HSI classification method of LDA based on quaternion rayleigh entropy and e-commerce application device

Abstract

The invention discloses an HSI classification method of LDA based on quaternion rayleigh entropy and an e-commerce application device, is applied to a hyperspectral image (HSI) classification method and an e-commerce device, and relates to the technical field of hyperspectral image classification. According to the method, a target projection vector matrix corresponding to a to-be-classified hyperspectral image is determined from a decomposition result of a global quaternion divergence matrix and a decomposition result of an inter-class quaternion divergence matrix, and the inter-class quaternion divergence matrix facilitates effective data classification in a quaternion space. And the quaternion projection vector matrix expresses the characteristics of the quaternion in a specific direction. Therefore, the calculation complexity and the calculation amount when the quaternion discrimination method is used for processing HSI classification are reduced, and the HSI classification efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of hyperspectral image classification, and in particular to an HSI classification method and an e-commerce application device based on LDA of quaternion Rayleigh entropy. Background Art

[0002] Hyperspectral imaging technology combines imaging technology with spectral technology. While obtaining photos of samples, it can also obtain the chemical composition and spatial distribution information of the samples. It can detect the two-dimensional geometric spatial information and one-dimensional spectral information of the target. Therefore, it has a wide range of applications in food safety, medical diagnosis, aerospace, etc. An important step in hyperspectral imaging technology is hyperspectral image (HSI) classification. Hyperspectral image (HSI) classification first converts the spatial, spectral and radiation information contained in the hyperspectral image into an effective feature vector through a specific method, and then classifies and analyzes the sample components based on the feature vector.

[0003] Traditional HSI classification methods usually rely on technologies such as spatial texture and morphological feature extraction, spatial neighborhood information acquisition, and spatial information post-processing. Although these methods have small computational requirements and sample requirements, their classification accuracy is difficult to meet the requirements. With the development of technology, the use of quaternion discriminant methods to process HSI classification is becoming more and more widespread. Quaternion is an extended complex number that contains a real part and three imaginary parts, which can simultaneously represent multiple dimensions in space. In HSI classification, the application of quaternions can better capture and utilize the spatial and spectral information of data, thereby improving classification accuracy. Among them, Linear Discriminant Analysis (LDA) is a classic statistical method for dimensionality reduction and classification. The goal of LDA is to find a linear combination of features while maintaining the separability of the categories, so that data of different categories are separated as much as possible in this new feature space, while data of the same category are as close as possible. Rayleigh entropy refers to the use of Rayleigh distribution to describe the dispersion characteristics of signals in the field of signal processing and communication. Rayleigh distribution is a continuous probability distribution that can be used to describe, for example, the distribution of light intensity of different spectra. However, the use of quaternion discrimination methods to process HSI classification requires analysis based on a large amount of sample data. If the traditional quaternion discrimination method is used, the computational complexity and amount are relatively large, resulting in low efficiency of HSI classification.

[0004] Therefore, it is necessary to propose a new HSI classification method to reduce the computational complexity and amount of calculation when using the quaternion discrimination method to process HSI classification, thereby improving the efficiency of HSI classification. Summary of the invention

[0005] The present invention provides an HSI classification method and an e-commerce application device based on LDA of quaternion rayleigh entropy, which can reduce the computational complexity and amount when using quaternion discrimination method to process HSI classification, thereby improving the efficiency of HSI classification.

[0006] In order to solve the above technical problems, the first aspect of the present invention discloses an HSI classification method based on LDA of quaternion Rayleigh entropy, the method comprising:

[0007] Use quaternion vectors to represent all pixel vectors of the hyperspectral image to be classified, and obtain the quaternion set Q corresponding to the hyperspectral image to be classified h×N , where N is the number of the pixel vectors, and h is the number of the quaternion vectors corresponding to each pixel vector;

[0008] Determine m training pixel vectors from the N pixel vectors to obtain a training data set Q h×m The data category set R corresponding to the training data set C×m , where m is the preset number of training samples and C is the total number of spectral categories;

[0009] According to the training data set Q h×m The data category set R corresponding to the training data set C×m , determine the global quaternion divergence matrix S corresponding to the hyperspectral image to be classified g and the inter-class quaternion divergence matrix S b ;

[0010] The global quaternion divergence matrix S P and the inter-class quaternion divergence matrix S b Perform quaternion eigenvalue decomposition to obtain the global quaternion divergence matrix S g The decomposition result and the inter-class quaternion divergence matrix S b The decomposition results of

[0011] According to the global quaternion divergence matrix S P The decomposition result and the inter-class quaternion divergence matrix S b Decomposition results of the target projection vector matrix corresponding to the hyperspectral image to be classified are determined;

[0012] The hyperspectral image to be classified is classified according to the target projection vector matrix.

[0013] As an optional implementation, in the first aspect of the present invention, the hyperspectral image to be classified includes N pixel vectors, each of the pixel vectors includes M spectral bands and the radiation value corresponding to each of the spectral bands, wherein N and M are both positive integers; the method of using quaternion vectors to represent all pixel vectors of the hyperspectral image to be classified includes:

[0014] For each pixel vector on the hyperspectral image to be classified, the M spectral bands of the pixel vector are determined as corresponding h quaternion vectors, where To round up the sign, each of the first h-1 quaternion vectors corresponding to the pixel vector is used to represent the 4 consecutive spectral bands on the pixel vector and the radiation value corresponding to each spectral band, and the hth quaternion vector corresponding to the pixel vector is used to represent the M-4·h consecutive spectral bands on the pixel vector and the radiation value corresponding to each spectral band.

[0015] As an optional implementation, in the first aspect of the present invention, the training data set Q h×m The data category set R corresponding to the training data set C×m , determine the global quaternion divergence matrix S corresponding to the hyperspectral image to be classified P and the inter-class quaternion divergence matrix S b ,include:

[0016] According to the training data set Q h×m The data category set R corresponding to the training data set C×m , determine the test data matrix A te ∈Q h×(N-m) , training data matrix A t ∈Q h×m The category matrix C corresponding to the training data matrix t ∈R C×m ;

[0017] According to the training data matrix A t and the category matrix C t , determine the first intermediate matrix A g and the second intermediate matrix A b , and determine the global quaternion divergence matrix corresponding to the hyperspectral image to be classified and the inter-class quaternion divergence matrix

[0018] Among them, the first intermediate matrix A g and the second intermediate matrix A b Determined by the following formula:

[0019]

[0020]

[0021] In the above formula, The training data set Q h×m The mean vector of is the mean vector of all c-th quaternion vectors corresponding to the hyperspectral image to be classified, L∈R 1×m is a row vector with 1 row and m columns, each element of which is 1.

