Hair follicle positioning hair removal device and method based on multi-frequency skin electrical impedance scanning

Abstract

The invention discloses a hair follicle positioning hair removal device and method based on multi-frequency skin electrical impedance scanning, and belongs to the technical field of medical cosmetology. The device comprises a microelectrode array module, an impedance measurement module, a visual detection module, a control module, a laser control module and a display prompt module, and all the modules cooperate to achieve the overall functions of skin electric signal transmission, impedance data collection and calculation, hair follicle visual detection, bimodal fusion positioning, laser precise irradiation and device work and skin contact state prompt. The method comprises the following five steps: S1, initializing and calibrating a microelectrode array; S2, acquiring a multi-frequency impedance signal; S3, constructing and filtering an impedance spectrum; S4, detecting candidate positions of hair follicles; impedance and vision dual-mode fusion positioning is adopted, a four-electrode measurement method is combined to eliminate contact resistance and interference resistance, the hair follicle positioning precision is improved through multi-step data optimization and refinement processing, and the practicability is high.

Description

Technical Field

[0001] This invention relates to the field of medical aesthetics technology, specifically a hair removal device and method based on multi-frequency skin electrical impedance scanning for hair follicle localization. Background Technology

[0002] The core of laser hair removal technology is the precise location of hair follicles to ensure that laser energy effectively targets the hair follicle root for hair removal while minimizing damage to surrounding normal skin tissue. Current technologies primarily rely on visual detection for hair follicle localization. This method uses a camera to capture skin images and then employs image recognition algorithms to detect hair follicles. However, it is significantly affected by the skin's surface condition, limiting its accuracy and leading to missed or false detections. Firstly, light-colored hair has extremely low optical contrast with skin. While visual models can achieve over 95% accuracy in recognizing dark hair, their accuracy for light-colored hair such as blonde, white, and gray is only 60-70%, resulting in significant missed detections. Given that approximately 30% of the global population has light-colored body hair, this represents a clear market blind spot. Secondly, the skin surface may exhibit both residual black hair roots after shaving and the presence of hair follicles without visible hair growth, making it impossible for purely visual solutions to detect hair growing under the skin. Furthermore, factors such as skin sweat reflection and skin texture interference can also lead to misjudgments in visual recognition.

[0003] Hair follicles, as appendages of the skin, are rich in cells, blood vessels, and water. This structure causes their electrical properties to differ significantly from the surrounding stratum corneum and dermal connective tissue. Specifically, the electrical impedance value of the hair follicle area is usually lower than that of the surrounding skin without hair follicles. This difference can be detected and captured by a high-sensitivity microelectrode array. Based on this, some existing methods use skin impedance measurement as an auxiliary means of hair follicle localization. However, these methods are mostly single-frequency impedance measurements, which cannot fully reflect the differences in electrical properties of different skin tissues, resulting in low impedance identification between hair follicles and surrounding skin. At the same time, conventional two-electrode measurement methods are susceptible to interference from the electrode-skin contact impedance, further increasing the localization error and ultimately affecting the effectiveness and safety of laser hair removal. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a hair removal device and method based on multi-frequency skin electrical impedance scanning for hair follicle localization. It improves the electrical identification of hair follicles through multi-frequency impedance feature analysis, eliminates contact impedance interference by combining a dedicated measurement method, and improves overall positioning accuracy through visual and impedance dual-modal fusion positioning. At the same time, it adapts to the needs of different usage scenarios, taking into account both positioning accuracy and real-time performance.

[0005] To achieve its technical objectives, the present invention employs the following technical solution:

[0006] A hair removal device based on multi-frequency skin electrical impedance scanning for follicle localization includes a microelectrode array module, an impedance measurement module, a visual detection module, a control module, a laser control module, and a display and prompting module.

[0007] The microelectrode array module is electrically connected to the impedance measurement module via a flexible circuit board, and is used to contact the skin and transmit electrical signals.

[0008] The impedance measurement module is communicatively connected to the control module and is used to generate AC excitation signals, perform electrode gating, and collect and calculate local impedance data of the skin.

[0009] The visual detection module is communicatively connected to the control module and is used to acquire skin images and detect hair follicle locations, and output a set of visual hair follicle locations.

[0010] The control module communicates with each module to achieve coordinated control. It receives impedance data from the impedance measurement module and hair follicle location data from the visual detection module. Based on the impedance data, it constructs an impedance spectrum and determines the candidate hair follicle location. Then, it performs spatial matching and fusion positioning between the candidate hair follicle location and the detection result of the visual detection module, and outputs the fused hair follicle coordinate set.

[0011] The laser control module is communicatively connected to the control module and is used to perform laser irradiation based on the fused hair follicle coordinate set output by the control module.

[0012] The display and prompt module is electrically connected to the control module and is used to issue skin contact status prompts and device working status prompts.

[0013] Preferably, the microelectrode array module includes a 4×4=16 microelectrode array, with an electrode spacing between 1-3 mm and a diameter between 0.5-0.8 mm; the electrode material of the microelectrode array module is gold-plated copper.

[0014] Preferably, the impedance measurement module includes a signal generation unit, a signal acquisition unit, and an electrode selection and connection unit. The signal generation unit is used to generate a multi-frequency sinusoidal AC excitation signal. The signal acquisition unit is used to acquire voltage signals and combine them with the excitation signal to complete local impedance calculation. The electrode selection and connection unit is used to realize electrode selection and path connection of the microelectrode array. The excitation current output by the signal generation unit is ≤1mA, and the excitation frequency of the multi-frequency sinusoidal AC excitation signal generated by the signal generation unit is 10k / 100k / 1M Hz.

[0015] More preferably, the signal generation unit and the signal acquisition unit are integrated into the AFE chip, and the electrode selection and connection unit is composed of an analog multiplexer; the control module is an MCU, and the MCU is connected to the AFE chip and the analog multiplexer of the impedance measurement module via an SPI bus; the spatial matching threshold is 1mm; the impedance measurement module allocates the electrode path through the electrode selection and connection unit to realize the separation of the current injection and voltage acquisition path in the four-electrode measurement method, and eliminates the interference of the electrode-skin contact impedance.

[0016] A hair follicle localization method based on multi-frequency skin electrical impedance scanning, applied to the hair removal device, includes:

[0017] S1 Microelectrode Array Initialization and Calibration: The control module completes the power-on self-test, electrode connectivity detection, air baseline acquisition, and skin contact detection of the microelectrode array, providing a reference for subsequent impedance measurements;

[0018] S2 Multi-frequency impedance signal acquisition: For the microelectrode array that has been calibrated and achieved effective skin contact in step S1, a multi-frequency AC excitation signal with preset parameters is applied through the impedance measurement module, and the four-electrode measurement method is used to complete the full array impedance scan and acquire local impedance data at multiple frequencies.

