CN106022358A - Hyper-spectral image classification method and hyper-spectral image classification device - Google Patents

Abstract

The present invention provides a hyperspectral image classification method and device, wherein the method includes: acquiring a hyperspectral image and extracting various image features from the hyperspectral image; constructing training samples and test samples for the extracted various image features respectively ;Extract an adaptive shape neighborhood for each test sample, and construct an adaptive shape neighborhood matrix corresponding to the extracted adaptive shape neighborhood; calculate each adaptive shape neighborhood matrix, and obtain a multi-feature sparse coefficient matrix ; Reconstructing the test samples through the multi-feature sparse coefficient matrix and the training samples, and classifying the test samples. The hyperspectral image classification method and device provided by the embodiments of the present invention can improve classification accuracy.

Description

A hyperspectral image classification method and device

technical field

The present invention relates to the technical field of image processing, in particular to a hyperspectral image (Hyperspectral Image, HSI) classification method and device.

Background technique

At present, the continuous development of hyperspectral remote sensing technology has made hyperspectral image processing technology more and more widely used in various fields, and an important branch in the field of hyperspectral image processing is hyperspectral image classification technology. According to the different feedbacks of surface materials to the spectral signals of different bands of the hyperspectral sensor, the hyperspectral images can be classified, and the spectral curves of different materials can be depicted in the chart. The difference in the spectral curves provides the basis for the classification of ground objects.

In related technologies, when classifying hyperspectral images, spectral value features can be used as the basic basis for classification, and various other spatial features of hyperspectral images can be used to fully mine and utilize spatial information to classify hyperspectral images.

In the course of realizing the present invention, the inventor finds that there are at least the following problems in the prior art:

The above-mentioned hyperspectral image classification method has a large amount of calculation and a long operation time, and it is difficult to effectively use the correlation between multiple features, and it does not consider the differences between them.

Contents of the invention

In view of this, the purpose of the embodiments of the present invention is to provide a hyperspectral image classification method and device, so as to improve classification accuracy.

In a first aspect, an embodiment of the present invention provides a hyperspectral image classification method, including:

Obtaining a hyperspectral image and extracting multiple image features from the hyperspectral image, the multiple image features include: spectral value features, extended morphological profile features, Gabor texture features, and differential morphological profile features;

Constructing training samples and test samples respectively for the various image features extracted;

Extracting an adaptive shape neighborhood for each test sample, and constructing an adaptive shape neighborhood matrix corresponding to each of the extracted adaptive shape neighborhoods;

Calculating each of the adaptive shape neighborhood matrices to obtain a multi-feature sparse coefficient matrix;

The test sample is reconstructed by the multi-feature sparse coefficient matrix and the training sample, and the test sample is classified.

With reference to the first aspect, the embodiment of the present invention provides a first possible implementation manner of the first aspect, wherein:

The multiple described image features extracted are respectively constructed training samples, including:

Select a sample of a preset ratio from each of the various image features as a training sample, expressed as {D ^k } _k=1,2,3,4 ;

Among them, D ^k represents the training dictionary of the kth feature, C represents the total number of sample categories, N ^k represents the dimension of the sample vector of the kth image feature, M represents the total number of training samples, and R represents a real number.

With reference to the first aspect, the embodiment of the present invention provides a second possible implementation manner of the first aspect, wherein:

The extraction of an adaptive shape neighborhood for each test sample, and constructing an adaptive shape neighborhood matrix corresponding to each of the extracted adaptive shape neighborhoods, includes:

Extracting an adaptive shape neighborhood from each of the obtained test samples through a shape adaptive algorithm;

Constructing an adaptive shape neighborhood matrix corresponding to each of the extracted adaptive shape neighborhoods by using pixels in the extracted adaptive shape neighborhood

in, Represents the adaptive shape neighborhood matrix of the kth image feature, Represents the first sample vector in the adaptive shape neighborhood matrix of the kth image feature, and Γ represents the total number of pixels in the adaptive shape neighborhood.