[0022] As an optional implementation, in the first aspect of the present invention, the global quaternion divergence matrix S is respectively g and the inter-class quaternion divergence matrix S b Perform quaternion eigenvalue decomposition, including:

[0023] The global quaternion divergence matrix S is calculated according to the following formula: g Perform quaternion eigenvalue decomposition:

[0024]

[0025] Among them, U g is the first decomposition matrix, D g is the second decomposition matrix;

[0026] The inter-class quaternion divergence matrix S is calculated according to the following formula: b Perform quaternion eigenvalue decomposition:

[0027]

[0028] in, is the global quaternion divergence matrix S g The corresponding matching matrix, U b is the third decomposition matrix, D b is the fourth decomposition matrix.

[0029] As an optional implementation, in the first aspect of the present invention, the global quaternion divergence matrix S g The decomposition result and the inter-class quaternion divergence matrix S b The decomposition result of is used to determine the target projection vector matrix corresponding to the hyperspectral image to be classified, including:

[0030] According to the global quaternion divergence matrix S g The decomposition result and the inter-class quaternion divergence matrix S b Decomposition result, determine the initial projection vector matrix

[0031] Determine the target projection vector matrix Here, q is a preset threshold, and the threshold q∈{1, 2, 3, ..., C-1}.

[0032] As an optional implementation, in the first aspect of the present invention, classifying the hyperspectral image to be classified according to the target projection vector matrix includes:

[0033] According to the target projection vector matrix Determine the quaternion classification feature in the spectral direction corresponding to the hyperspectral image to be classified

[0034] The quaternion classification feature A in the spectral direction f Perform spatial information processing to obtain the quaternion classification feature A in the spatial direction corresponding to the hyperspectral image to be classified s ;

[0035] According to the quaternion classification feature A in the spatial direction s The hyperspectral image to be classified is classified.

[0036] As an optional implementation, in the first aspect of the present invention, the hyperspectral image to be classified is a hyperspectral image corresponding to the target sample, and the method further includes:

[0037] According to the classification result obtained after classifying the hyperspectral image to be classified, multiple target chemical components on the target sample and the spatial distribution information corresponding to each of the target chemical components are determined.

[0038] The second aspect of the present invention discloses an HSI classification e-commerce application device based on LDA of quaternion rayleigh entropy, which is used to classify hyperspectral images to be classified in e-commerce, and the device includes:

[0039] The data conversion module is used to use quaternion vectors to represent all pixel vectors of the hyperspectral image to be classified, and obtain the quaternion set Q corresponding to the hyperspectral image to be classified. h×N , where N is the number of the pixel vectors, and h is the number of the quaternion vectors corresponding to each pixel vector;

[0040] A vector analysis module is used to determine m training pixel vectors from the N pixel vectors to obtain a training data set Q h×m The data category set R corresponding to the training data set C×m , where m is the preset number of training samples and C is the total number of spectral categories;

[0041] Divergence analysis module, used to analyze the training data set Q h×m The data category set R corresponding to the training data set C×m , determine the global quaternion divergence matrix S corresponding to the hyperspectral image to be classified g and the inter-class quaternion divergence matrix S b ;

[0042] The eigendecomposition module is used to respectively decompose the global quaternion divergence matrix S g and the inter-class quaternion divergence matrix S b Perform quaternion eigenvalue decomposition to obtain the global quaternion divergence matrix S P The decomposition result and the inter-class quaternion divergence matrix S b The decomposition results of

[0043] A projection decomposition module is used to calculate the global quaternion divergence matrix S according to the global quaternion divergence matrix S g The decomposition result and the inter-class quaternion divergence matrix S b Decomposition results of the target projection vector matrix corresponding to the hyperspectral image to be classified are determined;

[0044] An image analysis module is used to classify the hyperspectral image to be classified according to the target projection vector matrix.

[0045] As an optional implementation, in the second aspect of the present invention, the hyperspectral image to be classified includes N pixel vectors, each of the pixel vectors includes M spectral bands and the radiation value corresponding to each of the spectral bands, wherein N and M are both positive integers; the specific manner in which the data conversion module uses quaternion vectors to represent all pixel vectors of the hyperspectral image to be classified includes:

[0046] For each pixel vector on the hyperspectral image to be classified, the M spectral bands of the pixel vector are determined as corresponding h quaternion vectors, where To round up the sign, each of the first h-1 quaternion vectors corresponding to the pixel vector is used to represent the 4 consecutive spectral bands on the pixel vector and the radiation value corresponding to each spectral band, and the hth quaternion vector corresponding to the pixel vector is used to represent the M-4·h consecutive spectral bands on the pixel vector and the radiation value corresponding to each spectral band.

[0047] As an optional implementation, in the second aspect of the present invention, the divergence analysis module is based on the training data set Q h×m The data category set R corresponding to the training data set C×m, determine the global quaternion divergence matrix S corresponding to the hyperspectral image to be classified g and the inter-class quaternion divergence matrix S b Specific ways include:

[0048] According to the training data set Q h×m The data category set R corresponding to the training data set C×m , determine the test data matrix A te ∈Q h×(N-m) , training data matrix A t ∈Q h×m The category matrix C corresponding to the training data matrix t ∈R C×m ;

[0049] According to the training data matrix A t and the category matrix C t , determine the first intermediate matrix A P and the second intermediate matrix A b , and determine the global quaternion divergence matrix corresponding to the hyperspectral image to be classified and the inter-class quaternion divergence matrix

[0050] Among them, the first intermediate matrix A g and the second intermediate matrix A b Determined by the following formula:

[0051]

[0052]

[0053] In the above formula, The training data set Q h×m The mean vector of is the mean vector of all c-th quaternion vectors corresponding to the hyperspectral image to be classified, L∈R 1×m is a row vector with 1 row and m columns, each element of which is 1.