[0019] S3 Impedance Map Construction and Filtering: Based on the multi-frequency local impedance data collected in step S2, an impedance distribution map is generated by interpolation, and then Gaussian smoothing and noise reduction are performed to obtain low-frequency, mid-frequency and high-frequency impedance characteristic maps; Cole-Cole model parameters are extracted to obtain high-quality impedance characteristic maps.

[0020] S4 Hair follicle candidate location detection: On the low-frequency impedance feature map in step S3, local minimum search, threshold screening and sub-pixel refinement are completed in sequence, and multi-frequency cross-validation is performed to obtain a set of hair follicle candidate locations with confidence.

[0021] S5 integrates visual and impedance detection results to achieve precise hair follicle localization and output: The set of candidate hair follicle positions obtained in step S4 is aligned with the set of hair follicle positions output by the visual detection module, spatially matched and the results are fused, and the fused hair follicle coordinate set is output to the laser control module to perform laser irradiation.

[0022] More preferably, step S1 includes the following steps:

[0023] S101 Power-on self-test: After the device is powered on, the control module sends an initialization command to the impedance measurement module to configure the gain, excitation frequency, and sampling rate parameters of the impedance measurement. The impedance measurement module performs self-calibration on its internal signal generation unit and signal acquisition unit.

[0024] S102 Electrode Connectivity Detection: The control module sequentially selects and connects each electrode of the microelectrode array through the electrode selection and connection unit, applies a 100kHz / 100μA test signal, and measures the impedance value from each electrode to the common reference electrode. If the open-circuit impedance of the electrode is >10MΩ, it is marked as a faulty electrode and skipped in subsequent calculations.

[0025] S103 Air Baseline Acquisition: When the device is not in contact with the skin, it completes a full array scan of the microelectrode array through the impedance measurement module and records the baseline impedance value Z_air of each electrode pair in the air;

[0026] S104 Skin Contact Detection: After the device is attached to the skin, the impedance measurement module detects the electrode impedance value in real time. When more than 75% of the electrode pair impedance values ​​drop to less than 1 / 10 of Z_air, it is determined that the skin is in effective contact and the measurement mode is entered. Otherwise, the display prompt module will issue a contact adjustment prompt.

[0027] More preferably, step S2 includes the following steps:

[0028] S201 Excitation Signal Configuration: The control module controls the signal generation unit to generate 10kHz, 100kHz, and 1MHz sinusoidal excitation signals. The excitation current is in constant current mode with an amplitude ≤1mA, and the peak-to-peak excitation voltage Vpp <1V.

[0029] S202 Four-Electrode Measurement Method: Four adjacent electrodes are selected in the microelectrode array to form a measurement unit. The two electrodes on either side are current injection electrodes, applying an excitation current Iexc. The two electrodes in the middle are voltage acquisition electrodes, measuring the potential difference Vbc. Finally, according to… Calculate the local impedance;

[0030] S203 full array scanning sequence: The microelectrode array is sequentially selected and connected in the horizontal, vertical and diagonal directions by the electrode selection and connection unit. Each group of four electrodes completes an impedance measurement once at three frequencies, with a single measurement time of 0.5-1ms.

[0031] S204 Data Recording: The control module records the dataset {Z(i,j,f_k)} of the full array scan, where i and j are the electrode space coordinate indices, k=1,2,3 are the frequency indices, and f_k is the frequency value. The dataset is temporarily stored in the storage unit of the control module.

[0032] More preferably, step S3 includes the following steps:

[0033] S301 raw data interpolation: Bilinear interpolation is used to map the discrete local impedance measurements onto a regular grid, generating impedance distribution spectra Z10k(x,y), Z100k(x,y), and Z1M(x,y) for the three frequencies of 10kHz, 100kHz, and 1MHz, respectively.

[0034] S302 Gaussian Smoothing Denoising: A two-dimensional Gaussian filter is applied to each impedance distribution spectrum. The sigma of the Gaussian kernel function is set to 0.5-1 times the electrode spacing to eliminate random noise and retain the effective signal at the hair follicle scale, resulting in low-frequency impedance characteristic spectrum Z10k_smooth(x,y), mid-frequency impedance characteristic spectrum Z100k_smooth(x,y) and high-frequency impedance characteristic spectrum Z1Mk_smooth(x,y).

[0035] S303Cole-Cole parameter extraction: For each spatial location (x,y), the Cole-Cole model is fitted using impedance characteristic spectra at three frequencies to extract low-frequency impedance R_0, high-frequency impedance R_inf, and characteristic time constant tau. The dispersion coefficient alpha is set to a fixed empirical value of 0.8 to distinguish between hair follicle regions and non-hair follicle regions. The tau value of the hair follicle region is smaller than that of the non-hair follicle region, and the alpha value is closer to 1.

[0036] More preferably, step S4 includes the following steps:

[0037] S401 Local Minimum Search: On the smoothed low-frequency impedance feature map Z10k_smooth(x,y), traverse each pixel and compare its impedance value with the impedance values ​​of its 8 neighboring pixels. Pixels with impedance values ​​less than those of the 8 neighboring pixels are marked as candidate local minima.

[0038] S402 Threshold Screening: Calculate the ratio Ratio of the impedance value of each local minimum candidate point to the average impedance value of its 8 neighboring pixels. When Ratio < 0.85, it is confirmed as a candidate location for hair follicles.

[0039] S403 Subpixel Localization Refinement: For candidate hair follicle locations that have passed the threshold screening, subpixel localization is performed using two-dimensional parabolic fitting within a 3×3 window centered on the follicle and containing its 8 neighboring pixels, in order to improve the spatial resolution of hair follicle localization.

[0040] S404 Multi-frequency Cross-validation: Repeat steps S401-S403 for impedance characteristic spectra of low-frequency, mid-frequency, and high-frequency bands respectively to obtain the candidate hair follicle locations in each band; cross-compare the candidate points detected by the three frequency impedance characteristic spectra; mark the candidate points detected in at least two frequency impedance characteristic spectra and the spatial distance between the two candidate points is less than one electrode spacing as high confidence; mark the candidate points detected in only one frequency spectrum as medium confidence; output the set of hair follicle candidate locations I_set={(x_i,y_i,confidence_i)}, where confidence is either high or medium.

