and elasticity. Skin with abundant collagen and a firm structure reflects red light more evenly and brightly. Conversely, areas that are loose and have structural depressions (i.e., wrinkles) will appear as obvious shadows and dark areas on the red light image. By analyzing the distribution and depth of these shadows, deeper wrinkles and overall skin fullness can be effectively assessed.

[0248] The penetration depth of yellow light falls between that of green and red light, reaching the superficial dermis. Its primary function is to assess skin texture and microcirculation. Regarding skin texture, yellow light effectively highlights the surface texture, providing excellent insight into roughness and fine lines caused by dryness or aging. For microcirculation and skin tone, yellow light promotes blood and lymphatic circulation. By observing the skin's reaction and color under yellow light irradiation (e.g., whether it improves dullness), the health of the skin's microcirculation and overall complexion can be indirectly assessed. This offers guidance for improving the common "dull complexion" problem in mature skin.

[0249] In one scenario, the skin analysis task focuses on whitening and fading pigmentation / uneven skin tone. Based on this, the target diffraction features can be determined, including epidermal pigmentation features and pigment depth features. On one hand, these features are identified as corresponding to the green and red light bands, respectively. On the other hand, the first surface can be divided into at least two regions (preferably, the first surface is divided into m regions, where m is a multiple of 2). Microstructural units in different regions are used to modulate the green and red light bands in the incident light and generate diffraction feature information respectively. Based on the diffraction feature information or the diffraction feature images generated therefrom, the processor, through the implementation of a convolutional neural network, determines skin physiological indicators representing the quantitative status of surface pigmentation or uneven skin tone.

[0250] The green light has an extremely high absorption rate for melanin, making it the best choice for identifying epidermal pigmentation (such as sunspots, freckles, age spots, and brown acne scars). Under green light, these pigmented areas appear very dark, creating a strong contrast with the surrounding skin, facilitating precise location and area calculation. It also clearly shows uneven skin tone across the entire face.

[0251] Red light can provide clues about pigmentation levels. Because red light penetrates deeper, if a spot is very noticeable under green light (indicating a high level of melanin in the surface layer) but less so under red light (because red light "passes through" the surface and is reflected more by deeper tissues), then the spot is likely primarily located in the epidermis. Conversely, if a spot remains clearly visible under red light, it may indicate deeper pigmentation or a structural shadow. This contrast can provide a reference for assessing the stubbornness of pigmentation.

[0252] Figure 5 (a) shows a test subject targeted by the skin analysis system provided in this application during operation (the part involving the face in the figure is blurred, but when the technical solution of this application is implemented, the test subject itself has a relatively clear face, and the image of the test subject after mirror reflection can be presented as a relatively clear face). Figure 5 Figure (b) shows the diffraction feature image and related data generated by the skin analysis system provided in this application for the test subject. This related data can be used as a kind of skin physiological indicator. It can be seen that the facial features and details of the test subject cannot be distinguished in the diffraction feature image, but the high-order skin features of the test subject are clearly presented in the diffraction feature image. In this way, refined skin analysis is achieved while protecting privacy.

[0253] In summary, the configuration method of the skin analysis system provided in this application enables the configured skin analysis system to determine skin physiological indicators through diffraction feature information corresponding to multiple bands, and to detect the physiological condition of the skin in multiple dimensions. The skin analysis system performs analysis based on diffraction feature information, so that skin analysis is not limited to clear imaging of the test object, thus effectively protecting the user's privacy. The skin analysis system adopts an end-to-end configuration and training method, so that the optical system containing optical elements and the electrical system containing processor can be configured and fixed as a whole, which can improve the accuracy and reliability of the output results and prevent error accumulation.

[0254] It should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This way of describing the specification is only for clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

[0255] The detailed descriptions listed above are merely specific descriptions of feasible implementation methods of this application and are not intended to limit the scope of protection of this application. All equivalent implementation methods or modifications made without departing from the spirit of the art of this application should be included within the scope of protection of this application.