[0054] As an optional implementation, in the second aspect of the present invention, the eigendecomposition module respectively performs the global quaternion divergence matrix S g and the inter-class quaternion divergence matrix S b The specific methods for quaternion eigenvalue decomposition include:

[0055] The global quaternion divergence matrix S is calculated according to the following formula: g Perform quaternion eigenvalue decomposition:

[0056]

[0057] Among them, U g is the first decomposition matrix, D g is the second decomposition matrix;

[0058] The inter-class quaternion divergence matrix S is calculated according to the following formula: b Perform quaternion eigenvalue decomposition:

[0059]

[0060] in, is the global quaternion divergence matrix S g The corresponding matching matrix, U b is the third decomposition matrix, D b is the fourth decomposition matrix.

[0061] As an optional implementation, in the second aspect of the present invention, the projection decomposition module is based on the global quaternion divergence matrix S g The decomposition result and the inter-class quaternion divergence matrix S b The specific method of determining the target projection vector matrix corresponding to the hyperspectral image to be classified based on the decomposition result of , includes:

[0062] According to the global quaternion divergence matrix S g The decomposition result and the inter-class quaternion divergence matrix S b Decomposition result, determine the initial projection vector matrix

[0063] Determine the target projection vector matrix Here, q is a preset threshold, and the threshold q∈{1, 2, 3, ..., C-1}.

[0064] As an optional implementation, in the second aspect of the present invention, the specific manner in which the image analysis module classifies the hyperspectral image to be classified according to the target projection vector matrix includes:

[0065] According to the target projection vector matrix Determine the quaternion classification feature in the spectral direction corresponding to the hyperspectral image to be classified

[0066] The quaternion classification feature A in the spectral direction f Perform spatial information processing to obtain the quaternion classification feature A in the spatial direction corresponding to the hyperspectral image to be classified s ;

[0067] According to the quaternion classification feature A in the spatial direction sThe hyperspectral image to be classified is classified.

[0068] As an optional implementation, in the second aspect of the present invention, the hyperspectral image to be classified is a hyperspectral image corresponding to the target sample, and the device further includes:

[0069] The component analysis module is used to determine multiple target chemical components on the target sample and the spatial distribution information corresponding to each of the target chemical components according to the classification result obtained after classifying the hyperspectral image to be classified.

[0070] The third aspect of the present invention discloses another HSI classification e-commerce application device based on LDA of quaternion Rayleigh entropy, which is used to classify hyperspectral images to be classified in e-commerce, and the device includes:

[0071] A memory storing executable program code;

[0072] a processor coupled to the memory;

[0073] The processor calls the executable program code stored in the memory to execute the HSI classification method based on LDA of quaternion Rayleigh entropy disclosed in the first aspect of the present invention.

[0074] The fourth aspect of the present invention discloses a computer storage medium, which stores computer instructions. When the computer instructions are called, they are used to execute the HSI classification method based on LDA of quaternion Rayleigh entropy disclosed in the first aspect of the present invention.

[0075] Compared with the prior art, the present invention has the following beneficial effects:

[0076] The HSI classification method of the LDA based on quaternion rayleigh entropy of the present invention determines the target projection vector matrix corresponding to the hyperspectral image to be classified from the decomposition results of the global quaternion divergence matrix and the decomposition results of the inter-class quaternion divergence matrix. The inter-class quaternion divergence matrix is ​​helpful for effective data classification in the quaternion space, and the quaternion projection vector matrix expresses the characteristics of the quaternion in a specific direction. The present invention reduces the computational complexity and amount of calculation when using the quaternion discrimination method to process HSI classification, thereby improving the efficiency of HSI classification. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. 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 creative work.

[0078] Figure 1 It is a flow chart of an HSI classification method based on LDA of quaternion Rayleigh entropy disclosed in an embodiment of the present invention;

[0079] Figure 2 It is a structural schematic diagram of an HSI classification e-commerce application device based on LDA of quaternion Rayleigh entropy disclosed in an embodiment of the present invention;

[0080] Figure 3 It is a structural schematic diagram of another HSI classification e-commerce application device based on LDA of quaternion Rayleigh entropy disclosed in an embodiment of the present invention;

[0081] Figure 4 It is a structural schematic diagram of another HSI classification e-commerce application device based on LDA of quaternion Rayleigh entropy disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0082] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. 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 creative work are within the scope of protection of the present invention.

[0083] The terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish different objects rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, device, product or end including a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units that are not listed, or may optionally include other steps or units inherent to these processes, methods, products or ends.

[0084] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0085] The present invention discloses an HSI classification method and an e-commerce application device based on LDA of quaternion rayleigh entropy, which can reduce the computational complexity and amount when using quaternion discrimination method to process HSI classification, thereby improving the efficiency of HSI classification.

[0086] Embodiment 1

[0087] See also Figure 1 , Figure 1 : is a flow chart of an HSI classification method based on LDA of quaternion Rayleigh entropy disclosed in an embodiment of the present invention. Figure 1 The described HSI classification method based on quaternion Rayleigh entropy LDA can be applied to an HSI classification e-commerce application device based on quaternion Rayleigh entropy LDA, and the HSI classification e-commerce application device based on quaternion Rayleigh entropy LDA can be integrated in a cloud server or a local server, which is not limited in the embodiments of the present invention. Figure 1 As shown, the HSI classification method based on LDA of quaternion Rayleigh entropy may include the following operations:

[0088] Step 101: Use quaternion vectors to represent all pixel vectors of the hyperspectral image to be classified, and obtain the quaternion set Q corresponding to the hyperspectral image to be classified h×N .

[0089] In the embodiment of the present invention, N is the number of pixel vectors, and h is the number of quaternion vectors corresponding to each pixel vector; a quaternion is an extended complex number, including a real part and three imaginary parts, and can simultaneously represent multiple dimensions in space. Optionally, the spectral vector of each pixel can be mapped to the imaginary part of the quaternion, and the real part can be set to zero or a constant.

[0090] For example, the hyperspectral image to be classified may include N pixel vectors, each pixel vector includes M spectral bands and the radiation value corresponding to each spectral band, where N and M are both positive integers. In the embodiment of the present invention, further optionally, using quaternion vectors to represent all pixel vectors of the hyperspectral image to be classified may include:

[0091] For each pixel vector on the hyperspectral image to be classified, the M spectral bands of the pixel vector are determined as corresponding h quaternion vectors, where: To round up the sign, each of the first h-1 quaternion vectors corresponding to the pixel vector is used to represent the 4 consecutive spectral bands on the pixel vector and the radiation value corresponding to each spectral band, and the hth quaternion vector corresponding to the pixel vector is used to represent the M-4·h consecutive spectral bands on the pixel vector and the radiation value corresponding to each spectral band.