[0041] More preferably, step S5 includes:

[0042] S501 Coordinate System Alignment: During the device manufacturing stage, the relative positional relationship between the camera's field of view and the microelectrode array is calibrated to obtain the affine transformation matrix M_calib. During runtime, the corresponding coordinates of the hair follicle candidate location set obtained from impedance detection in step S4 are transformed to image coordinates. imp is impedance detection, and img is image detection;

[0043] S502 Spatial Matching: The impedance hair follicle candidate location set I_set, which is transformed into image coordinates, is spatially adjacent to the visual hair follicle candidate location set V_set. The matching threshold is 1mm. For each candidate point in I_set, the nearest visual hair follicle point is searched in V_set. When the distance between the two points is less than the matching threshold, they are determined to be a dual-modal detection matching pair of the same hair follicle. After matching, three types of results are obtained: dual-modal matching, visual detection only, and impedance detection only.

[0044] S503 Result Fusion: For bimodal matching results, the qualitative confidence levels "high" and "medium" marked in S4 are first mapped to quantitative values, with high confidence = 0.7 and medium confidence = 0.5. The visual detection module outputs a quantitative confidence level of 0~1. Then, coordinates and confidence levels are fused. The formulas for calculating fused coordinates and fused confidence levels are as follows:

[0045] , ,

[0046] The fusion confidence bimodal reward is 0.1, with a maximum of 1.0; the weight calculation in the fusion coordinate formula is as follows:

[0047] ,

[0048] The laser irradiation energy is the standard energy. V_j and I_i are the coordinates of the hair follicle candidate points output by the vision and impedance modules, w_v / w_i are the fusion weights of the corresponding coordinates, and conf_v / conf_i are the quantitative confidence values ​​of the vision and impedance modules.

[0049] For visual inspection results only, the output coordinates and confidence level of the visual inspection module are directly used, and the laser irradiation energy is the standard energy. For impedance inspection results only, the confidence level is set to 0.5-0.7 according to its confidence level, and 0.7 is used for "high" and 0.5 for "medium" according to the confidence level of the impedance candidate points. The laser irradiation energy is 70% of the standard energy.

[0050] The S504 outputs to the laser control module: The control module generates a fused hair follicle set: F_set{(x_k,y_k,conf_k,energy_k)}, where x_k and y_k are the coordinates of the k-th hair follicle, conf_k is the confidence level of the k-th hair follicle, and energy_k is the laser irradiation energy of the k-th hair follicle; and then sorts the hair follicles by confidence level from high to low before transmitting them to the laser control module.

[0051] The beneficial effects of this invention are as follows:

[0052] This invention employs a dual-modal fusion positioning method combining multi-frequency skin impedance scanning and visual detection. This overcomes the technical shortcomings of existing single visual detection, which is susceptible to interference from skin surface conditions, and single-frequency impedance detection, which has low recognition accuracy. At the same time, it eliminates the interference of electrode-skin contact impedance at the hardware level through a four-electrode measurement method. Combined with Gaussian smoothing noise reduction, sub-pixel positioning refinement, and multi-frequency cross-validation, this invention significantly improves the accuracy and anti-interference ability of hair follicle positioning, effectively reduces laser damage to surrounding normal skin tissue, and ensures hair removal effect and safety.

[0053] This invention is specifically designed to meet the practical needs of home-use hair removal. It adopts a 4×4 microelectrode array and a 10k / 100k / 1M Hz tri-frequency excitation signal design. While ensuring positioning accuracy, it simplifies the hardware structure and reduces the amount of data calculation. The full array scanning and result fusion can achieve millisecond-level response, meeting the requirements for real-time use. Moreover, all modules of the device use commercially available common components and the connection method is a conventional technical means, making the manufacturing cost controllable and possessing good prospects for popularization in civilian use and industrial application.

[0054] The technical solution of this invention has good scalability. The basic home version can be upgraded to a high-precision version suitable for medical aesthetic institutions by simply adjusting the electrode array density, increasing the number of excitation frequencies, and reducing the spatial matching threshold. This achieves complete fitting of the four parameters of the Cole-Cole model, further improving the spatial resolution of hair follicle positioning, and forming a complete technical system that can cover both consumer and professional medical applications, meeting the hair removal positioning needs of different usage scenarios. Attached Figure Description

[0055] Figure 1 This is a schematic diagram of the overall process of the present invention;

[0056] Figure 2 This is a flowchart illustrating step S3 of the present invention;

[0057] Figure 3 This is a flowchart illustrating step S4 of the present invention;

[0058] Figure 4 This is a flowchart illustrating step S5 of the present invention;

[0059] Figure 5 This is a schematic diagram of the visual-impedance fusion localization strategy in step S5 of the present invention;

[0060] Figure 6 This is a schematic diagram of the four-electrode method principle of the present invention;

[0061] Figure 7 This is a schematic diagram of the microelectrode array layout in this invention;

[0062] Figure 8 yes Figure 7 The corresponding impedance distribution spectrum. Detailed Implementation

[0063] In the description of this invention, it should be noted that the terms "upper / lower", "front / rear", "inner / outer", "top / bottom", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limiting this invention.

[0064] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installed," "equipped with," and "connected," etc., should be interpreted broadly. For example, "connection" can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium; it can be a connection within the components of a compatible model. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances. Furthermore, unless otherwise specified, all components used in this application are commercially available components, and the connection between different components can be achieved through conventional technical means.

[0065] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0066] Example 1

[0067] This embodiment provides a home-use hair removal device and corresponding hair follicle positioning method based on multi-frequency skin electrical impedance scanning, which is suitable for daily home hair removal use scenarios, balancing positioning accuracy and ease of operation, while controlling equipment costs.

[0068] Home-use hair removal device structure

[0069] The hair removal device includes a microelectrode array module with 4×4=16 electrodes, an impedance measurement module, a visual inspection module, a control module, a laser control module, and a display and prompt module. These modules work together to achieve precise hair follicle positioning and laser hair removal. Specific structural parameters are as follows:

[0070] Microelectrode array module: Electrically connected to the impedance measurement module via a flexible circuit board, used for contact with the skin and transmission of electrical signals; wherein, the 4×4=16 microelectrode array has an electrode spacing of 1-3mm, an electrode diameter between 0.5-0.8mm, and the electrode material is gold-plated copper to ensure stable transmission of electrical signals;

[0071] Impedance Measurement Module: Integrates an AFE chip (model AD5933 or AD5940) and an analog multiplexer (model CD74HC4067), including a signal generation unit, a signal acquisition unit, and an electrode gating unit. The signal generation unit can generate 10kHz / 100kHz / 1MHz three-frequency sinusoidal AC excitation signals, with a constant current excitation current and amplitude ≤1mA, and a peak-to-peak excitation voltage Vpp<1V. The signal acquisition unit completes voltage signal acquisition and local impedance calculation. The electrode gating unit, composed of an analog multiplexer, is the core of the impedance measurement module's path control. Driven by MCU instructions, it can accurately and orderly select and connect specified electrode paths in the microelectrode array, realizing on-demand electrode switching and assembly. At the same time, it completes the separation of current injection and voltage acquisition paths in the four-electrode measurement method, eliminating electrode and skin contact impedance interference from the hardware level.

