Fabric polyester fiber detection method based on spectral imaging and curve fitting

Abstract

The invention discloses a textile fabric polyester fiber detection method based on spectral imaging and curve fitting, and relates to the technical field of hyperspectral imaging analysis, and the method comprises the steps: carrying out the characteristic decomposition and principal component analysis of a reflectivity function after baseline correction, and obtaining a characteristic curve; extracting feature information of the hyperspectral image, and fusing the feature information of the hyperspectral image with the feature curve to generate a multi-dimensional feature vector; and constructing a hyperspectral fiber content detection model, inputting the multi-dimensional feature vector and the feature parameter vector into the hyperspectral fiber content detection model, and generating the polyester fiber content. Not only are the problems of background noise and drift interference solved, but also the multi-dimensional characteristics of the hyperspectral data are fully mined. The content of the polyester fibers in the waste textile fabrics can be efficiently detected on line. Finally, according to the method, the detection precision and reliability are remarkably improved, and an efficient, environment-friendly and accurate technical means is provided for resource utilization of the waste textile fabrics.

Description

Technical Field

[0001] The invention relates to the technical field of hyperspectral imaging analysis, in particular to a textile polyester fiber detection method based on spectral imaging and curve fitting. Background Art

[0002] With the rapid development of the global textile industry, a large amount of waste textiles generated every year not only causes serious pollution to the environment, but also leads to a huge waste of resources. In order to improve the recycling efficiency of resources, the classification and recycling technology of waste textiles has gradually become a research hotspot. In the classification of textiles, polyester fiber is the most common chemical fiber, and the accurate detection of its content is of great significance to the recycling of waste textiles. At present, the detection methods of polyester fiber content mainly include chemical method, thermal analysis method and optical analysis method. The chemical method separates the fiber components through solvent extraction and other methods, but its process is complicated and pollutes the environment; the thermal analysis method infers the fiber composition by detecting the thermal performance indicators of the fiber, but the equipment requirements are high and the detection accuracy is limited; the optical analysis method uses the spectral characteristics of the fiber for component analysis, which is currently a more advanced detection technology. Among them, hyperspectral imaging technology has attracted much attention because of its ability to obtain spectral and spatial information at the same time. However, there are still technical bottlenecks for the online detection of polyester fiber content in waste textiles.

[0003] The existing polyester fiber detection technology based on hyperspectral imaging has certain limitations. First, the acquisition and correction process of hyperspectral data is complicated, especially when facing the complex background of waste textiles, it is easily interfered by factors such as background noise and spectral baseline drift, resulting in low reliability of reflectance data and affecting the final detection results. Secondly, most existing technologies fail to fully utilize the multidimensional characteristics of hyperspectral data in the process of spectral data modeling. They often only focus on a single spectral feature or use a simple linear model, which cannot effectively reveal the complex spectral and compositional relationship of waste textiles, resulting in insufficient detection accuracy, especially when the fiber mixing ratio is high or the spectral difference is small. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a textile polyester fiber detection method based on spectral imaging and curve fitting to solve the problems of background noise and spectral baseline drift interference and insufficient utilization of the multi-dimensional characteristics of hyperspectral data.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a method for detecting polyester fibers in textiles based on spectral imaging and curve fitting, which includes collecting hyperspectral data of waste textiles, correcting the collected hyperspectral data, and generating a corrected reflectance function and a hyperspectral image; based on the corrected reflectance function, generating a baseline-corrected reflectance function by dynamically adjusting the baseline parameters; performing feature decomposition and principal component analysis on the baseline-corrected reflectance function to obtain a characteristic curve; extracting feature information of the hyperspectral image, fusing the feature information of the hyperspectral image with the characteristic curve to generate a multi-dimensional feature vector; performing peak search on the characteristic curve after the principal component analysis to generate a feature parameter vector; constructing a hyperspectral fiber content detection model, inputting the multi-dimensional feature vector and the feature parameter vector into the hyperspectral fiber content detection model to generate the polyester fiber content.

[0008] As a preferred solution of the textile polyester fiber detection method based on spectral imaging and curve fitting of the present invention, wherein: the hyperspectral data of waste textiles are collected, the collected hyperspectral data are corrected, and the corrected reflectance function and hyperspectral image are generated. The specific steps are as follows:

[0009] Use a hyperspectral instrument to collect hyperspectral data of waste textiles;

[0010] The collected hyperspectral data are corrected through dark current correction, white reference correction and noise filtering to generate corrected reflectance function and hyperspectral image.