[0092] This embodiment determines the M spectral bands of each pixel vector as corresponding h quaternion vectors, wherein each quaternion vector represents four consecutive spectra in the M spectral bands of the corresponding pixel vector. For example, the pixel vector A is:

[0093] {(Band: infrared light; intensity: 21)(Band: red light 1; intensity: 15)(Band: red light 2; intensity: 14)(Band: yellow light; intensity: 45)(Band: green light; intensity: 33)(Band: blue light; intensity: 78)(Band: purple light; intensity: 27)(Band: ultraviolet light; intensity: 33)}

[0094] Represent the pixel vector A with a quaternion, then the four consecutive bands in the pixel vector A: {(band: infrared light; intensity: 21)(band: red light 1; intensity: 15)(band: red light 2; intensity: 14)(band: yellow light; intensity: 45)} are represented by a quaternion q1; the other four consecutive bands in the pixel vector A: {(band: green light; intensity: 33)(band: blue light; intensity: 78)(band: purple light; intensity: 27)(band: ultraviolet light; intensity: 33)} are represented by a quaternion q2.

[0095] It can be seen that this embodiment can more efficiently use quaternions to represent pixel vectors, thereby obtaining a more efficient and practical quaternion set corresponding to the hyperspectral image to be classified.

[0096] Step 102: Determine m training pixel vectors from the N pixel vectors to obtain a training data set Q h×m The data category set R corresponding to the training data set C×m .

[0097] In the embodiment of the present invention, m is the preset number of training samples, and C is the total number of spectral categories.

[0098] Step 103: Based on the training data set Q h×m The data category set R corresponding to the training data set C×m , determine the global quaternion divergence matrix S corresponding to the hyperspectral image to be classifiedg and the inter-class quaternion divergence matrix S b .

[0099] In the embodiment of the present invention, the quaternion divergence matrix is ​​used to describe the distribution characteristics of quaternion data. The global quaternion divergence matrix S P and the inter-class quaternion divergence matrix S b are two statistical tools for analyzing quaternion data sets.

[0100] Among them, the global quaternion divergence matrix S g It describes the overall distribution characteristics of all quaternion samples in the entire data set. It is usually constructed by calculating the covariance matrix of all sample quaternions, where each sample quaternion represents an observation. The global quaternion divergence matrix S g It helps to understand the geometric structure and distribution range of the entire data set and can be used for data preprocessing, feature extraction and dimensionality reduction.

[0101] Inter-class quaternion divergence matrix S b It is a matrix specifically used to describe the differences in quaternion samples between different categories. In supervised learning, the inter-class quaternion divergence matrix S b The inter-class quaternion divergence matrix S is constructed by calculating the difference between the center points (class means) of the quaternions of samples of different categories. b Helps to distinguish different categories of data.

[0102] Step 104: respectively calculate the global quaternion divergence matrix S P and the inter-class quaternion divergence matrix S b Perform quaternion eigenvalue decomposition to obtain the global quaternion divergence matrix S P The decomposition result and the inter-class quaternion divergence matrix S b The decomposition result.

[0103] In the embodiment of the present invention, the global quaternion divergence matrix S P and the inter-class quaternion divergence matrix S b Perform quaternion eigenvalue decomposition. Through eigenvalue decomposition, we can study the intrinsic characteristics of the quaternion matrix and obtain eigenvectors and eigenvalues.

[0104] Step 105: According to the global quaternion divergence matrix S P The decomposition result and the inter-class quaternion divergence matrix S b The decomposition result is used to determine the target projection vector matrix corresponding to the hyperspectral image to be classified.

[0105] Traditional methods such as principal component analysis (PCA) and linear discriminant analysis (LDA) are projection classification methods in the real number domain. They have limited vector expression capabilities and cannot solve the crosstalk of strong correlations between adjacent spectral bands, nor can they fully utilize the dimensional information of hyperspectral images. Linear discriminant analysis in the real number domain can be implemented by Rayleigh entropy or generalized eigenvalue decomposition based on 2-norm constraints. However, in the quaternion domain, since it cannot be guaranteed that the generalized eigenvalues ​​of the quaternion are real numbers, targeted design is required to solve the above problems. In this embodiment of the present invention, the global quaternion divergence matrix S is used to obtain the linear discriminant analysis. P The decomposition result and the inter-class quaternion divergence matrix S b The target projection vector matrix corresponding to the hyperspectral image to be classified is determined from the decomposition result, the inter-class quaternion divergence matrix is ​​helpful for effective data classification in the quaternion space, and the quaternion projection vector matrix expresses the characteristics of the quaternion in a specific direction. Therefore, the amount of calculation in the embodiment of the present invention is less, and the interference between adjacent spectra is reduced.

[0106] Step 106: classify the hyperspectral image to be classified according to the target projection vector matrix.

[0107] In an embodiment of the present invention, optionally, a spatial filtering algorithm such as 2DGF, 2DSSA or AWF can be used to process the spatial information of the standard projection vector matrix, and then input into a kernel extreme learning machine (KELM) or support vector machine (SVM) classifier to obtain a classification result of the hyperspectral image to be classified.

[0108] It can be seen that the HSI classification method based on LDA of quaternion rayleigh entropy in the embodiment of the present invention determines the target projection vector matrix corresponding to the hyperspectral image to be classified from the decomposition results of the global quaternion divergence matrix and the decomposition results of the inter-class quaternion divergence matrix. The inter-class quaternion divergence matrix is ​​helpful for effective data classification in the quaternion space, and the quaternion projection vector matrix expresses the characteristics of the quaternion in a specific direction. Therefore, the computational complexity and amount of calculation when using the quaternion discrimination method to process HSI classification are reduced, thereby improving the efficiency of HSI classification.

[0109] In an optional embodiment, according to the training data set Q h×m The data category set R corresponding to the training data set C×m , determine the global quaternion divergence matrix S corresponding to the hyperspectral image to be classified g and the inter-class quaternion divergence matrix S b , which may include:

[0110] According to the training data set Q h×m The data category set R corresponding to the training data set C×m , determine the test data matrix A te∈Q h×(N-m) , training data matrix A t ∈Q h×m The category matrix C corresponding to the training data matrix t ∈R C×m ;

[0111] According to the training data matrix A t and the category matrix C t , determine the first intermediate matrix A P and the second intermediate matrix A b , and determine the global quaternion divergence matrix corresponding to the hyperspectral image to be classified and the inter-class quaternion divergence matrix

[0112] Among them, the first intermediate matrix A g and the second intermediate matrix A b Determined by the following formula:

[0113]

[0114]

[0115] In the above formula, The training data set Q h×m The mean vector of is the mean vector of all c-th quaternion vectors corresponding to the hyperspectral image to be classified, L∈R 1×m is a row vector with 1 row and m columns, each element of which is 1.

