Comprehensive analysis method for hair types

Abstract

The invention provides a hair type comprehensive analysis method, and relates to the technical field of hair analysis, and the method comprises the following steps: collecting a hair sample, and pretreating the surface of the hair sample; performing electrical data measurement on the preprocessed hair sample by using electrical measurement equipment to obtain an electrical measurement signal; the electrical measurement signal comprises a conductivity signal, a capacitance characteristic signal and an impedance spectrum signal; performing signal de-noising and normalization processing on the electrical measurement signal to obtain a pre-processed signal; performing feature extraction on the preprocessed signal to obtain a sample feature parameter; inputting the sample characteristic parameters into an initial neural network for training to obtain a hair type analysis model; and inputting the electrical measurement signal of the to-be-measured hair into the hair type analysis model to obtain an analysis result. Based on an electrical data processing technology, detection and analysis of a hair sample can be completed within a few minutes, and the efficiency of customs inspection is remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of hair analysis, and in particular to a comprehensive analysis method for hair types. Background Art

[0002] During customs inspections, hair sample analysis is an important means of identifying individuals, detecting prohibited substances (such as drugs and heavy metals), and assessing health conditions. Traditional hair analysis methods rely primarily on chemical testing and microscopic observation, which have the following problems:

[0003] Low efficiency: Traditional methods require a long time to process samples, making it difficult to meet customs requirements for rapid clearance.

[0004] Complex equipment: Chemical testing usually requires expensive and complex laboratory equipment, which is not suitable for use in high-traffic scenarios such as customs.

[0005] Singleness: Existing methods can usually only detect one aspect of hair characteristics and cannot achieve comprehensive analysis.

[0006] In recent years, analytical methods based on electrical properties have gained increasing attention. As a biomaterial, hair's electrical properties (such as conductivity, capacitance, and impedance spectrum) can reflect its physical structure, chemical composition, and contaminant content. Therefore, developing a comprehensive hair type analysis method based on electrical data processing could enable rapid and accurate hair testing, meeting customs inspection requirements. Summary of the Invention

[0007] In order to overcome the shortcomings of the existing technology, the purpose of the present invention is to provide a comprehensive analysis method for hair types. Based on electrical data processing technology, it can complete the detection and analysis of hair samples in a few minutes, significantly improving the efficiency of customs inspection. Through electrical characteristic analysis, it can simultaneously detect the physical properties, health status and pollutant content of hair, and realize comprehensive analysis.

[0008] To achieve the above object, the present invention provides the following solutions:

[0009] A comprehensive hair type analysis method comprising:

[0010] collecting a hair sample and pre-treating the surface of the hair sample;

[0011] Using electrical measurement equipment to measure electrical data of the pre-processed hair sample to obtain electrical measurement signals; the electrical measurement signals include conductivity signals, capacitance characteristic signals, and impedance spectrum signals;

[0012] performing signal denoising and normalization processing on the electrical measurement signal to obtain a preprocessed signal;

[0013] Performing feature extraction on the preprocessed signal to obtain sample feature parameters;

[0014] Inputting the sample characteristic parameters into an initial neural network for training to obtain a hair type analysis model;

[0015] The electrical measurement signal of the hair to be tested is input into the hair type analysis model to obtain the analysis result.

[0016] Preferably, the surface of the hair sample is pre-treated, comprising:

[0017] performing a cleaning process on the hair sample;

[0018] The cleaned hair samples were dried to eliminate the interference of surface impurities on the electrical property measurement.

[0019] Preferably, performing signal denoising and normalization processing on the electrical measurement signal to obtain a preprocessed signal comprises:

[0020] performing a translation on the electrical measurement signal to obtain a translated signal;

[0021] performing wavelet decomposition on the shifted signal to obtain a plurality of wavelet coefficients;

[0022] Determine the filtering threshold according to the decomposition scale and length of the signal;

[0023] Constructing a denoising function according to the filtering threshold;

[0024] Using the denoising function to remove noise from the shifted signal to obtain a denoised signal;

[0025] Normalization is performed on the denoised signal to obtain the preprocessed signal.