[0011] As a preferred solution of the textile polyester fiber detection method based on spectral imaging and curve fitting of the present invention, wherein: based on the corrected reflectance function, by dynamically adjusting the baseline parameters, the reflectance function after baseline correction is generated, and the specific steps are as follows:

[0012] Based on the corrected reflectivity function, the background signal of the corrected reflectivity function is removed, and the dynamic nonlinear baseline function B(λ; α, β, γ) is constructed according to the trend, local change and periodic noise of the background signal. The expression is:

[0013]

[0014] Among them, α is the amplitude parameter, β is the trend change rate parameter, γ is the periodic noise amplitude parameter, and λ 0 is the segmentation point, λ is the wavelength;

[0015] By initializing parameters, dynamically determining segmentation points and optimizing baseline function parameters, the reflectivity function after baseline correction is generated, and the expression is:

[0016] R b (λ)=R c (λ)-B(λ; α* ,β * ,γ * );

[0017] Among them, R b (λ) is the reflectance function after baseline correction, R c (λ) is the corrected reflectivity function, B(λ; α * ,β * ,γ * ) is the optimized dynamic baseline function.

[0018] As a preferred solution of the textile polyester fiber detection method based on spectral imaging and curve fitting of the present invention, wherein: the reflectance function after baseline correction is subjected to feature decomposition and principal component analysis to obtain a characteristic curve, the specific steps are as follows:

[0019] Based on the reflectance function after baseline correction, a characteristic matrix is ​​constructed, and the wavelength of the characteristic matrix is ​​standardized;

[0020] Based on the standardized feature matrix, the covariance matrix is ​​calculated, the covariance matrix is ​​eigen-decomposed and the principal components are extracted;

[0021] Based on the principal components of the covariance matrix, the principal component analysis characteristic curve is constructed, and the expression is:

[0022]

[0023] Among them, Φ p is the principal component characteristic curve value, M j,p is the jth sample at wavelength λ p The normalized reflectance value, q kj is the eigenvector value of the kth principal component on sample j, λ k is the eigenvalue corresponding to the kth principal component, k is the number of principal components, N is the number of samples, and λ p is the wavelength of the characteristic matrix after normalization, and p is the serial number of the wavelength.

[0024] As a preferred solution of the textile polyester fiber detection method based on spectral imaging and curve fitting of the present invention, wherein: the feature information of the hyperspectral image is extracted, the feature information of the hyperspectral image is fused with the feature curve to generate a multi-dimensional feature vector, and the specific steps are as follows:

[0025] Extract the complete spectral information of each pixel in the hyperspectral image to form a spectral feature matrix;

[0026] Calculate the local spatial mean and local spatial entropy of each pixel, and concatenate the local spatial mean and entropy by band to form a spatial feature matrix;

[0027] The characteristic information of the hyperspectral image is fused with the characteristic curve after principal component analysis, and the expression is:

[0028]

[0029] Among them, Ψ is the fused multi-dimensional feature vector, Φ b is the principal component characteristic curve value, B represents the total number of spectral bands, N s is the total number of pixels in the hyperspectral image, P ′ (b, s) represents the wavelength λ in the standardized spectral matrix b For the spectral value of the sth pixel, ρ is the fusion weight parameter, F s (b,q) is the qth spatial feature in band λ b , κ is the fusion weight parameter in the denominator, ξ is a specific dynamic adjustment parameter, b is the spectral band number, s is the pixel number, and q is the dimension number of the spatial feature.

[0030] As a preferred solution of the textile polyester fiber detection method based on spectral imaging and curve fitting of the present invention, wherein: the characteristic curve obtained after principal component analysis is peaked to generate a characteristic parameter vector, and the specific steps are as follows:

[0031] The characteristic curve after principal component analysis is smoothed, and the first-order derivative and second-order derivative of the smoothed characteristic curve are calculated. The local maximum value is found through the joint analysis of the first-order derivative and the second-order derivative, and the characteristic parameters corresponding to the peak intensity, peak width and peak sharpness are extracted to generate a characteristic parameter vector.