With reference to the first aspect, the embodiment of the present invention provides a third possible implementation manner of the first aspect, wherein:

The calculation is performed on each of the adaptive shape neighborhood matrices to obtain a multi-feature sparse coefficient matrix, including:

The multi-feature sparse coefficient matrix is obtained by the following formula:

$\begin{array}{ccc}\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; S</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arg</span>}\underset{\mathrm{<span\; class="notranslate">\; S</span>}}{\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; no</span>}}\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; 4</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; S</span>}\mathrm{<span\; class="notranslate">\; A</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}{\mathrm{<span\; class="notranslate">\; S</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>{<span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; f</span>}}& \mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; .</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; .</span>& <span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; S</span>}<span\; class="notranslate">\; |</span>{<span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; p</span>}\mathrm{<span\; class="notranslate">\; t</span>}\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; v</span>}\mathrm{<span\; class="notranslate">\; e</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 0</span>}}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; K</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}\end{array}$

Among them, S and Represents the multi-feature sparse coefficient matrix, S ^k represents the sub-matrix belonging to the kth feature part in the multi-feature sparse coefficient matrix, F represents the Frobenius norm, ||S|| _{adaptive, 0} represents the adaptive matrix norm, K ₀ Indicates the sparsity.

In combination with the first aspect, the embodiment of the present invention provides a fourth possible implementation manner of the first aspect, wherein: the test sample is reconstructed through the multi-feature sparse coefficient matrix and the training sample, and the test sample is classified , comprising: reconstructing a test signal of multiple features of the test sample through the multi-feature sparse coefficient matrix and the training sample;

According to the test signal of multiple characteristics of the reconstructed test sample and the following formula, the category of the test sample is determined:

$\mathrm{<span\; class="notranslate">\; c</span>}\mathrm{<span\; class="notranslate">\; l</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arg</span>}\underset{\mathrm{<span\; class="notranslate">\; c</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; ...</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; C</span>}}{\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; no</span>}}\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; 4</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; S</span>}\mathrm{<span\; class="notranslate">\; A</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; c</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; S</span>}}}_{\mathrm{<span\; class="notranslate">\; c</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>{<span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; f</span>}}$

Among them, class(x) represents the category of the test sample, Denotes the test signal of the adaptively shaped neighborhood of the test sample in the reconstructed kth feature image, Represents the c-th class training sample in the k-th image feature, Indicates the part of the submatrix corresponding to the kth feature and the cth class sample in the multi-feature sparse coefficient matrix.

In the second aspect, the embodiment of the present invention also provides a hyperspectral image classification device, including:

An image acquisition module, configured to acquire a hyperspectral image and extract a variety of image features from the hyperspectral image, the various image features include: spectral value features, extended morphological profile features, Gabor texture features and differential morphological profiles feature;

A sample construction module, used to respectively construct training samples and test samples for the extracted multiple image features;

An image processing module, configured to extract an adaptive shape neighborhood for each test sample, and construct an adaptive shape neighborhood matrix corresponding to each of the extracted adaptive shape neighborhoods;

A calculation module, configured to calculate each of the adaptive shape neighborhood matrices to obtain a multi-feature sparse coefficient matrix;

A classification module, configured to reconstruct a test sample through the multi-feature sparse coefficient matrix and training samples, and classify the test sample.

In combination with the second aspect, the embodiment of the present invention provides a first possible implementation manner of the second aspect, wherein: the sample construction module includes:

A selection unit, configured to select a sample of a preset ratio from each of the various image features as a training sample, expressed as {D ^k } _k=1,2,3,4 ;

Among them, D ^k represents the training dictionary of the kth feature, C represents the total number of sample categories, N ^k represents the dimension of the sample vector of the kth image feature, M represents the total number of training samples, and R represents a real number.