Claims

1. A configuration method of a skin analysis system, the skin analysis system comprising: an optical element comprising a first surface configured to receive incident light from a subject and modulate diffraction feature information, the optical element being configured according to at least one of: the optical element modulating the incident light to generate the diffraction feature information for forming a blurred image of the subject, the optical element modulating the incident light to generate the diffraction feature information for forming an image of high-order skin features of a face of the subject, and a processor configured to determine a skin physiological indicator of the subject based on the diffraction feature information, the configuration method comprising: obtaining a skin analysis task and determining at least one target diffraction feature corresponding to the skin analysis task, determining m regions at the first surface for modulating incident light of m target wavebands corresponding to the target diffraction feature, m being an integer greater than or equal to 1, different target wavebands corresponding to different diffraction feature information corresponding to actual skin features, obtaining a detected physiological indicator and a preset calibration physiological indicator, the detected physiological indicator being detected based on the skin analysis system and a preset calibration subject, the calibration physiological indicator corresponding to the calibration subject, determining a loss value based on the detected physiological indicator and the calibration physiological indicator, and updating structure information of the optical element, updating configuration information of the processor, or updating both the structure information of the optical element and the configuration information of the processor based on the loss value, to obtain a configured skin analysis system corresponding to the skin analysis task. The optical element is configured to generate diffraction feature information including only high-order skin features. The configuration method comprises: determining m target wavebands corresponding to the target diffraction feature, and configuring the structure of the m regions to modulate only incident light of the corresponding target wavebands. The target wavebands comprise at least one of: a waveband with a wavelength of 380 nm to 780 nm or other natural light wavebands, a waveband with a wavelength of 780 nm to 2526 nm or other near-infrared light wavebands, a waveband with a wavelength of 590 nm to 610 nm or other orange light wavebands, a waveband with a wavelength of 440 nm to 475 nm or other blue light wavebands, a waveband with a wavelength of 492 nm to 577 nm or other green light wavebands, a waveband with a wavelength of 570 nm to 585 nm or other yellow light wavebands, and a waveband with a wavelength of 625 nm to 740 nm or other red light wavebands. At least one of the following corresponds to the natural light waveband: obvious spot features, pigment deposition features, skin texture features, and flatness features, high-density color spot features correspond to the orange light waveband, hemoglobin features correspond to the near-infrared light waveband, acne features and epidermal inflammation features correspond to the blue light waveband, epidermal color spot features, acne mark features, skin redness features, and blood vessel dilation features correspond to the green light waveband, sensitive state features and microcirculation features correspond to the yellow light waveband, and collagen features, deep inflammation features, structural depression features, and pigment depth features correspond to the red light waveband. 6.The configuration method of claim 1, wherein the configuration method further comprises: determining a target waveband corresponding to a target diffraction feature, and configuring the structure of the m regions to modulate only incident light of the corresponding target waveband. ​ ​ ​ ​ ​ ​ 2. The configuration method of claim 1, wherein, ​ 3. The configuration method of claim 1, wherein, ​ ​ ​ 4. The configuration method of claim 3, wherein, ​ ​ ​ ​ ​ ​ ​ ​ 5. The configuration method of claim 3, wherein, ​ ​ ​ ​ ​ ​ ​ ​ ​ The first surface is provided with microstructure units to form a metasurface, The configuration method comprises at least one of the following: Based on the loss value, updating the phase arrangement of the microstructure units in the m regions at the first surface of the optical element, Based on the loss value, updating the material of the microstructure units in the m regions at the first surface of the optical element.

7. The configuration method of claim 1, wherein, The skin analysis system comprises: A light sensor provided on the light-emitting side of the optical element, the light sensor being used to receive diffraction feature information and generate a diffraction feature image, The processor is used to determine the skin physiological indicators of the subject according to the diffraction feature image.

8. The configuration method of claim 1, wherein, The processor is configured according to at least one of the following: The processor determines the skin physiological indicators of the subject by implementing a convolutional neural network, The processor is used to determine the skin health degree according to the diffraction feature information and a preset database.

9. A skin analysis system characterized by, Comprise: An optical element comprising a first surface for receiving incident light from a subject and generating diffraction feature information, A processor for determining the skin physiological indicators of the subject according to the diffraction feature information, The skin analysis system is configured according to the configuration method of any one of claims 1-8.

10. Skin analysis system according to claim 9, characterized in that The first surface is provided with microstructure units to form a metasurface, The first surface is circular, and the first surface is configured according to one of the following: The m regions are arranged in a strip shape along a first diameter of the first surface, The m regions are arranged in a concentric circle and m-1 concentric annuli along a first radius of the first surface from the center of the first surface, The m regions are arranged in a fan shape around the center of the first surface.

11. A display device, characterized by comprising: Comprise: The skin analysis system of claim 9 or 10, A display screen for displaying the skin physiological indicators.

12. The display device of claim 11, wherein, The display device comprises: A mirror for presenting an image of the subject when the display device is implemented.

Images