[0032] As a preferred solution of the textile polyester fiber detection method based on spectral imaging and curve fitting of the present invention, the specific steps of constructing a high-spectral fiber content detection model are as follows:

[0033] A regression model based on nonlinear kernel function is used to construct a hyperspectral fiber content detection model, which is expressed as follows:

[0034]

[0035] Wherein, C is the polyester fiber content, Υ is the comprehensive characteristics, is the peak intensity of the rth spectral peak, is the peak width of the rth spectral peak, is the peak sharpness of the rth spectral peak, H is the total number of spectral peaks, η and δ are regularization parameters, Ψ t represents the tth component of the multidimensional feature vector, and D is the number of dimensions of the multidimensional feature vector.

[0036] As a preferred solution of the textile polyester fiber detection method based on spectral imaging and curve fitting of the present invention, wherein: the multi-dimensional feature vector and the feature parameter vector are input into the hyperspectral fiber content detection model to generate the polyester fiber content, the specific steps are as follows:

[0037] Calculate the comprehensive feature Υ of the multi-dimensional feature vector and feature parameter vector, the expression is:

[0038]

[0039] Among them, t is the component of the multi-dimensional feature vector, ω and θ are dynamic adjustment parameters;

[0040] The comprehensive features are input into the hyperspectral fiber content detection model to generate the polyester fiber content of waste textiles.

[0041] In a second aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the textile polyester fiber detection method based on spectral imaging and curve fitting as described in the first aspect of the present invention is implemented.

[0042] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the method for detecting polyester fibers in textiles based on spectral imaging and curve fitting as described in the first aspect of the present invention is implemented.

[0043] The beneficial effects of the present invention are as follows: the present invention not only solves the background noise and drift interference problems, but also fully exploits the multidimensional characteristics of hyperspectral data through two core innovative steps: dynamic baseline correction and multidimensional feature fusion. Combining principal component analysis with characteristic curve extraction, the present invention can efficiently perform online detection of the polyester fiber content of waste textiles. Ultimately, this method significantly improves detection accuracy and reliability, and provides an efficient, environmentally friendly and accurate technical means for the resource utilization of waste textiles. BRIEF DESCRIPTION OF THE DRAWINGS

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

[0045] Figure 1 This is a flow chart of the method for detecting polyester fibers in textiles based on spectral imaging and curve fitting in Example 1.

[0046] Figure 2This is a schematic diagram of the polyester fiber content produced in Example 1. DETAILED DESCRIPTION

[0047] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0048] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0049] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0050] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, provides a method for detecting polyester fibers in textiles based on spectral imaging and curve fitting, comprising the following steps:

[0051] Collecting hyperspectral data of waste textiles, correcting the collected hyperspectral data, and generating corrected reflectance functions and hyperspectral images;

[0052] Furthermore, a hyperspectral instrument is used to collect hyperspectral data of waste textiles;

[0053] It should be noted that the surface of the waste textile samples should be free of obvious wrinkles, stains and dust to reduce the interference of spectral data;

[0054] Configure the acquisition parameters of the hyperspectral instrument in terms of wavelength range, spectral resolution and spatial resolution;

[0055] The collected hyperspectral data are corrected by dark current correction, white reference correction and noise filtering to generate a corrected reflectance function and a hyperspectral image;

[0056] It should be noted that dark current is the background noise of the hyperspectral instrument under lightless conditions, which will affect the measurement accuracy. The influence of dark current can be eliminated by collecting dark reference data and subtracting the dark reference value from the original data.

[0057] White reference correction is used to eliminate the influence of light source non-uniformity on measurement by collecting spectral data of a standard white plate with high reflectivity and normalizing the collected textile data;

[0058] Noise filtering is to remove random noise through smoothing filtering while retaining the main characteristic information of the spectrum.