[0026] Preferably, the expression of the filtering threshold is:

[0027]

[0028] Among them, λ is the filtering threshold, j is the decomposition scale of the shifted signal, d j is the decomposition of the wavelet coefficients with a scale of j, N represents the length of the shifted signal, and median represents the median operation.

[0029] Preferably, removing noise from the shifted signal using the denoising function to obtain a denoised signal comprises:

[0030] The corresponding wavelet coefficients are removed using the denoising function to obtain a smoothed signal; wherein the denoising function is:

[0031]

[0032] Among them, X represents the wavelet coefficient, n represents the adjustment coefficient;

[0033] Performing inverse translation on each of the smoothed signals to obtain an inverse translated signal;

[0034] An average of multiple inverse-shifted signals is obtained to obtain the denoised signal.

[0035] Preferably, the sample characteristic parameters include: conductivity characteristic sub-parameters, capacitance characteristic sub-parameters, impedance spectrum characteristic sub-parameters, time domain characteristic sub-parameters and frequency domain characteristic sub-parameters; the conductivity characteristic sub-parameters include: DC conductivity, AC conductivity and conductivity change rate; the capacitance characteristic sub-parameters include: static capacitance value, dynamic capacitance value and dielectric loss factor; the impedance spectrum characteristic sub-parameters include: impedance amplitude, impedance phase angle, impedance frequency response curve and characteristic frequency points; the time domain characteristic sub-parameters include: rise time and fall time of the electrical signal, signal attenuation rate; the frequency domain characteristic sub-parameters include: spectrum energy distribution and frequency bandwidth.

[0036] Preferably, the sample characteristic parameters are input into an initial neural network for training to obtain a hair type analysis model, including:

[0037] Acquire a preset conductivity characteristic sub-parameter dataset, a capacitance characteristic sub-parameter dataset, an impedance spectrum characteristic sub-parameter dataset, a time domain characteristic sub-parameter dataset, and a frequency domain characteristic sub-parameter dataset;

[0038] The initial neural network is trained according to the conductivity characteristic sub-parameter data set, the capacitance characteristic sub-parameter data set, the impedance spectrum characteristic sub-parameter data set, the time domain characteristic sub-parameter data set, and the frequency domain characteristic sub-parameter data set to obtain a trained first classifier, a second classifier, a third classifier, a fourth classifier, and a fifth classifier;

[0039] Cascade the trained classifiers to obtain a classification network;

[0040] The trained LSTM neural network is connected to the classification network to obtain the hair type analysis model.

[0041] Preferably, the analysis results include: hair type, health condition, and whether the hair contains prohibited substances.

[0042] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0043] The present invention provides a comprehensive analysis method for hair types, comprising: collecting a hair sample and pre-treating the surface of the hair sample; using an electrical measurement device to perform electrical data measurement on the pre-treated hair sample to obtain an electrical measurement signal; the electrical measurement signal includes a conductivity signal, a capacitance characteristic signal, and an impedance spectrum signal; performing signal denoising and normalization on the electrical measurement signal to obtain a pre-processed signal; performing feature extraction on the pre-processed signal to obtain sample characteristic parameters; inputting the sample characteristic parameters into an initial neural network for training to obtain a hair type analysis model; and inputting the electrical measurement signal of the hair to be tested into the hair type analysis model to obtain an analysis result. Based on electrical data processing technology, the present invention can complete the detection and analysis of hair samples within a few minutes, significantly improving the efficiency of customs inspections. Through electrical characteristic analysis, the physical properties, health status, and pollutant content of the hair can be simultaneously detected to achieve a comprehensive analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0045] Figure 1 A flowchart of a method provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0046] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0047] The purpose of the present invention is to provide a comprehensive analysis method for hair types. Based on electrical data processing technology, it can complete the detection and analysis of hair samples within a few minutes, significantly improving the efficiency of customs inspection. Through electrical characteristic analysis, it can simultaneously detect the physical properties, health status and pollutant content of hair, and realize comprehensive analysis.