[0116] In another optional embodiment, the global quaternion divergence matrix S g and the inter-class quaternion divergence matrix S b Perform quaternion eigenvalue decomposition, including:

[0117] The global quaternion divergence matrix S is calculated according to the following formula g Perform quaternion eigenvalue decomposition:

[0118]

[0119] Among them, U g is the first decomposition matrix, D g is the second decomposition matrix;

[0120] According to the following formula, the inter-class quaternion divergence matrix S b Perform quaternion eigenvalue decomposition:

[0121]

[0122] in, is the global quaternion divergence matrix S g The corresponding matching matrix, U b is the third decomposition matrix, D b is the fourth decomposition matrix.

[0123] In this optional embodiment, it is obvious that S b is a semi-positive definite Hermite quaternion matrix; when m> h When (m is the preset number of training samples, h is the number of quaternion vectors corresponding to each pixel vector), S g In general, it is a positive Hermite quaternion matrix; because the quaternion matrix does not satisfy the commutative law of multiplication, the S used in the solution of the generalized eigenvalue problem in the real number field g -1 S b is no longer a conjugate quaternion matrix, which will result in S g -1 S b The eigenvalue of is not a real number, so it is impossible to directly use the method of solving the generalized eigenvalue to obtain the discriminant projection vector.

[0124] Therefore, in this optional embodiment, because S b is a semi-positive quaternion Hermite matrix, so It is also a semi-positive quaternion Hermite matrix for Perform quaternion eigenvalue decomposition Where D b is a non-negative real diagonal matrix, that is, D b =(U b H ) -1 (U g D g 0.5 )S b ((U g D g 0.5 ) H ) -1 (U b ) -1 .

[0125] It can be seen that this optional embodiment proposes a new quaternion divergence matrix decomposition method, which is conducive to obtaining the discriminant projection vector by solving the generalized eigenvalue method.

[0126] Furthermore, the specific derivation process of this optional embodiment is as follows:

[0127] For S g The result of quaternion eigenvalue decomposition is When the quaternion matrix Sg When D is a positive Hermite quaternion matrix, g The diagonal elements of are all positive real numbers, and the diagonal matrix D is calculated. g The square root of the diagonal elements of g The square root matrix D g 0.5 , then the following formula can be used to calculate S g To break it down:

[0128]

[0129] Because S g is a reversible matrix, so U g D g 0.5 is a reversible quaternion matrix, and we can get the following quaternion Hermite matrix:

[0130]

[0131] Because S b is a semi-positive quaternion Hermite matrix, so It is also a semi-positive quaternion Hermite matrix for Perform quaternion eigenvalue decomposition Where D b is a non-negative real diagonal matrix, that is, D b =(U b H ) -1 (U g D g 0.5 ) -1 S b ((U g D g 0.5 ) H ) -1 (U b ) -1 .

[0132] In yet another optional embodiment, according to the global quaternion divergence matrix S g The decomposition result and the inter-class quaternion divergence matrix S b The decomposition result of the target projection vector matrix corresponding to the hyperspectral image to be classified can be determined by:

[0133] According to the global quaternion divergence matrix S g The decomposition result and the inter-class quaternion divergence matrix S b Decomposition result, determine the initial projection vector matrix

[0134] Determine the target projection vector matrix Wherein, q is a preset threshold, and the threshold q∈{1, 2, 3, …, C-1}.

[0135] The specific derivation process of this optional embodiment is as follows:

[0136] Let W s =((U g D g 0.5 ) H ) -1 (U b ) -1 , then

[0137] (W s ) H S b W s =D b

[0138] and:

[0139] (W s ) H S g W s =( U b H ) -1 (U g D g 0.5 ) -1 U g D g 0.5 (U g D g 0.5 ) H ((U g D g 0.5 ) H ) -1 (U b ) -1 =(U b H ) -1 (U b ) -1

[0140] Because U b is an orthogonal quaternion matrix, so:

[0141] (U b H ) -1 (U b ) -1 =(U b-1 ) -1 (U b ) -1 =U b (U b ) -1 =I, that is (W s ) -1 S g W s =I.

[0142] So the inter-class scatter matrix S obtained by the quaternion matrix bundle simultaneous diagonalization algorithm is b With the global divergence matrix S g The judgment result between them is:

[0143]

[0144] Due to S b The rank of is C-1, so take W s o =W s (:, 1:q) is used as the QDA quaternion projection vector matrix, where q∈{1, 2,…, C-1}, and the value of q can be preset in advance according to different situations.

[0145] In yet another optional embodiment, classifying the hyperspectral image to be classified according to the target projection vector matrix may include:

[0146] According to the target projection vector matrix Determine the quaternion classification features in the spectral direction corresponding to the hyperspectral image to be classified

[0147] Quaternion classification feature A in the spectral direction f Perform spatial information processing to obtain the quaternion classification feature A in the spatial direction corresponding to the hyperspectral image to be classified s ;

[0148] Quaternion classification feature A according to spatial direction s Classify the hyperspectral images to be classified.

[0149] In the embodiment of the present invention, the quaternion classification feature in the spectral direction corresponding to the hyperspectral image to be classified is Among them, the test data matrix A te ∈Q h×(N-m) , training data matrix A t ∈Q h×m , The target projection vector matrix is ​​used to reasonably represent all the features of the hyperspectral image to be classified, providing more efficient data support for subsequent classification.

[0150] In yet another optional embodiment, the hyperspectral image to be classified is a hyperspectral image corresponding to the target sample, and the method may further include:

[0151] According to the classification results obtained after classifying the hyperspectral image to be classified, multiple target chemical components on the target sample and the spatial distribution information corresponding to each target chemical component are determined.

[0152] An optional application scenario of this optional embodiment is: quality monitoring of agricultural products in e-commerce.