[0059] Based on the corrected reflectance function, a baseline-corrected reflectance function is generated by dynamically adjusting the baseline parameters;

[0060] Furthermore, based on the corrected reflectivity function, the background signal of the corrected reflectivity function is removed, and according to the trend, local change and periodic noise of the background signal, a dynamic nonlinear baseline function B(λ; α, β, γ) is constructed, and the expression is:

[0061]

[0062] Among them, α is the amplitude parameter, which controls the overall amplitude of the background signal; β is the trend change rate parameter, which describes the rate at which the background signal changes with the wavelength, affecting the frequency of trend changes and periodic noise; γ is the periodic noise amplitude parameter, which controls the fluctuation intensity of the periodic noise; and λ 0 is the segmentation point, which defines the dividing point between the two signal change modes. It is usually dynamically selected according to the characteristics of background changes. λ is the wavelength;

[0063] It should be noted that trend changes are background signals that may decay exponentially or grow logarithmically with wavelength, local changes are due to sample surface differences or local interference, background signals may mutate within certain wavelength ranges, and periodic noise is periodic interference from the light source or instrument that may cause sine or cosine fluctuations in the background signal;

[0064] The expression of the dynamic baseline function is divided into two parts.

[0065] When λ≤λ 0 When , then B(λ;α,β,γ)=αe -βλ +γcos(βλ), the first term represents the exponential decay trend of the background signal, and the second term describes the cosine fluctuation of the periodic noise;

[0066] When λ>λ 0 When B(λ; α, β, γ) = αln(1+βλ)+γsin(βλ), the first term represents the logarithmic growth trend of the background signal, and the second term describes the sinusoidal fluctuation of the periodic noise;

[0067] By initializing parameters, dynamically determining segmentation points and optimizing baseline function parameters, the reflectivity function after baseline correction is generated, and the expression is:

[0068] R b (λ)=R c (λ)-B(λ; α * , β *, γ * );

[0069] Where Rb(λ) is the reflectance function after baseline correction, R c (λ) is the corrected reflectivity function, B(λ; α * , β * , γ * ) is the optimized dynamic baseline function.

[0070] It should be noted that the initialization parameters are set based on the preliminary analysis of the background signal. The segmentation points are dynamically adjusted using the changing characteristics of the background signal to ensure the fitting accuracy of the two segments of the baseline function. The parameters are adjusted using the optimization algorithm to minimize the difference between the baseline function and the background signal. The reflectance function after baseline correction is calculated based on the optimized baseline function and the corrected reflectance function.

[0071] Performing feature decomposition and principal component analysis on the reflectance function after baseline correction to obtain a characteristic curve;

[0072] Furthermore, based on the reflectance function after baseline correction, a characteristic matrix is ​​constructed, and the wavelength of the characteristic matrix is ​​standardized;

[0073] It should be noted that the baseline-corrected reflectance function Rb(λ) of all samples is collected and arranged by sample to form a matrix M, which is expressed as:

[0074]

[0075] Among them, the dimension of matrix M is N×P, N is the number of samples, and P is the number of wavelengths;

[0076] The wavelength column is standardized to eliminate the magnitude differences of reflectance values ​​at different wavelengths. The expression is:

[0077]

[0078] Among them, M t，p is the sample t at wavelength λ p The normalized reflectance value, σ p is the wavelength λ p The standard deviation of p is the wavelength λ p The mean reflectance value of all samples at ;

[0079] Based on the standardized feature matrix, the covariance matrix is ​​calculated, the covariance matrix is ​​eigen-decomposed and the principal components are extracted;

[0080] It should be noted that the covariance matrix is ​​calculated as follows:

[0081]

[0082] Where S is the covariance matrix;

[0083] The covariance matrix is ​​decomposed into its own features, and the expression is:

[0084] C k =v k q k ;

[0085] Among them, v k represents the eigenvalue of the kth principal component, reflecting the variance contribution of the principal component, q k Represents the eigenvector of the kth principal component, reflecting the weight of the principal component at each wavelength;

[0086] Select the first k principal components according to the eigenvalues ​​for subsequent feature analysis;

[0087] Based on the principal components of the covariance matrix, the principal component analysis characteristic curve is constructed, and the expression is:

[0088]

[0089] Among them, Φ p is the principal component characteristic curve value, M j，p is the jth sample at wavelength λ p The normalized reflectance value, q kj is the eigenvector value of the kth principal component on sample j, λ k is the eigenvalue corresponding to the kth principal component, k is the number of principal components, N is the number of samples, and λ p is the wavelength of the characteristic matrix after normalization, and p is the serial number of the wavelength;

[0090] It should be noted that the significance of the principal component is to describe the principal component characteristics of the hyperspectral data at different wavelengths, which can help identify the key bands related to the fiber components.