[0072] Control Module: Utilizing an MCU as the main control unit, it communicates with the impedance measurement module via an SPI bus and is electrically connected to the vision inspection module, laser control module, and display / prompt module, respectively, enabling coordinated control and data processing of each module. It receives impedance data from the impedance measurement module and hair follicle location data from the vision inspection module, constructs an impedance spectrum based on the impedance data, determines candidate hair follicle locations, and then performs spatial matching and fusion positioning between the candidate hair follicle locations and the detection results from the vision inspection module, outputting a fused set of hair follicle coordinates.

[0073] Visual detection module: Employs a miniature camera to acquire real-time images of the skin surface, detects hair follicle locations using image recognition algorithms, and outputs a visual hair follicle location set V_set;

[0074] Laser control module: Receives the fused hair follicle coordinate set output by the control module, controls the laser to perform precise irradiation, and performs irradiation with 70% of the standard irradiation energy for hair follicles determined only by impedance detection, so as to avoid excessive damage to normal skin tissue;

[0075] Display prompt module: LED indicator lights are used as prompt units to provide feedback on the skin contact status. When the skin contact is poor, the indicator light flashes to remind the user to adjust the fitting angle of the device.

[0076] The hair follicle localization method based on multi-frequency skin electrical impedance scanning, applied in the aforementioned home-use hair removal device, sequentially executes five steps: microelectrode array initialization and calibration, multi-frequency impedance signal acquisition, impedance spectrum construction and filtering, hair follicle candidate location detection, and visual and impedance fusion localization. Finally, the fused hair follicle coordinate set is output and sent to the laser control module for irradiation. Figures 1-4 As shown, the specific operation is as follows:

[0077] The S1 microelectrode array is initialized and calibrated to provide a reference for subsequent impedance measurements, eliminating interference from factors such as electrode failure and poor contact. It then sequentially performs power-on self-test, electrode connectivity detection, air baseline acquisition, and skin contact detection.

[0078] S101 Power-on Self-Test: After the device is powered on, the MCU sends an initialization command to the AFE chip (model AD5933 or AD5940) to configure register parameters such as gain, excitation frequency and sampling rate. The AFE chip performs self-calibration on the internal DDS (direct digital frequency synthesizer) and ADC to ensure the accuracy of signal generation and acquisition.

[0079] S102 Electrode Connectivity Detection: The MCU sequentially selects 16 electrodes through an analog multiplexer (model CD74HC4067), applies a 100kHz / 100μA test signal, and measures the impedance value from each electrode to the common reference electrode. If the open-circuit impedance of an electrode is >10MΩ, it is marked as a faulty electrode and skipped in subsequent calculations.

[0080] S103 Air Baseline Acquisition: When the device is not in contact with the skin, it completes a full array scan of the microelectrode array through the impedance measurement module and records the baseline impedance value Z_air of each electrode pair (two) in the air, that is, the impedance reference value of all electrode pairs in the air-free state.

[0081] S104 Skin Contact Detection: When the device is placed against the skin of a human arm, the impedance measurement module detects the impedance value of each electrode pair in real time. When more than 75% of the electrode pair impedance values ​​drop below 1 / 10 of Z_air, it is determined that the skin is in effective contact and the measurement mode is entered. Otherwise, the LED indicator of the display prompt module will flash to remind the user to adjust the contact angle.

[0082] S2 Multi-frequency Impedance Signal Acquisition

[0083] Based on a calibrated microelectrode array that achieves effective skin contact, a four-electrode measurement method was used to complete a full array impedance scan, acquiring local impedance data at three frequencies to eliminate interference between electrode and skin contact impedance. Specific steps include:

[0084] S201 excitation signal configuration: The MCU controls the AFE chip to generate sinusoidal AC excitation signals of 10kHz, 100kHz, and 1MHz in sequence. The excitation current is kept in constant current mode and the amplitude is ≤1mA (far below the human perception threshold of 5mA to ensure safety). The peak-to-peak value of the excitation voltage Vpp is <1V.

[0085] It should be noted that 10kHz, 100kHz, and 1MHz were chosen because the penetration depth and flow path of alternating current at different frequencies in skin tissue vary significantly. These differences can be used to capture the specific differences in electrical properties and tissue structure between the hair follicle area and the surrounding normal skin tissue: at the low frequency of 10kHz, the current mainly flows in the extracellular fluid and is sensitive to the water-containing area of ​​the hair follicle; at the mid frequency of 100kHz, some current can penetrate the cell membrane, thus reflecting differences in tissue structure; at the high frequency of 1MHz, the current can penetrate the cell membrane and obtain information from inside the cell.

[0086] The S202 four-electrode measurement method is performed as follows: Four adjacent electrodes are selected in the microelectrode array to form a measurement unit. The two electrodes on either side are current injection electrodes, applying an excitation current ≤1mA Iexc. The two electrodes in the middle are voltage acquisition electrodes, measuring the potential difference Vbc. Finally, the result is calculated according to the formula... Calculate the local impedance;

[0087] Specifically, such as Figure 6 As shown, the four-electrode method selects four adjacent electrodes, A, B, C, and D. Current is injected from A and flows out from D, and the potential difference Vbc is measured between B and C. Since no current flows through B and C, the interference of electrode-skin interface contact impedance is eliminated.

[0088] S203 full array scanning sequence: The electrode array is scanned in four-electrode groups in the horizontal, vertical and diagonal directions. Each group of four electrodes completes an impedance measurement once at three frequencies: 10kHz, 100kHz and 1MHz. The measurement time is between 0.5-1ms. The reason for scanning in the diagonal direction is to further improve the spatial resolution.

[0089] Specifically, such as Figure 6 and Figure 7As shown, the electrodes are numbered E1-E16 and arranged in 4 rows and 4 columns. The four-electrode group of E1-E2-E3-E4 is selected in the horizontal direction, and the four-electrode group of E1-E5-E9-E13 is selected in the vertical direction. Each of the three frequencies is scanned once.

[0090] S204 Data Recording: The MCU records the impedance data of the full array scan in real time, generating a dataset {Z(i,j,f_k)} (i,j are the electrode space coordinate indices, k=1,2,3 are the frequency indices, and f_k is the frequency value). The data is temporarily stored in the MCU's SRAM buffer. Each round of the 16-electrode array generates approximately 36-48 sets of measurement values ​​* 3 frequencies = 108-144 complex impedance data points (including the real part R and the imaginary part X). The total time for the full array scan is approximately 30-50ms.