[0048] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0049] Figure 1 A flow chart of the method provided in the embodiment of the present invention is shown in FIG. Figure 1As shown, the present invention provides a comprehensive analysis method for hair types, comprising:

[0050] Step 100: Collect a hair sample and pre-treat the surface of the hair sample;

[0051] Step 200: Using electrical measurement equipment to measure electrical data of the pre-processed hair sample to obtain electrical measurement signals; the electrical measurement signals include conductivity signals, capacitance characteristic signals, and impedance spectrum signals;

[0052] Step 300: performing signal denoising and normalization processing on the electrical measurement signal to obtain a preprocessed signal;

[0053] Step 400: extracting features from the preprocessed signal to obtain sample feature parameters;

[0054] Step 500: Inputting sample feature parameters into the initial neural network for training to obtain a hair type analysis model;

[0055] Step 600: Input the electrical measurement signal of the hair to be measured into the hair type analysis model to obtain the analysis result.

[0056] Preferably, the surface of the hair sample is pre-treated, comprising:

[0057] performing a cleaning process on the hair sample;

[0058] The cleaned hair samples were dried to eliminate the interference of surface impurities on the electrical property measurement.

[0059] Specifically, in this embodiment, hair samples are cleaned by rinsing with deionized water or anhydrous ethanol. First, the collected hair sample is placed in a clean container and gently rinsed with deionized water to remove dust, grease, and other soluble impurities from the hair surface. If there may be oily or chemical residues on the hair surface, it can be further soaked or wiped with anhydrous ethanol to ensure surface cleanliness. The entire cleaning process should avoid mechanical damage to the hair sample to ensure its structural integrity.

[0060] After cleaning, hair samples need to be dried to eliminate any interference from residual moisture or solvents on electrical property measurements. Place the cleaned hair sample in a clean, dry environment and dry it using a low-temperature blower dryer or vacuum drying equipment to ensure the hair surface is completely dry and free from external contamination. The temperature should be strictly controlled during the drying process (e.g., no more than 40°C) to prevent high temperatures from affecting the physical and chemical properties of the hair, thereby ensuring the accuracy of subsequent electrical measurements.

[0061] Specifically, step 300 of this embodiment includes:

[0062] The hair sample is secured to the measuring device's test platform, ensuring full contact between the hair and the electrodes. This platform typically consists of a highly sensitive electrode array capable of applying electrical signals at varying frequencies to the hair sample. Adjusting the contact pressure and position of the electrodes ensures stable and consistent signal transmission during measurement while avoiding mechanical damage to the hair sample.

[0063] During conductivity signal measurement, the device applies a DC or low-frequency AC current to the hair sample and calculates its conductivity by measuring the magnitude of the current and the change in voltage through the hair. The conductivity can reflect the hair's moisture content, internal structure, and the distribution of conductive substances (such as minerals or contaminants). To improve measurement accuracy, the device automatically calibrates background noise and repeats the measurement multiple times to obtain an average value.

[0064] To measure capacitance and impedance spectrum signals, the device applies multi-frequency AC signals to the hair sample and records the hair's response to these signals at different frequencies. By analyzing the curves of the hair's capacitance, impedance amplitude, and phase angle as they change with frequency, the device reveals the hair's dielectric and polarization properties, as well as the complexity of its internal structure. During the measurement process, the device automatically collects electrical data within the frequency range and stores it as a digital signal for subsequent data processing and feature extraction.

[0065] Preferably, performing signal denoising and normalization processing on the electrical measurement signal to obtain a preprocessed signal comprises:

[0066] performing a translation on the electrical measurement signal to obtain a translated signal;

[0067] performing wavelet decomposition on the shifted signal to obtain a plurality of wavelet coefficients;

[0068] Determine the filtering threshold according to the decomposition scale and length of the signal;

[0069] Constructing a denoising function according to the filtering threshold;

[0070] Using the denoising function to remove noise from the shifted signal to obtain a denoised signal;

[0071] Normalization is performed on the denoised signal to obtain the preprocessed signal.