[0091] Extract the characteristic information of the hyperspectral image, fuse the characteristic information of the hyperspectral image with the characteristic curve, and generate a multi-dimensional feature vector;

[0092] Furthermore, the complete spectral information of each pixel of the hyperspectral image is extracted to form a spectral feature matrix;

[0093] It should be noted that the hyperspectral image is a three-dimensional data cube. The hyperspectral processing tool is used to read the three-dimensional data cube, extract the complete spectral information of each pixel in the hyperspectral image, and form a spectral feature matrix according to the pixel arrangement;

[0094] Calculate the local spatial mean and local spatial entropy of each pixel, and concatenate the local spatial mean and entropy by band to form a spatial feature matrix;

[0095] It should be noted that calculating the local spatial mean of each pixel is to calculate the mean of its neighborhood for each pixel in each band; calculating the local spatial entropy is to calculate the distribution complexity of pixel values ​​in the local area;

[0096] The characteristic information of the hyperspectral image is fused with the characteristic curve after principal component analysis, and the expression is:

[0097]

[0098] Among them, Ψ is the fused multi-dimensional feature vector, Φ b is the principal component characteristic curve value, B represents the total number of spectral bands, N s is the total number of pixels in the hyperspectral image, and P′(b, s) represents the wavelength λ in the standardized spectral matrix. b For the spectral value of the sth pixel, ρ is the fusion weight parameter, F s (b, q) is the qth spatial feature in band λ b , κ is the fusion weight parameter in the denominator, ξ is a specific dynamic adjustment parameter, b is the spectral band number, s is the pixel number, and q is the dimension number of the spatial feature.

[0099] It should be noted that the fusion step combines the spectral information, spatial features and characteristic curves extracted by principal component analysis of the hyperspectral image to generate a multi-dimensional feature vector. The multi-dimensional feature vector integrates spectral, spatial and statistical characteristic information, providing a more comprehensive feature expression for subsequent classification or recognition tasks.

[0100] Peak search is performed on the characteristic curve obtained after principal component analysis to generate a characteristic parameter vector;

[0101] Furthermore, the characteristic curve after principal component analysis is smoothed, and the first-order derivative and the second-order derivative of the characteristic curve after smoothing are calculated. The local maximum value is found through the joint analysis of the first-order derivative and the second-order derivative, and the characteristic parameters corresponding to the peak intensity, peak width and peak sharpness are extracted to generate a characteristic parameter vector;

[0102] It should be noted that the characteristic curve after principal component analysis is the characteristic weight distribution on the spectral band. The peak corresponds to the significant feature of a specific band. Through the peak search operation, the key characteristic points on the band can be extracted to provide focus information for subsequent analysis.

[0103] The first-order derivative is used to determine the growth or decline trend of the characteristic curve, and the second-order derivative is used to determine the curvature change of the characteristic curve and assist in determining the peak point;

[0104] Peak intensity is the characteristic curve value of the peak point, peak width is the width of the peak at a specific height, and peak sharpness is the ratio of peak intensity to peak width, reflecting the shape characteristics of the peak.

[0105] A hyperspectral fiber content detection model is constructed, and the multi-dimensional feature vector and feature parameter vector are input into the hyperspectral fiber content detection model to generate the polyester fiber content;

[0106] Furthermore, a regression model based on nonlinear kernel function is used to construct a hyperspectral fiber content detection model, which is expressed as follows:

[0107]

[0108] Among them, C is the polyester fiber content, Y is the comprehensive characteristics, is the peak intensity of the rth spectral peak, is the peak width of the rth spectral peak, is the peak sharpness of the rth spectral peak, H is the total number of spectral peaks, η and δ are regularization parameters, Ψ t represents the tth component of the multidimensional feature vector, and D is the number of dimensions of the multidimensional feature vector;

[0109] It should be noted that the hyperspectral fiber content detection model can capture the nonlinear relationship between complex features through the nonlinear kernel function. The weighted integral of the comprehensive features is represented. The nonlinear kernel function cos(πz) is used to perform nonlinear transformation on the comprehensive features, which enhances the fitting ability of the hyperspectral fiber content detection model and captures the distribution characteristics of the comprehensive features at different scales. Reflects the contribution of spectral peak parameters to polyester fiber content, represents the combined effect of peak intensity and peak width, represents the nonlinear effect of peak sharpness, η is a regularization parameter used to prevent overfitting, It reflects the change gradient of the multi-dimensional eigenvector and measures the complexity of the eigenvector by summing the absolute values ​​of the differences between adjacent components. δ is a dynamic adjustment parameter used to balance the influence of the eigenvector on the fiber content.