[0153] The rapid development of e-commerce has enabled agricultural products to be quickly sold to customers, but the quality of agricultural products sold online is mixed, and the quality of agricultural products cannot be effectively guaranteed. For example, the bright and shiny fruit on the surface may have been eaten by insects or rotted inside. For another example, there are many varieties of crop seeds in the seed market. The counterfeiting behavior of some illegal vendors makes it difficult to distinguish the authenticity of seeds, and seed buyers cannot distinguish the authenticity of seeds. Therefore, the detection and authenticity of agricultural product quality is an urgent problem to be solved. Most traditional detection methods are based on manual detection and supplemented by chemical analysis methods. These methods are time-consuming and laborious, and cannot achieve large-scale detection. Machine vision technology can identify the size, shape, color and texture of objects; but it is not helpful for detecting damage under the peel, nor can it be used to evaluate the internal quality of fruits. Hyperspectral imaging technology combines imaging technology and spectroscopy technology to obtain spatial and continuous spectral information from samples, which can greatly enhance the ability to identify and detect various chemical components and their spatial distribution of fruits, and judge the internal quality of fruits.

[0154] In another optional embodiment, the sample spectrum can be divided into a training sample spectrum and a test sample spectrum, and the training sample spectrum can be divided into agricultural products bitten by insects, infected by fungi such as aflatoxin, penicillium, glaucophorus, and black aspergillus, and normal agricultural products. This optional embodiment is more suitable for application scenarios when the number of training samples is greater than the quaternion dimension of the sample. In this application scenario, the HSI classification method of LDA based on quaternion rayleigh entropy in the embodiment of the present invention is used to obtain the projection vector matrix of the quaternion spectrum, and then the features in the spatial direction are extracted through the spatial information processing method; finally, the classification operation is performed using a classifier such as a kernel extreme learning machine (KELM) or a support vector machine. Traditional methods such as principal component analysis (PCA) and linear discriminant analysis (LDA) are projection classification methods in the real number domain. The expression ability of vectors is limited, and they cannot solve the crosstalk of strong correlations between adjacent spectral bands, nor can they make full use of the dimensional information of hyperspectral images. This optional embodiment can complete the feature extraction task more quickly and accurately when the number of training samples is greater than the quaternion dimension of HSI pixels.

[0155] Embodiment 2

[0156] See also Figure 2 , Figure 2 Schematic diagram of the structure of an HSI classification e-commerce application device based on LDA of quaternion Rayleigh entropy disclosed in an embodiment of the present invention. The device is used to classify hyperspectral images to be classified in e-commerce, such as Figure 2 As shown, the HSI classification e-commerce application device based on LDA of quaternion Rayleigh entropy may include:

[0157] The data conversion module 201 is used to use quaternion vectors to represent all pixel vectors of the hyperspectral image to be classified, and obtain the quaternion set Q corresponding to the hyperspectral image to be classified h×N , where N is the number of pixel vectors, and h is the number of quaternion vectors corresponding to each pixel vector;

[0158] The vector analysis module 202 is used to determine m training pixel vectors from N pixel vectors and obtain a training data set Q h×m The data category set R corresponding to the training data set C×m , where m is the preset number of training samples and C is the total number of spectral categories;

[0159] Divergence analysis module 203, used to analyze the training data set Q h×m The data category set R corresponding to the training data set C×m , determine the global quaternion divergence matrix S corresponding to the hyperspectral image to be classified g and the inter-class quaternion divergence matrix S b ;

[0160] The feature decomposition module 204 is used to respectively decompose the global quaternion scatter matrix S g and the inter-class quaternion divergence matrix S b Perform quaternion eigenvalue decomposition to obtain the global quaternion divergence matrix S P The decomposition result and the inter-class quaternion divergence matrix S b The decomposition results of

[0161] The projection decomposition module 205 is used to calculate the projection decomposition matrix S according to the global quaternion divergence matrix S g The decomposition result and the inter-class quaternion divergence matrix S b Decomposition results are used to determine the target projection vector matrix corresponding to the hyperspectral image to be classified;

[0162] The image analysis module 206 is used to classify the hyperspectral image to be classified according to the target projection vector matrix.

[0163] It can be seen that the HSI classification e-commerce application device based on LDA of quaternion rayleigh entropy in the embodiment of the present invention determines the target projection vector matrix corresponding to the hyperspectral image to be classified from the decomposition results of the global quaternion divergence matrix and the decomposition results of the inter-class quaternion divergence matrix. The inter-class quaternion divergence matrix is ​​helpful for effective data classification in the quaternion space, and the quaternion projection vector matrix expresses the characteristics of the quaternion in a specific direction. Therefore, the computational complexity and amount of calculation when using the quaternion discrimination method to process HSI classification are reduced, thereby improving the efficiency of HSI classification.

[0164] In an optional embodiment, the hyperspectral image to be classified includes N pixel vectors, each pixel vector includes M spectral bands and the radiation value corresponding to each spectral band, wherein N and M are both positive integers; the specific manner in which the data conversion module 201 uses quaternion vectors to represent all pixel vectors of the hyperspectral image to be classified may include:

[0165] For each pixel vector on the hyperspectral image to be classified, the M spectral bands of the pixel vector are determined as corresponding h quaternion vectors, where: To round up the sign, each of the first h-1 quaternion vectors corresponding to the pixel vector is used to represent the 4 consecutive spectral bands on the pixel vector and the radiation value corresponding to each spectral band, and the hth quaternion vector corresponding to the pixel vector is used to represent the M-4·h consecutive spectral bands on the pixel vector and the radiation value corresponding to each spectral band.

[0166] It can be seen that this embodiment can more efficiently use quaternions to represent pixel vectors, thereby obtaining a more efficient and practical quaternion set corresponding to the hyperspectral image to be classified.

[0167] In another optional embodiment, the divergence analysis module 203 calculates the training data set Q according to the training data set Q. h×m The data category set R corresponding to the training data set C×m , determine the global quaternion divergence matrix S corresponding to the hyperspectral image to be classified g and the inter-class quaternion divergence matrix S b Specific ways may include:

[0168] According to the training data set Q h×m The data category set R corresponding to the training data set C×m , determine the test data matrix A te ∈Q h×(N-m) , training data matrix A t ∈Q h×m The category matrix C corresponding to the training data matrix t ∈RC×m ;

[0169] According to the training data matrix A t and the category matrix C t , determine the first intermediate matrix A g and the second intermediate matrix A b , and determine the global quaternion divergence matrix corresponding to the hyperspectral image to be classified and the inter-class quaternion divergence matrix

[0170] Among them, the first intermediate matrix A g and the second intermediate matrix A b Determined by the following formula:

[0171]

[0172]

[0173] In the above formula, The training data set Q h×m The mean vector of is the mean vector of all c-th quaternion vectors corresponding to the hyperspectral image to be classified, L∈R 1×m is a row vector with 1 row and m columns, each element of which is 1.