[0072] Specifically, wavelet transform denoising is a common denoising technique. By decomposing the original signal, wavelet coefficients of different sizes can be generated. Then, a filtering threshold is set. Among the many wavelet coefficients, the coefficients with smaller absolute values ​​are set to zero, while the coefficients with larger absolute values ​​are retained or shrunk. The thresholded wavelet coefficients are then reconstructed to achieve the purpose of denoising.

[0073] Preferably, the expression of the filtering threshold is:

[0074]

[0075] Among them, λ is the filtering threshold, j is the decomposition scale of the shifted signal, d j is the decomposition of the wavelet coefficients with a scale of j, N represents the length of the shifted signal, and median represents the median operation.

[0076] Specifically, in the wavelet threshold denoising process, the quality of threshold selection directly affects the denoising effect. Since the amplitude of the noise wavelet coefficient decreases with the increase of the decomposition scale, this embodiment introduces the decomposition scale to adaptively adjust the filtering threshold.

[0077] Commonly used threshold functions include hard and soft threshold functions. Due to the discontinuity of the shrinkage function, the hard threshold function can produce artificial noise points in the recovered signal, causing oscillations in the reconstructed signal. The soft threshold function shrinks wavelet coefficients greater than the threshold. While this process improves the continuity of the processed wavelet coefficients, it also loses some useful high-frequency information. This embodiment addresses the shortcomings of hard and soft threshold functions and proposes a denoising function.

[0078] Preferably, removing noise from the shifted signal using the denoising function to obtain a denoised signal comprises:

[0079] The corresponding wavelet coefficients are removed using the denoising function to obtain a smoothed signal; wherein the denoising function is:

[0080]

[0081] Among them, X represents the wavelet coefficient, n represents the adjustment coefficient;

[0082] Performing inverse translation on each of the smoothed signals to obtain an inverse translated signal;

[0083] An average of multiple inverse-shifted signals is obtained to obtain the denoised signal.

[0084] Specifically, this embodiment can not only effectively suppress the pseudo-Gibbs effect by continuously shifting the signal, but also obtain a smaller mean square error than the threshold method denoising, thereby further improving the signal-to-noise ratio.

[0085] Preferably, the sample characteristic parameters include: conductivity characteristic sub-parameters, capacitance characteristic sub-parameters, impedance spectrum characteristic sub-parameters, time domain characteristic sub-parameters and frequency domain characteristic sub-parameters; the conductivity characteristic sub-parameters include: DC conductivity, AC conductivity and conductivity change rate; the capacitance characteristic sub-parameters include: static capacitance value, dynamic capacitance value and dielectric loss factor; the impedance spectrum characteristic sub-parameters include: impedance amplitude, impedance phase angle, impedance frequency response curve and characteristic frequency points; the time domain characteristic sub-parameters include: rise time and fall time of the electrical signal, signal attenuation rate; the frequency domain characteristic sub-parameters include: spectrum energy distribution and frequency bandwidth.

[0086] Specifically, this embodiment performs data segmentation and time-domain analysis on the preprocessed electrical measurement signals (including conductivity signals, capacitance characteristic signals, and impedance spectrum signals). By segmenting the signals, key parameters such as DC conductivity, low-frequency conductivity, and high-frequency conductivity can be extracted. These parameters can reflect the electrical conductivity and moisture content of hair. Simultaneously, time-domain analysis methods are used to extract dynamic characteristics such as the signal's rise time, fall time, and signal decay rate. These characteristics are closely related to the physical structure and health of the hair. Secondly, frequency-domain analysis is performed on the electrical signals to extract frequency response characteristic parameters. Using a fast Fourier transform (FFT), the signals are converted from the time domain to the frequency domain, analyzing the variations in the impedance amplitude, phase angle, and capacitance values ​​of the hair at different frequencies. Characteristic parameters include the peak frequency, inflection frequency, frequency bandwidth, and dielectric loss factor of the impedance spectrum. These parameters can reveal the internal structure, surface properties, and polarization characteristics of the hair. Next, statistical methods are used to quantify the overall characteristics of the signal and extract global characteristic parameters. For example, statistical features such as the mean, variance, peak value, and energy distribution of the conductivity, capacitance, and impedance spectrum signals are calculated. These global features reflect the overall electrical properties of the hair sample and provide foundational data for subsequent classification and analysis. Then, by incorporating the nonlinear characteristics of the signals, high-level feature parameters are extracted. By analyzing the nonlinear dynamic behavior of the signals (such as the nonlinear changes in the impedance phase angle and the nonlinear fitting parameters of the frequency response curve), the complex electrical properties of hair can be further revealed. Furthermore, multi-scale analysis of the signals using methods such as wavelet transforms is performed to extract feature parameters at different scales, capturing the multi-level information of the hair sample. Finally, all extracted feature parameters are integrated to form a high-dimensional feature vector. To improve feature validity and reduce redundancy, principal component analysis (PCA) or linear discriminant analysis (LDA) can be used to reduce the dimensionality of the feature vector, retaining the most discriminative feature parameters. These final sample feature parameters serve as input data for subsequent hair type analysis model training and classification.