[0091] S3 Impedance Map Construction and Filtering

[0092] The collected discrete impedance data are interpolated, smoothed, denoised, and model parameters are extracted to generate high-quality impedance feature maps, providing a reliable basis for hair follicle candidate location detection.

[0093] S301 Raw Data Interpolation: Since the four-electrode method measures the local impedance between adjacent electrodes, the measurement point is located in the middle of the electrode rather than at the electrode itself. For a 4x4 electrode array, 3x4=12 measurement points are obtained in the horizontal direction and 4x3=12 measurement points are obtained in the vertical direction. Bilinear interpolation is used to map the discrete measurement values ​​onto a regular grid to generate an impedance distribution spectrum Z(x,y); one spectrum is generated for each frequency, resulting in a total of 3 impedance spectra: Z10k(x,y), Z100k(x,y), and Z1M(x,y), which compensates for the insufficient spatial resolution of discrete measurements.

[0094] S302 Gaussian Smoothing Denoising: Apply a two-dimensional Gaussian filter to each impedance spectrum: Where G is the Gaussian kernel function, sigma = 0.5-1 times the electrode spacing (for an array with a 2mm spacing, sigma is 1-2mm), to eliminate random noise caused by poor contact of a single electrode or local skin surface conditions (sweat, uneven thickness of the stratum corneum), while retaining the effective signal at the hair follicle scale (diameter about 0.1-0.3mm, but its electrical influence range is about 1-2mm); and obtain the smoothed low-frequency, mid-frequency, and high-frequency impedance characteristic spectra Z10k_smooth(x,y), Z100k_smooth(x,y), Z1Mk_smooth(x,y);

[0095] S303 Cole-Cole Model Parameter Extraction: For each spatial position (x, y), use the impedance characteristic spectra at three frequencies to fit the Cole-Cole model: , to distinguish between the follicular area and the non-follicular area; where R_0 is the low-frequency impedance, R_inf is the high-frequency impedance, tau is the characteristic time constant, and alpha is the dispersion coefficient (0 < alpha <= 1); there are systematic differences in the Cole-Cole parameters between the follicular area and the area without follicles: the tau value in the follicular area is smaller (the dielectric relaxation of the water-containing tissue is faster), and the alpha value is closer to 1 (the tissue structure is more uniform).

[0096] It should be noted that 3 parameters (R_0, R_inf, tau) can be solved with 3 frequency points, and alpha takes an empirical fixed value of 0.8 when the number of data points is insufficient.

[0097] S4 Detection of Candidate Follicle Positions

[0098] Taking the smoothed low-frequency impedance characteristic spectrum Z10k_smooth(x, y) as the core (the impedance contrast between the follicle and the surrounding tissue is the largest at low frequencies), successively perform local minimum search, threshold screening, sub-pixel localization refinement, and multi-frequency cross-validation to obtain a set of follicle candidate positions with confidence:

[0099] S401 Local Minimum Search: Traverse each pixel point in the low-frequency impedance spectrum Z10k_smooth(x, y) after Gaussian smoothing and denoising one by one; for each pixel point currently traversed, obtain the impedance value of the pixel point itself and the impedance values of the 8 adjacent pixel points above, below, left, right, upper left, lower left, upper right, and lower right of the pixel point respectively; compare the impedance value of the current pixel point with the impedance values of the above 8 adjacent pixel points one by one; if the impedance value of the current pixel point is strictly less than the impedance values of all the above 8 adjacent pixel points, mark the pixel point as a local minimum candidate point;

[0100] S402 Threshold Screening: Calculate the ratio Ratio of the impedance value of each local minimum candidate point to the average impedance value of its 8 adjacent pixel points, and screen out the candidate points with Ratio < 0.85 (that is, the impedance value is at least 15% lower than the surrounding), and mark them as follicle candidate positions; The selection basis of the threshold 0.85: Based on the impedance characteristics measurement of different parts of the human skin, the impedance difference between the follicular area and the surrounding skin is within the range of 15% - 40% (varying with the follicle depth, thickness, and water content state). Taking 15% as the lowest detection threshold can control the false detection rate while ensuring the sensitivity;

[0101] S403 Sub-pixel Localization Refinement:

[0102] For candidate hair follicle locations selected through threshold filtering, sub-pixel localization is performed using two-dimensional parabolic fitting: a two-dimensional quadratic function is fitted within a 3x3 window consisting of the candidate point and its eight neighboring pixels. The precise location of the minimum value is: This step can improve the spatial resolution by about 2-3 times (to the 0.5-1mm level) from the electrode spacing (1-3mm), allowing for more accurate positioning of the hair follicle center;

[0103] It's important to note that the reason for sub-pixel positioning refinement is that the electrode spacing is 1–3 mm, while the candidate points detected by the electrodes can only be accurate to the scale of the electrode grid, such as falling between two electrodes or near the center of an electrode grid. However, the hair follicle itself is only 0.1–0.3 mm in size, and its true center must be within an even smaller range of the electrode grid. Therefore, relying solely on electrode grid positioning will result in an error of several millimeters, affecting the accuracy of laser targeting.

[0104] S404 Multi-Frequency Cross-Validation: Steps S401-S403 are repeated for impedance characteristic spectra in the low-frequency, mid-frequency, and high-frequency bands to obtain the locations of candidate hair follicle points in each band. Low frequency (10kHz): The impedance contrast between the hair follicle and the surrounding skin is the largest, making it the easiest to find the hair follicle with less interference. Mid-frequency (100kHz) and high-frequency (1MHz): These do not only correspond to tissue, but also contain hair follicle signals as well as other interference signals such as skin and connective tissue. However, the impedance contrast of the hair follicle is not as obvious as that of the low frequency, and it is difficult to distinguish between hair follicle and tissue when looking at the mid-frequency and high-frequency bands alone. The candidate points detected by the impedance characteristic spectra of the three frequencies are cross-compared. Candidate points detected in at least two frequency impedance characteristic spectra and whose spatial distance between the two candidate points is less than one electrode spacing are marked as high confidence. Candidate points detected in only one frequency spectrum are marked as medium confidence. The hair follicle candidate set I_set={(x_i,y_i,confidence_i)} is output, where confidence takes the value of {high, medium}.

[0105] S5 vision-impedance fusion positioning and output to the laser control module

[0106] The candidate set of impedance hair follicle locations is aligned with the visual set of hair follicle locations in terms of coordinate system, spatially matched, and the results are fused to generate a standardized fused hair follicle coordinate set, which is then output to the laser control module for laser irradiation.