[0174] In another optional embodiment, the eigendecomposition module 204 respectively performs the global quaternion divergence matrix S g and the inter-class quaternion divergence matrix S b Specific methods for performing quaternion eigenvalue decomposition may include:

[0175] The global quaternion divergence matrix S is calculated according to the following formula g Perform quaternion eigenvalue decomposition:

[0176]

[0177] Among them, U g is the first decomposition matrix, D g is the second decomposition matrix;

[0178] According to the following formula, the inter-class quaternion divergence matrix S b Perform quaternion eigenvalue decomposition:

[0179]

[0180] in, is the global quaternion divergence matrix S g The corresponding matching matrix, U b is the third decomposition matrix, Db is the fourth decomposition matrix.

[0181] It can be seen that this optional embodiment proposes a new quaternion divergence matrix decomposition method, which is conducive to obtaining the discriminant projection vector by solving the generalized eigenvalue method.

[0182] In another optional embodiment, the projection decomposition module 205 calculates the projection decomposition matrix S according to the global quaternion divergence matrix S g The decomposition result and the inter-class quaternion divergence matrix S b The specific method of determining the target projection vector matrix corresponding to the hyperspectral image to be classified based on the decomposition result may include:

[0183] According to the global quaternion divergence matrix S P The decomposition result and the inter-class quaternion divergence matrix S b Decomposition result, determine the initial projection vector matrix

[0184] Determine the target projection vector matrix Wherein, q is a preset threshold, and the threshold q∈{1, 2, 3, …, C-1}.

[0185] In yet another optional embodiment, the specific manner in which the image analysis module 206 classifies the hyperspectral image to be classified according to the target projection vector matrix may include:

[0186] According to the target projection vector matrix Determine the quaternion classification features in the spectral direction corresponding to the hyperspectral image to be classified

[0187] Quaternion classification feature A in the spectral direction f Perform spatial information processing to obtain the quaternion classification feature A in the spatial direction corresponding to the hyperspectral image to be classified s ;

[0188] Quaternion classification feature A according to spatial direction s Classify the hyperspectral images to be classified.

[0189] In the embodiment of the present invention, the quaternion classification feature in the spectral direction corresponding to the hyperspectral image to be classified is Among them, the test data matrix A te ∈Q h×(N-m) , training data matrix A t ∈Q h×m , The target projection vector matrix is ​​used to reasonably represent all the features of the hyperspectral image to be classified, providing more efficient data support for subsequent classification.

[0190] In yet another optional embodiment, Figure 3 As shown, the hyperspectral image to be classified is a hyperspectral image corresponding to the target sample, and the device may further include:

[0191] The component analysis module 207 is used to determine multiple target chemical components on the target sample and the spatial distribution information corresponding to each target chemical component according to the classification result obtained after classifying the hyperspectral image to be classified.

[0192] Embodiment 3

[0193] See also Figure 4 , Figure 4 is a schematic diagram of the structure of another HSI classification e-commerce application device based on LDA of quaternion Rayleigh entropy disclosed in an embodiment of the present invention. The device is used to classify hyperspectral images to be classified in e-commerce, such as Figure 4 As shown, the HSI classification e-commerce application device based on LDA of quaternion Rayleigh entropy may include:

[0194] A memory 301 storing executable program codes;

[0195] a processor 302 coupled to the memory 301;

[0196] The processor 302 calls the executable program code stored in the memory 301 to execute the steps of the HSI classification method based on LDA of quaternion Rayleigh entropy described in the first embodiment of the present invention.

[0197] Embodiment 4

[0198] The embodiment of the present invention discloses a computer storage medium, which stores computer instructions. When the computer instructions are called, they are used to execute the steps of the HSI classification method based on LDA of quaternion Rayleigh entropy described in the first embodiment of the present invention.

[0199] Embodiment 5

[0200] An embodiment of the present invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to enable a computer to execute the steps of the HSI classification method based on LDA of quaternion Rayleigh entropy described in Embodiment 1.

[0201] The device embodiments described above are only illustrative, wherein the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, i.e., they may be located in one place, or they may be distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Those of ordinary skill in the art may understand and implement it without creative work.

[0202] Through the specific description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the above technical solution can be essentially or partly contributed to the prior art in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, and the storage medium includes a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable rewritable read-only memory (EEPROM), a compact disc (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0203] Finally, it should be noted that the HSI classification method of LDA based on quaternion Rayleigh entropy and the e-commerce application device disclosed in the embodiment of the present invention are only the preferred embodiments of the present invention, which are only used to illustrate the technical scheme of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical schemes described in the aforementioned embodiments can still be modified, or some of the technical features therein can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical schemes from the spirit and scope of the technical schemes of the various embodiments of the present invention.

Claims

1. A HSI classification method based on LDA of quaternion Rayleigh entropy, characterized in that: The method is used to classify a hyperspectral image to be classified, and the method comprises: Use quaternion vectors to represent all pixel vectors of the hyperspectral image to be classified, and obtain the quaternion set Q corresponding to the hyperspectral image to be classified h×N , where N is the number of the pixel vectors, and h is the number of the quaternion vectors corresponding to each pixel vector; Determine m training pixel vectors from the N pixel vectors to obtain a training data set Q h×m The data category set R corresponding to the training data set C×m , where m is the preset number of training samples and C is the total number of spectral categories; According to the training data set Q h×m The data category set R corresponding to the training data set C×m , determine the global quaternion divergence matrix S corresponding to the hyperspectral image to be classified g and the inter-class quaternion divergence matrix S b ; The global quaternion divergence matrix S P and the inter-class quaternion divergence matrix S b Perform quaternion eigenvalue decomposition to obtain the global quaternion divergence matrix S g The decomposition result and the inter-class quaternion divergence matrix S b The decomposition results of According to the global quaternion divergence matrix S P The decomposition result and the inter-class quaternion divergence matrix S b Decomposition results of the target projection vector matrix corresponding to the hyperspectral image to be classified are determined; The hyperspectral image to be classified is classified according to the target projection vector matrix.