[0087] Furthermore, the sample characteristic parameters of this embodiment include:

[0088] (1) Conductivity characteristic sub-parameters

[0089] DC conductivity: reflects the electrical conductivity of the hair sample and is related to the water content, internal structure and chemical composition (such as protein and minerals) of the hair.

[0090] AC conductivity: Changes in conductivity at different frequencies reflect the frequency response characteristics of hair and can reveal the internal structure and surface properties of hair.

[0091] Conductivity change rate: The rate of change of conductivity within different frequency ranges, used to analyze how the conductivity of hair changes with frequency.

[0092] (2) Capacitance characteristic parameters

[0093] Static capacitance: The capacitance value of a hair sample under low-frequency conditions, reflecting the dielectric properties and surface structure of the hair.

[0094] Dynamic capacitance value: The change of capacitance value at different frequencies reflects the polarization characteristics and internal dielectric distribution of hair.

[0095] Dielectric loss factor: The energy loss of a hair sample under the action of an electric field can reflect the health status and pollutant content of the hair.

[0096] (3) Impedance spectrum characteristic sub-parameters

[0097] Impedance amplitude: The impedance at different frequencies reflects the overall electrical properties of hair.

[0098] Impedance phase angle: The change in the impedance phase angle can reveal the electrical response characteristics and internal structure of hair.

[0099] Impedance frequency response curve: The shape of the curve showing the change of impedance with frequency is used to analyze the conductivity, polarization characteristics and internal structure of hair.

[0100] Characteristic frequency point: The characteristic frequency (such as peak frequency, inflection point frequency) that appears in the impedance spectrum is used to distinguish different types of hair.

[0101] (4) Time domain characteristic sub-parameters

[0102] The rise time and fall time of the electrical signal: reflect the response speed of hair to electrical signals and are related to the conductivity and polarization properties of hair.

[0103] Signal decay rate: The rate at which the electrical signal decays in a hair sample can reveal the internal structure and health of the hair.

[0104] (5) Frequency domain characteristic parameters

[0105] Spectral energy distribution: The energy distribution of electrical signals within different frequency ranges, used to analyze the frequency response characteristics of hair.

[0106] Frequency bandwidth: The effective bandwidth of the electrical signal of a hair sample in the frequency domain, reflecting the conductivity and polarization properties of hair.

[0107] Preferably, the sample characteristic parameters are input into an initial neural network for training to obtain a hair type analysis model, including:

[0108] Acquire a preset conductivity characteristic sub-parameter dataset, a capacitance characteristic sub-parameter dataset, an impedance spectrum characteristic sub-parameter dataset, a time domain characteristic sub-parameter dataset, and a frequency domain characteristic sub-parameter dataset;

[0109] The initial neural network is trained according to the conductivity characteristic sub-parameter data set, the capacitance characteristic sub-parameter data set, the impedance spectrum characteristic sub-parameter data set, the time domain characteristic sub-parameter data set, and the frequency domain characteristic sub-parameter data set to obtain a trained first classifier, a second classifier, a third classifier, a fourth classifier, and a fifth classifier;

[0110] Cascade the trained classifiers to obtain a classification network;

[0111] The trained LSTM neural network is connected to the classification network to obtain the hair type analysis model.