[0110] Calculate the comprehensive feature Y of the multi-dimensional feature vector and feature parameter vector, the expression is:

[0111]

[0112] Among them, t is the component of the multi-dimensional feature vector, ω and θ are dynamic adjustment parameters;

[0113] It should be noted that is the normalization term of the multi-dimensional feature vector, which is used to suppress the influence of the feature dimension. is the normalization term of the spectral peak parameter, ω is used to control the weight of the peak parameter, and θ is a dynamic adjustment parameter used to balance the overall amplitude of the denominator;

[0114] The comprehensive features are input into the hyperspectral fiber content detection model to generate the polyester fiber content of waste textiles.

[0115] This embodiment also provides a computer device, which is suitable for the case of a textile polyester fiber detection method based on spectral imaging and curve fitting, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the textile polyester fiber detection method based on spectral imaging and curve fitting proposed in the above embodiment.

[0116] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0117] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for detecting polyester fibers in textiles based on spectral imaging and curve fitting as proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, disk or optical disk.

[0118] In summary, the present invention not only solves the problems of background noise and drift interference, but also fully exploits the multidimensional characteristics of hyperspectral data through two core innovative steps: dynamic baseline correction and multidimensional feature fusion. Combining principal component analysis with characteristic curve extraction, the present invention can efficiently perform online detection of the polyester fiber content of waste textiles. Ultimately, this method significantly improves detection accuracy and reliability, and provides an efficient, environmentally friendly and accurate technical means for the resource utilization of waste textiles.

[0119] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for detecting polyester fibers in textiles based on spectral imaging and curve fitting, characterized in that: include, Collecting hyperspectral data of waste textiles, correcting the collected hyperspectral data, and generating corrected reflectance functions and hyperspectral images; Based on the corrected reflectance function, a baseline-corrected reflectance function is generated by dynamically adjusting the baseline parameters; Performing feature decomposition and principal component analysis on the reflectance function after baseline correction to obtain a characteristic curve; Extract the characteristic information of the hyperspectral image, fuse the characteristic information of the hyperspectral image with the characteristic curve, and generate a multi-dimensional feature vector; Peak search is performed on the characteristic curve obtained after principal component analysis to generate a characteristic parameter vector; A hyperspectral fiber content detection model is constructed, and the multi-dimensional feature vector and feature parameter vector are input into the hyperspectral fiber content detection model to generate the polyester fiber content.

2. The method for detecting polyester fibers in textiles based on spectral imaging and curve fitting according to claim 1, characterized in that: The steps of collecting the hyperspectral data of waste textiles, correcting the collected hyperspectral data, and generating the corrected reflectance function and hyperspectral image are as follows: Use a hyperspectral instrument to collect hyperspectral data of waste textiles; The collected hyperspectral data are corrected through dark current correction, white reference correction and noise filtering to generate corrected reflectance function and hyperspectral image.

3. The method for detecting polyester fibers in textiles based on spectral imaging and curve fitting according to claim 1, characterized in that: The reflectivity function after baseline correction is generated by dynamically adjusting the baseline parameters based on the corrected reflectivity function. The specific steps are as follows: Based on the corrected reflectivity function, the background signal of the corrected reflectivity function is removed, and the dynamic nonlinear baseline function B(λ; α, β, γ) is constructed according to the trend, local change and periodic noise of the background signal. The expression is: Among them, α is the amplitude parameter, β is the trend change rate parameter, γ is the periodic noise amplitude parameter, λ0 is the segmentation point, and λ is the wavelength; By initializing parameters, dynamically determining segmentation points and optimizing baseline function parameters, the reflectivity function after baseline correction is generated, and the expression is: R b (λ)=R c (λ)-B(λ;a * ,b * ,c * ); Among them, R b (λ) is the reflectance function after baseline correction, R c (λ) is the corrected reflectivity function, B(λ; α * ,β * ,γ * ) is the optimized dynamic baseline function.