2. The HSI classification method based on LDA of quaternion rayleigh entropy according to claim 1, characterized in that: The hyperspectral image to be classified includes N pixel vectors, each of which includes M spectral bands and a radiation value corresponding to each spectral band, wherein N and M are both positive integers; the method of using quaternion vectors to represent all pixel vectors of the hyperspectral image to be classified includes: For each pixel vector on the hyperspectral image to be classified, the M spectral bands of the pixel vector are determined as corresponding h quaternion vectors, where To round up the sign, each of the first h-1 quaternion vectors corresponding to the pixel vector is used to represent the 4 consecutive spectral bands on the pixel vector and the radiation value corresponding to each spectral band, and the hth quaternion vector corresponding to the pixel vector is used to represent the M-4·h consecutive spectral bands on the pixel vector and the radiation value corresponding to each spectral band.

3. The HSI classification method based on LDA of quaternion rayleigh entropy according to claim 2, characterized in that: According to the training data set Q h×m The data category set R corresponding to the training data set C×m , determine the global quaternion divergence matrix S corresponding to the hyperspectral image to be classified g and the inter-class quaternion divergence matrix S b ,include: According to the training data set Q h×m The data category set R corresponding to the training data set C×m , determine the test data matrix At e ∈Q h×(N-m) , training data matrix A t ∈Q h×m The category matrix C corresponding to the training data matrix t ∈R C×m ; According to the training data matrix A t and the category matrix C t , determine the first intermediate matrix A g and the second intermediate matrix A b , and determine the global quaternion divergence matrix corresponding to the hyperspectral image to be classified and the inter-class quaternion divergence matrix Among them, the first intermediate matrix A g and the second intermediate matrix A b Determined by the following formula: In the above formula, The training data set Q h×m The mean vector of is the mean vector of all c-th quaternion vectors corresponding to the hyperspectral image to be classified, L∈R 1×m is a row vector with 1 row and m columns, each element of which is 1.

4. The HSI classification method based on LDA of quaternion Rayleigh entropy according to claim 3, characterized in that: The global quaternion divergence matrix S g and the inter-class quaternion divergence matrix S b Perform quaternion eigenvalue decomposition, including: The global quaternion divergence matrix S is calculated according to the following formula: g Perform quaternion eigenvalue decomposition: Among them, U g is the first decomposition matrix, D g is the second decomposition matrix; The inter-class quaternion divergence matrix S is calculated according to the following formula: b Perform quaternion eigenvalue decomposition: in, is the global quaternion divergence matrix S g The corresponding matching matrix, U b is the third decomposition matrix, D b is the fourth decomposition matrix.

5. The HSI classification method based on LDA of quaternion rayleigh entropy according to claim 4, characterized in that: The global quaternion divergence matrix S g The decomposition result and the inter-class quaternion divergence matrix S b The decomposition result of is used to determine the target projection vector matrix corresponding to the hyperspectral image to be classified, including: According to the global quaternion divergence matrix S g The decomposition result and the inter-class quaternion divergence matrix S b Decomposition result, determine the initial projection vector matrix Determine the target projection vector matrix Here, q is a preset threshold, and the threshold q∈{1, 2, 3, ..., C-1}.

6. The HSI classification method based on LDA of quaternion Rayleigh entropy according to claim 5, characterized in that: The classifying the hyperspectral image to be classified according to the target projection vector matrix includes: According to the target projection vector matrix Determine the quaternion classification feature in the spectral direction corresponding to the hyperspectral image to be classified The quaternion classification feature A in the spectral direction f Perform spatial information processing to obtain the quaternion classification feature A in the spatial direction corresponding to the hyperspectral image to be classified s ; According to the quaternion classification feature A in the spatial direction s The hyperspectral image to be classified is classified.

7. The HSI classification method based on LDA of quaternion Rayleigh entropy according to any one of claims 1 to 6, characterized in that: The hyperspectral image to be classified is a hyperspectral image corresponding to the target sample, and the method further includes: According to the classification result obtained after classifying the hyperspectral image to be classified, multiple target chemical components on the target sample and the spatial distribution information corresponding to each of the target chemical components are determined.

8. An HSI classification e-commerce application device based on LDA of quaternion rayleigh entropy, characterized in that: The device is used to classify hyperspectral images to be classified in e-commerce, and the device comprises: The data conversion module is used to use quaternion vectors to represent all pixel vectors of the hyperspectral image to be classified, and obtain the quaternion set Q corresponding to the hyperspectral image to be classified. h×N , where N is the number of the pixel vectors, and h is the number of the quaternion vectors corresponding to each pixel vector; A vector analysis module is used to determine m training pixel vectors from the N pixel vectors to obtain a training data set Q h×m The data category set R corresponding to the training data set C×m , where m is the preset number of training samples and C is the total number of spectral categories; Divergence analysis module, used to analyze the training data set Q h×m The data category set R corresponding to the training data set C×m , determine the global quaternion divergence matrix S corresponding to the hyperspectral image to be classified g and the inter-class quaternion divergence matrix S b ; The eigendecomposition module is used to respectively decompose the global quaternion divergence matrix S g and the inter-class quaternion divergence matrix S b Perform quaternion eigenvalue decomposition to obtain the global quaternion divergence matrix S P The decomposition result and the inter-class quaternion divergence matrix S b The decomposition results of A projection decomposition module is used to calculate the global quaternion divergence matrix S according to the global quaternion divergence matrix S g The decomposition result and the inter-class quaternion divergence matrix S b Decomposition results of the target projection vector matrix corresponding to the hyperspectral image to be classified are determined; An image analysis module is used to classify the hyperspectral image to be classified according to the target projection vector matrix.

9. An HSI classification e-commerce application device based on LDA of quaternion rayleigh entropy, characterized in that: The device is used to classify hyperspectral images to be classified in e-commerce, and the device includes: a memory storing executable program code; a processor coupled to the memory; the processor calls the executable program code stored in the memory to execute the HSI classification method based on LDA of quaternion rayleigh entropy as described in any one of claims 1-7.

10. A computer storage medium, characterized in that: The computer storage medium stores computer instructions, and when the computer instructions are called, they are used to execute the HSI classification method based on LDA of quaternion Rayleigh entropy as described in any one of claims 1 to 7.

Images