[0112] Preferably, the analysis results include: hair type, health condition, and whether the hair contains prohibited substances.

[0113] Specifically, this embodiment first acquires a dataset of pre-set conductivity characteristic sub-parameters, capacitance characteristic sub-parameters, impedance spectrum characteristic sub-parameters, time domain characteristic sub-parameters, and frequency domain characteristic sub-parameters. These datasets are obtained by extracting features from the electrical measurement signals of a large number of hair samples and are manually annotated (e.g., with labels such as hair type, health status, and prohibited substance content) to form training datasets. Each sub-parameter dataset contains input data for the corresponding characteristic parameter and a corresponding target label, which is used to train different classifiers.

[0114] Secondly, the initial neural network is trained independently according to each characteristic sub-parameter data set to obtain the first classifier (based on conductivity features), the second classifier (based on capacitance features), the third classifier (based on impedance spectrum features), the fourth classifier (based on time domain features), and the fifth classifier (based on frequency domain features). The training process of each classifier includes forward propagation, loss function calculation, and backpropagation optimization to ensure that the classifier can accurately identify the patterns in the corresponding feature data set. For example, the conductivity feature classifier can identify the relationship between changes in hair conductivity and health status, while the impedance spectrum feature classifier can capture the frequency response characteristics of hair types.

[0115] The trained classifiers are then cascaded to construct a classification network. This cascade process utilizes the output of each classifier as input to the intermediate layer of the classification network, enabling multi-feature fusion classification. Specifically, the classification network integrates the classification results of conductivity, capacitance, impedance spectrum, time domain, and frequency domain features to further improve the accuracy of hair type and health status identification. This multi-classifier cascade approach fully leverages the complementarity of different features and ensures model robustness.

[0116] Next, a trained LSTM (Long Short-Term Memory) neural network is connected to the backend of the classification network for time series data processing and dynamic characteristic analysis. The LSTM network can capture potential temporal correlations in hair electrodynamic signals, such as the dynamic patterns of impedance changes at different frequencies or the decay trend of capacitance characteristics over time. By combining the classification network and the LSTM network, a complete hair type analysis model is ultimately formed. This model not only classifies static features but also analyzes dynamic characteristics, further improving the ability to determine hair type, health status, and prohibited substance content.

[0117] Finally, the trained hair type analysis model is used to analyze the hair sample to be tested and generate analysis results. Specifically, the electrical measurement signal of the hair sample to be tested is input, and the model will first classify the hair type (such as straight hair, curly hair, etc.) and health status (such as whether it is damaged, whether the moisture content is normal) through the classification network. At the same time, the LSTM network is used to conduct a comprehensive analysis of the dynamic characteristics to determine whether the hair contains banned substances (such as drug residues, heavy metal pollution, etc.). Finally, the model outputs the analysis results, including the hair type, health status score, and banned substance detection results, providing users with a comprehensive hair analysis report.

[0118] The beneficial effects of the present invention are as follows:

[0119] (1) The present invention is based on electrical data processing technology and can complete the detection and analysis of hair samples within a few minutes, significantly improving the efficiency of customs inspection.

[0120] (2) Electrical measuring equipment is small in size and light in weight, making it suitable for use in mobile scenarios such as customs.

[0121] (3) Through electrical property analysis, the physical properties, health status and pollutant content of hair can be detected simultaneously to achieve comprehensive analysis.

[0122] (4) Multi-frequency electrical scanning technology and machine learning algorithms are used to ensure the accuracy and reliability of the test results.

[0123] (5) The present invention adopts a non-invasive electrical measurement method, which does not damage the hair sample and is suitable for subsequent further testing or storage.

[0124] (6) In addition to customs inspection, the present invention can also be applied to fields such as beauty care, medical diagnosis and forensic identification, providing technical support for related industries.