[0107] S501 Coordinate System Alignment: During the device manufacturing stage, the relative positional relationship between the camera's field of view and the microelectrode array is calibrated to obtain the affine transformation matrix M_calib (containing translation, rotation, and scaling parameters). During runtime, the corresponding coordinates of the hair follicle candidate location set obtained from impedance detection in step S4 are transformed to image coordinates. imp is impedance detection, and img is image detection;

[0108] S502 Spatial Matching: The impedance hair follicle candidate location set I_set, which is transformed into image coordinates, is spatially adjacent to the visual hair follicle candidate location set V_set. The matching threshold is 1mm. For each candidate point in I_set, the nearest visual hair follicle point is searched in V_set. When the distance between the two points is less than the matching threshold, they are determined to be a dual-modal detection matching pair of the same hair follicle. After matching, three types of results are obtained: dual-modal matching, visual detection only, and impedance detection only.

[0109] S503 Result Fusion: (e.g., ...) Figure 5 As shown, the three types of results are processed separately:

[0110] Type A bimodal matching: First, the high and medium qualitative confidence scores marked in S4 are mapped to quantitative values ​​(high confidence = 0.7, medium confidence = 0.5). The visual detection module outputs a quantitative confidence score of 0~1. Then, coordinates and confidence scores are fused uniformly. The formulas for calculating fused coordinates and fused confidence scores are as follows:

[0111] , ,

[0112] The fusion confidence bimodal reward is 0.1, with a maximum of 1.0; the weight calculation in the fusion coordinate formula is as follows:

[0113] ,

[0114] The laser irradiation energy is the standard energy; V_j and I_i are the coordinates of the hair follicle candidate points output by the vision and impedance modules; w_v / w_i are the fusion weights of the corresponding coordinates; conf_v / conf_i are the quantitative confidence values ​​of the vision and impedance modules, which are the core basis for calculating the weights and fusion confidence.

[0115] Type B Visual Inspection Only: The visual results remain unchanged, the coordinates and confidence scores are used from the visual module output, and the laser irradiation energy is 100% of the standard energy;

[0116] Type C Impedance Detection Only: The impedance detection results are adopted, and the confidence level is set to 0.5-0.7 (based on the confidence level of the impedance hair follicle candidate point: 0.7 for "high" and 0.5 for "medium"); medium energy (70% of standard energy) is used during laser irradiation to avoid over-irradiation of false detection locations;

[0117] The S504 outputs to the laser control module: The MCU generates a fused hair follicle set F_set{(x_k,y_k,conf_k,energy_k)}, where x_k and y_k are the coordinates of the k-th hair follicle, conf_k is the confidence level of the k-th hair follicle, and energy_k is the laser irradiation energy of the k-th hair follicle. F_set is sorted by confidence level from highest to lowest and then transmitted to the laser control module. After receiving the fused hair follicle set F_set, the laser control module performs precise laser irradiation on the hair follicles according to their coordinates, achieving a safe and efficient home hair removal effect. Simultaneously, the time taken for a single fusion calculation is less than 5ms, without affecting real-time performance.

[0118] Example 2

[0119] To further improve the spatial resolution and detection accuracy of hair follicle localization and meet the high-precision hair removal needs of professional scenarios such as medical aesthetic institutions, this invention provides a high-precision preferred implementation method (professional version) based on the above-mentioned home-use version technical solution. The professional version optimizes and upgrades core parameters such as electrode array specifications, number of excitation frequencies, and positioning matching thresholds. The overall module composition is consistent with the home-use version, still including a microelectrode array module, impedance measurement module, visual detection module, control module, laser control module, and display prompt module. The core adaptation parameters and methods are optimized as follows:

[0120] High-precision optimized hair removal device structure

[0121] Microelectrode array module: Employs a high-density array of 32×32=1024 electrodes with an electrode spacing of 0.5-1mm and an electrode diameter of 0.3-0.5mm. Platinum is the preferred electrode material to enhance corrosion resistance and electrical signal transmission accuracy. Impedance measurement module: Generates five-frequency sinusoidal AC excitation signals of 10kHz / 50kHz / 100kHz / 500kHz / 1MHz, with a constant excitation current of ≤1mA. More frequency points enable complete fitting of the four parameters of the Cole-Cole model, improving the distinction between follicular and non-follicular areas. Spatial matching threshold: The spatial matching threshold for visual-impedance fusion positioning is adjusted to 0.5mm to meet the high-precision positioning requirements of the high-density electrode array. The functions and connections of the remaining modules are consistent with the home version of Example 1 and will not be repeated here.

[0122] Therapeutic version of hair follicle localization and hair removal methods

[0123] The hair follicle localization method sequentially executes five steps, S1 to S5. The overall logical steps are completely consistent with the home-use version of Example 1. The core difference lies only in the steps related to frequency and fitting, and the specific optimizations are as follows:

[0124] S2 Multi-Frequency Impedance Signal Acquisition Stage: The excitation frequency is configured with five frequencies: 10kHz / 50kHz / 100kHz / 500kHz / 1MHz. Each group of four electrodes completes an impedance measurement once at each of the five frequencies. The single measurement time is still 0.5-1ms. Impedance data from more frequencies can more comprehensively reflect the differences in electrical properties of different skin tissues.

[0125] S3 Cole-Cole model parameter extraction stage: Using impedance data at five frequency points, the four parameters (R_0, R_inf, tau, alpha) of the Cole-Cole model are fully fitted to each spatial location (x,y). There is no need to take a fixed empirical value for the dispersion coefficient alpha, which further improves the parameter discrimination between the hair follicle region and the non-hair follicle region.

[0126] S5 Spatial Matching Stage: The spatial proximity matching threshold of visual-impedance fusion positioning is adjusted to 0.5mm to meet the high-precision positioning requirements of high-density electrode arrays and improve the accuracy of hair follicle positioning.

[0127] This high-precision professional-grade laser hair removal system achieves ultra-precise positioning of hair follicles through optimization of high-density electrode arrays, multi-frequency excitation signals, and high-precision matching thresholds. Combined with precise irradiation by the laser control module, it can achieve medical-grade, efficient, and safe hair removal results, adapting to professional clinical hair removal needs of different skin types and hair follicle conditions.

[0128] General Instructions

[0129] The hair removal device and hair follicle positioning method of the present invention make up for the shortcomings of single visual detection being easily interfered with by skin surface conditions and single impedance detection having insufficient resolution by using a dual-modal fusion positioning method that combines multi-frequency skin impedance scanning with visual detection. At the same time, a four-electrode measurement method is used to eliminate the interference of electrode-skin contact impedance. Gaussian smoothing and noise reduction, sub-pixel positioning refinement and other steps are used to improve positioning accuracy. Two specifications are designed, a home version and a professional version, to meet the needs of daily home use and high-precision use in medical aesthetic institutions, respectively.