4. The method for detecting polyester fibers in textiles based on spectral imaging and curve fitting according to claim 1, characterized in that: The characteristic decomposition and principal component analysis are performed on the reflectivity function after baseline correction to obtain the characteristic curve. The specific steps are as follows: Based on the reflectance function after baseline correction, a characteristic matrix is ​​constructed, and the wavelength of the characteristic matrix is ​​standardized; Based on the standardized feature matrix, the covariance matrix is ​​calculated, the covariance matrix is ​​eigen-decomposed and the principal components are extracted; Based on the principal components of the covariance matrix, the principal component analysis characteristic curve is constructed, and the expression is: Among them, Φ p is the principal component characteristic curve value, M j,p is the jth sample at wavelength λ p The normalized reflectance value, q kj is the eigenvector value of the kth principal component on sample j, λ k is the eigenvalue corresponding to the kth principal component, k is the number of principal components, N is the number of samples, and λ p is the wavelength of the characteristic matrix after normalization, and p is the serial number of the wavelength.

5. The method for detecting polyester fibers in textiles based on spectral imaging and curve fitting according to claim 1, characterized in that: The method of extracting the characteristic information of the hyperspectral image and fusing the characteristic information of the hyperspectral image with the characteristic curve to generate a multi-dimensional feature vector is as follows: Extract the spectral information of each pixel in the hyperspectral image to form a spectral feature matrix; Calculate the local spatial mean and local spatial entropy of each pixel, and concatenate the local spatial mean and local spatial entropy by band to form a spatial feature matrix; The characteristic information of the hyperspectral image is fused with the characteristic curve after principal component analysis, and the expression is: Among them, Ψ is the fused multi-dimensional feature vector, Φ b is the principal component characteristic curve value, B represents the total number of spectral bands, N s is the total number of pixels in the hyperspectral image, and P′(b,s) represents the wavelength λ in the standardized spectral matrix. b For the spectral value of the sth pixel, ρ is the fusion weight parameter, F s (b,q) is the qth spatial feature in band λ b , κ is the fusion weight parameter in the denominator, ξ is a specific dynamic adjustment parameter, b is the spectral band number, s is the pixel number, and q is the dimension number of the spatial feature.

6. The method for online monitoring of polyester fiber content in waste textiles based on hyperspectral imaging and curve fitting as claimed in claim 1, characterized in that: The characteristic curve obtained after principal component analysis is peaked to generate a characteristic parameter vector. The specific steps are as follows: The characteristic curve after principal component analysis is smoothed, and the first-order derivative and second-order derivative of the smoothed characteristic curve are calculated. The local maximum value is found through the joint analysis of the first-order derivative and the second-order derivative, and the characteristic parameters corresponding to the peak intensity, peak width and peak sharpness are extracted to generate a characteristic parameter vector.

7. The method for detecting polyester fibers in textiles based on spectral imaging and curve fitting according to claim 1, characterized in that: The specific steps of constructing the hyperspectral fiber content detection model are as follows: A regression model based on nonlinear kernel function is used to construct a hyperspectral fiber content detection model, which is expressed as follows: Among them, C is the polyester fiber content, γ is the comprehensive characteristics, is the peak intensity of the rth spectral peak, is the peak width of the rth spectral peak, is the peak sharpness of the rth spectral peak, H is the total number of spectral peaks, η and δ are regularization parameters, Ψ t represents the tth component of the multidimensional feature vector, and D is the number of dimensions of the multidimensional feature vector.

8. The method for detecting polyester fibers in textiles based on spectral imaging and curve fitting according to claim 1, characterized in that: The multi-dimensional feature vector and the feature parameter vector are input into the hyperspectral fiber content detection model to generate the polyester fiber content. The specific steps are as follows: Calculate the comprehensive feature γ of the multi-dimensional feature vector and feature parameter vector, the expression is: Among them, t is the component of the multi-dimensional feature vector, ω and θ are dynamic adjustment parameters; The comprehensive features are input into the hyperspectral fiber content detection model to generate the polyester fiber content of waste textiles.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the textile polyester fiber detection method based on spectral imaging and curve fitting are implemented as described in any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for detecting polyester fibers in textiles based on spectral imaging and curve fitting are implemented as described in any one of claims 1 to 8.

Images