[0125] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0126] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the method and core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A comprehensive analysis method for hair types, characterized in that: include: collecting a hair sample and pre-treating the surface of the hair sample; Using electrical measurement equipment to measure electrical data of the pre-processed hair sample to obtain electrical measurement signals; the electrical measurement signals include conductivity signals, capacitance characteristic signals, and impedance spectrum signals; performing signal denoising and normalization processing on the electrical measurement signal to obtain a preprocessed signal; Performing feature extraction on the preprocessed signal to obtain sample feature parameters; Inputting the sample characteristic parameters into an initial neural network for training to obtain a hair type analysis model; The electrical measurement signal of the hair to be tested is input into the hair type analysis model to obtain the analysis result.

2. The comprehensive analysis method of hair type according to claim 1, characterized in that: Pre-treating the surface of the hair sample, comprising: performing a cleaning process on the hair sample; The cleaned hair samples were dried to eliminate the interference of surface impurities on the electrical property measurement.

3. The comprehensive analysis method of hair type according to claim 2, characterized in that: Performing signal denoising and normalization processing on the electrical measurement signal to obtain a preprocessed signal, including: performing a translation on the electrical measurement signal to obtain a translated signal; performing wavelet decomposition on the shifted signal to obtain a plurality of wavelet coefficients; Determine the filtering threshold according to the decomposition scale and length of the signal; Constructing a denoising function according to the filtering threshold; Using the denoising function to remove noise from the shifted signal to obtain a denoised signal; Normalization is performed on the denoised signal to obtain the preprocessed signal.

4. The comprehensive analysis method of hair type according to claim 3, characterized in that: The expression of the filtering threshold is: Among them, λ is the filtering threshold, j is the decomposition scale of the shifted signal, d j is the decomposition of the wavelet coefficients with a scale of j, N represents the length of the shifted signal, and median represents the median operation.

5. The comprehensive analysis method of hair type according to claim 4, characterized in that: Removing noise from the shifted signal using the denoising function to obtain a denoised signal includes: The corresponding wavelet coefficients are removed using the denoising function to obtain a smoothed signal; wherein the denoising function is: Among them, X represents the wavelet coefficient, n represents the adjustment coefficient; Performing inverse translation on each of the smoothed signals to obtain an inverse translated signal; An average of multiple inverse-shifted signals is obtained to obtain the denoised signal.

6. The comprehensive analysis method of hair type according to claim 1, characterized in that: The sample characteristic parameters include: conductivity characteristic sub-parameters, capacitance characteristic sub-parameters, impedance spectrum characteristic sub-parameters, time domain characteristic sub-parameters and frequency domain characteristic sub-parameters; the conductivity characteristic sub-parameters include: DC conductivity, AC conductivity and conductivity change rate; the capacitance characteristic sub-parameters include: static capacitance value, dynamic capacitance value and dielectric loss factor; the impedance spectrum characteristic sub-parameters include: impedance amplitude, impedance phase angle, impedance frequency response curve and characteristic frequency points; the time domain characteristic sub-parameters include: rise time and fall time of the electrical signal, and signal attenuation rate; the frequency domain characteristic sub-parameters include: spectrum energy distribution and frequency bandwidth.

7. The comprehensive analysis method of hair type according to claim 6, characterized in that: The sample characteristic parameters are input into the initial neural network for training to obtain a hair type analysis model, including: Acquire a preset conductivity characteristic sub-parameter dataset, a capacitance characteristic sub-parameter dataset, an impedance spectrum characteristic sub-parameter dataset, a time domain characteristic sub-parameter dataset, and a frequency domain characteristic sub-parameter dataset; The initial neural network is trained according to the conductivity characteristic sub-parameter data set, the capacitance characteristic sub-parameter data set, the impedance spectrum characteristic sub-parameter data set, the time domain characteristic sub-parameter data set, and the frequency domain characteristic sub-parameter data set to obtain a trained first classifier, a second classifier, a third classifier, a fourth classifier, and a fifth classifier; Cascade the trained classifiers to obtain a classification network; The trained LSTM neural network is connected to the classification network to obtain the hair type analysis model.

8. The comprehensive analysis method of hair type according to claim 1, characterized in that: The analysis results include: hair type, health status and whether it contains banned substances.

Images