[0130] During the laser irradiation stage, hair follicles identified solely by impedance detection are irradiated with 70% standard energy. This ensures effective hair removal while minimizing damage to normal skin tissue, thus enhancing the device's safety. The calculation time for each step is strictly controlled to guarantee real-time detection and hair removal, ensuring a superior user experience.

[0131] Furthermore, all modules in this invention use commercially available common components, and the connection between different components is achieved through conventional technical means. The manufacturing cost of the device is controllable, and it has good prospects for industrial application.

[0132] Although embodiments of the present invention have been described in the specification, these embodiments are merely illustrative and should not be construed as limiting the scope of protection of the present invention. Various omissions, substitutions, and modifications made without departing from the spirit of the present invention should be included within the scope of protection of the present invention.

Claims

1. A hair removal device based on multi-frequency skin electrical impedance scanning for follicle localization, characterized in that, It includes a microelectrode array module, an impedance measurement module, a vision inspection module, a control module, a laser control module, and a display and prompting module; The microelectrode array module is electrically connected to the impedance measurement module via a flexible circuit board, and is used to contact the skin and transmit electrical signals. The impedance measurement module is communicatively connected to the control module and is used to generate AC excitation signals, perform electrode gating, and collect and calculate local impedance data of the skin. The visual detection module is communicatively connected to the control module and is used to acquire skin images and detect hair follicle locations, and output a set of visual hair follicle locations. The control module communicates with each module to achieve coordinated control. It receives impedance data from the impedance measurement module and hair follicle location data from the visual detection module. Based on the impedance data, it constructs an impedance spectrum and determines the candidate hair follicle location. Then, it performs spatial matching and fusion positioning between the candidate hair follicle location and the detection result of the visual detection module, and outputs the fused hair follicle coordinate set. The laser control module is communicatively connected to the control module and is used to perform laser irradiation based on the fused hair follicle coordinate set output by the control module. The display and prompt module is electrically connected to the control module and is used to issue skin contact status prompts and device working status prompts.

2. The hair removal device based on multi-frequency skin electrical impedance scanning for follicle localization as described in claim 1, characterized in that, The microelectrode array module includes a 4×4=16 microelectrode array with an electrode spacing between 1-3 mm and a diameter between 0.5-0.8 mm; the electrode material of the microelectrode array module is gold-plated copper.

3. The hair removal device based on multi-frequency skin electrical impedance scanning for hair follicle localization as described in claim 1, characterized in that, The impedance measurement module includes a signal generation unit, a signal acquisition unit, and an electrode selection and connectivity unit. The signal generation unit generates a multi-frequency sinusoidal AC excitation signal. The signal acquisition unit acquires voltage signals and combines them with the excitation signal to perform local impedance calculations. The electrode selection and connectivity unit enables electrode selection and path connectivity for the microelectrode array. The excitation current output by the signal generation unit is ≤1mA, and the excitation frequency of the multi-frequency sinusoidal AC excitation signal generated by the signal generation unit is 10k / 100k / 1M Hz.

4. The hair removal device based on multi-frequency skin electrical impedance scanning for follicle localization as described in claim 3, characterized in that, The signal generation unit and signal acquisition unit are integrated into the AFE chip, and the electrode selection and connection unit is composed of an analog multiplexer; the control module is an MCU, and the MCU communicates with the AFE chip and analog multiplexer of the impedance measurement module through the SPI bus; the impedance measurement module allocates the electrode path through the electrode selection and connection unit to realize the separation of current injection and voltage acquisition path in the four-electrode measurement method and eliminate the interference of electrode-skin contact impedance.

5. A hair follicle localization method based on multi-frequency skin electrical impedance scanning, applied to the hair removal device according to any one of claims 1 to 4, characterized in that, include: S1 Microelectrode Array Initialization and Calibration: The control module completes the power-on self-test, electrode connectivity detection, air baseline acquisition, and skin contact detection of the microelectrode array, providing a reference for subsequent impedance measurements; S2 Multi-frequency impedance signal acquisition: For the microelectrode array that has been calibrated and achieved effective skin contact in step S1, a multi-frequency AC excitation signal with preset parameters is applied through the impedance measurement module, and the four-electrode measurement method is used to complete the full array impedance scan and acquire local impedance data at multiple frequencies. S3 Impedance Map Construction and Filtering: Based on the multi-frequency local impedance data collected in step S2, an impedance distribution map is generated by interpolation, and then Gaussian smoothing and noise reduction are performed to obtain low-frequency, mid-frequency and high-frequency impedance characteristic maps; Cole-Cole model parameters are extracted to obtain high-quality impedance characteristic maps. S4 Hair follicle candidate location detection: On the low-frequency impedance feature map in step S3, local minimum search, threshold screening and sub-pixel refinement are completed in sequence, and multi-frequency cross-validation is performed to obtain a set of hair follicle candidate locations with confidence. S5 integrates visual and impedance detection results to achieve precise hair follicle localization and output: The set of candidate hair follicle positions obtained in step S4 is aligned with the set of hair follicle positions output by the visual detection module, spatially matched and the results are fused, and the fused hair follicle coordinate set is output to the laser control module to perform laser irradiation.

6. The hair follicle localization method based on multi-frequency skin electrical impedance scanning as described in claim 5, characterized in that, Step S1 includes the following steps: S101 Power-on self-test: After the device is powered on, the control module sends an initialization command to the impedance measurement module to configure the gain, excitation frequency, and sampling rate parameters of the impedance measurement. The impedance measurement module performs self-calibration on its internal signal generation unit and signal acquisition unit. S102 Electrode Connectivity Detection: The control module sequentially selects and connects each electrode of the microelectrode array through the electrode selection and connection unit, applies a 100kHz / 100μA test signal, and measures the impedance value from each electrode to the common reference electrode. If the open-circuit impedance of the electrode is >10MΩ, it is marked as a faulty electrode and skipped in subsequent calculations. S103 Air Baseline Acquisition: When the device is not in contact with the skin, it completes a full array scan of the microelectrode array through the impedance measurement module and records the baseline impedance value Z_air of each electrode pair in the air; S104 Skin Contact Detection: After the device is attached to the skin, the impedance measurement module detects the electrode impedance value in real time. When more than 75% of the electrode pair impedance values ​​drop to less than 1 / 10 of Z_air, it is determined that the skin is in effective contact and the measurement mode is entered. Otherwise, the display prompt module will issue a contact adjustment prompt.

7. The hair follicle localization method based on multi-frequency skin electrical impedance scanning as described in claim 6, characterized in that, Step S2 includes the following steps: S201 Excitation Signal Configuration: The control module controls the signal generation unit to generate 10kHz, 100kHz, and 1MHz sinusoidal excitation signals. The excitation current is in constant current mode with an amplitude ≤1mA, and the peak-to-peak excitation voltage Vpp <1V. S202 Four-Electrode Measurement Method: Four adjacent electrodes are selected in the microelectrode array to form a measurement unit. The two electrodes on either side are current injection electrodes, applying an excitation current Iexc. The two electrodes in the middle are voltage acquisition electrodes, measuring the potential difference Vbc. Finally, according to… Calculate the local impedance; S203 full array scanning sequence: The microelectrode array is sequentially selected and connected in the horizontal, vertical and diagonal directions by the electrode selection and connection unit. Each group of four electrodes completes an impedance measurement once at three frequencies, with a single measurement time of 0.5-1ms. S204 Data Recording: The control module records the dataset {Z(i,j,f_k)} of the full array scan, where i and j are the electrode space coordinate indices, k=1,2,3 are the frequency indices, and f_k is the frequency value. The dataset is temporarily stored in the storage unit of the control module.

8. The hair follicle localization method based on multi-frequency skin electrical impedance scanning as described in claim 7, characterized in that, Step S3 includes the following steps: S301 raw data interpolation: Bilinear interpolation is used to map the discrete local impedance measurements onto a regular grid, generating impedance distribution spectra Z10k(x,y), Z100k(x,y), and Z1M(x,y) for the three frequencies of 10kHz, 100kHz, and 1MHz, respectively. S302 Gaussian Smoothing Denoising: A two-dimensional Gaussian filter is applied to each impedance distribution spectrum. The sigma of the Gaussian kernel function is set to 0.5-1 times the electrode spacing to eliminate random noise and retain the effective signal at the hair follicle scale, resulting in low-frequency impedance characteristic spectrum Z10k_smooth(x,y), mid-frequency impedance characteristic spectrum Z100k_smooth(x,y) and high-frequency impedance characteristic spectrum Z1Mk_smooth(x,y). S303Cole-Cole parameter extraction: For each spatial location (x,y), the Cole-Cole model is fitted using impedance characteristic spectra at three frequencies to extract low-frequency impedance R_0, high-frequency impedance R_inf, and characteristic time constant tau. The dispersion coefficient alpha is set to a fixed empirical value of 0.8 to distinguish between hair follicle regions and non-hair follicle regions. The tau value of the hair follicle region is smaller than that of the non-hair follicle region, and the alpha value is closer to 1.

9. The hair follicle localization method based on multi-frequency skin electrical impedance scanning as described in claim 8, characterized in that, Step S4 includes the following steps: S401 Local Minimum Search: On the smoothed low-frequency impedance feature map Z10k_smooth(x,y), traverse each pixel and compare its impedance value with the impedance values ​​of its 8 neighboring pixels. Pixels with impedance values ​​less than those of the 8 neighboring pixels are marked as candidate local minima. S402 Threshold Screening: Calculate the ratio Ratio of the impedance value of each local minimum candidate point to the average impedance value of its 8 neighboring pixels. When Ratio < 0.85, it is confirmed as a candidate location for hair follicles. S403 Subpixel Localization Refinement: For candidate hair follicle locations that have passed the threshold screening, subpixel localization is performed using two-dimensional parabolic fitting within a 3×3 window centered on the follicle and containing its 8 neighboring pixels, in order to improve the spatial resolution of hair follicle localization. S404 Multi-frequency Cross-validation: Repeat steps S401-S403 for impedance characteristic spectra of low-frequency, mid-frequency, and high-frequency bands respectively to obtain the candidate hair follicle locations in each band; cross-compare the candidate points detected by the three frequency impedance characteristic spectra; mark the candidate points detected in at least two frequency impedance characteristic spectra and the spatial distance between the two candidate points is less than one electrode spacing as high confidence; mark the candidate points detected in only one frequency spectrum as medium confidence; output the set of hair follicle candidate locations I_set={(x_i,y_i,confidence_i)}, where confidence is either high or medium.

10. The hair follicle localization method based on multi-frequency skin electrical impedance scanning as described in claim 9, characterized in that, Step S5 includes: S501 Coordinate System Alignment: During the device manufacturing stage, the relative positional relationship between the camera's field of view and the microelectrode array is calibrated to obtain the affine transformation matrix M_calib. During runtime, the corresponding coordinates of the hair follicle candidate location set obtained from impedance detection in step S4 are transformed to image coordinates. imp is impedance detection, and img is image detection; S502 Spatial Matching: The impedance hair follicle candidate location set I_set, which is transformed into image coordinates, is spatially adjacent to the visual hair follicle candidate location set V_set. The matching threshold is 1mm. For each candidate point in I_set, the nearest visual hair follicle point is searched in V_set. When the distance between the two points is less than the matching threshold, they are determined to be a dual-modal detection matching pair of the same hair follicle. After matching, three types of results are obtained: dual-modal matching, visual detection only, and impedance detection only. S503 Result Fusion: For bimodal matching results, the qualitative confidence scores "high" and "medium" marked in S4 are first mapped to quantitative values, with high confidence score = 0.7 and medium confidence score = 0.

5. The visual detection module outputs a quantitative confidence score of 0~1. Then, coordinates and confidence scores are fused. The formulas for calculating fused coordinates and fused confidence scores are as follows: ， ， The fusion confidence bimodal reward is 0.1, with a maximum of 1.0; the weight calculation in the fusion coordinate formula is as follows: ， The laser irradiation energy is the standard energy; V_j and I_i are the coordinates of the hair follicle candidate points output by the vision and impedance modules; w_v / w_i are the fusion weights of the corresponding coordinates; conf_v / conf_i are the quantitative confidence values ​​of the vision and impedance modules. For visual inspection results only, the output coordinates and confidence level of the visual inspection module are directly used, and the laser irradiation energy is the standard energy. For impedance inspection results only, the confidence level is set to 0.5-0.7 according to its confidence level. Based on the confidence level of the impedance candidate points: 0.7 for "high" and 0.5 for "medium". The laser irradiation energy is 70% of the standard energy. The S504 outputs to the laser control module: The control module generates a fused hair follicle set: F_set{(x_k,y_k,conf_k,energy_k)}, where x_k and y_k are the coordinates of the k-th hair follicle, conf_k is the confidence level of the k-th hair follicle, and energy_k is the laser irradiation energy of the k-th hair follicle; and then sorts the hair follicles by confidence level from high to low before transmitting them to the laser control module.

Images