With reference to the second aspect, the embodiment of the present invention provides a second possible implementation manner of the second aspect, wherein: the image processing module includes:

A computing unit is used to extract an adaptive shape neighborhood for each test sample through a shape adaptive algorithm;

A matrix construction unit, configured to construct an adaptive shape neighborhood matrix corresponding to each of the extracted adaptive shape neighborhoods by using pixels in the extracted adaptive shape neighborhood

in, Represents the adaptive shape neighborhood matrix of the kth image feature, Represents the first sample vector in the adaptive shape neighborhood matrix of the kth image feature, and Γ represents the total number of pixels in the adaptive shape neighborhood.

In combination with the second aspect, the embodiment of the present invention provides a third possible implementation manner of the second aspect, wherein: the calculation module includes:

The determination unit is used to obtain the multi-feature sparse coefficient matrix by the following formula:

$\begin{array}{ccc}\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; S</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arg</span>}\underset{\mathrm{<span\; class="notranslate">\; S</span>}}{\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; no</span>}}\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; 4</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; S</span>}\mathrm{<span\; class="notranslate">\; A</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}{\mathrm{<span\; class="notranslate">\; S</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>{<span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; f</span>}}& \mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; .</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; .</span>& <span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; S</span>}<span\; class="notranslate">\; |</span>{<span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; p</span>}\mathrm{<span\; class="notranslate">\; t</span>}\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; v</span>}\mathrm{<span\; class="notranslate">\; e</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 0</span>}}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; K</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}\end{array}$

Among them, S and Represents a multi-feature sparse coefficient matrix, S ^k represents the sub-matrix belonging to the k-th feature part in the multi-feature sparse coefficient matrix, F represents the Frobenius norm, ||S|| _{adaptive, 0} represents the adaptive matrix norm, K ₀ means sparsity.

In combination with the second aspect, the embodiment of the present invention provides a fourth possible implementation manner of the second aspect, wherein: the classification module includes:

a reconstruction unit, configured to reconstruct a test signal of multiple features of the test sample through the multi-feature sparse coefficient matrix and the training sample;

A classification unit, configured to determine the category of the test sample according to the reconstructed test signal of multiple features of the test sample and the following formula:

$\mathrm{<span\; class="notranslate">\; c</span>}\mathrm{<span\; class="notranslate">\; l</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arg</span>}\underset{\mathrm{<span\; class="notranslate">\; c</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; ...</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; C</span>}}{\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; no</span>}}\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; 4</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; S</span>}\mathrm{<span\; class="notranslate">\; A</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; c</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; S</span>}}}_{\mathrm{<span\; class="notranslate">\; c</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>{<span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; f</span>}}$

Among them, class(x) represents the category of the test sample, Denotes the test signal of the adaptively shaped neighborhood of the test sample in the reconstructed kth feature image, Represents the c-th class training sample in the k-th image feature, Indicates the part of the submatrix corresponding to the kth feature and the cth class sample in the multi-feature sparse coefficient matrix.

The hyperspectral image classification method and device provided in the embodiments of the present invention extract spectral value features, extended morphological profile features, Gabor texture features, and differential morphological profile features from the acquired hyperspectral image, and extract A variety of image features of the training samples and test samples are respectively constructed, and on this basis, the adaptive shape neighborhood matrix and the multi-feature sparse coefficient matrix corresponding to the adaptive shape neighborhood of each test sample are obtained, so as to pass each test The self-adaptive shape neighborhood matrix of the sample and the multi-feature sparse coefficient matrix of various image features classify the test samples. Neighborhood pixel blocks with higher similarity, and taking into account the similarity and difference between multiple features at the same time, improve the classification accuracy.

In order to make the above-mentioned objects, features and advantages of the present invention more comprehensible, preferred embodiments will be described in detail below together with the accompanying drawings.

Description of drawings

In order to illustrate the technical solutions of the embodiments of the present invention more clearly, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention, and thus It should be regarded as a limitation on the scope, and those skilled in the art can also obtain other related drawings based on these drawings without creative work.

FIG. 1 shows a flowchart of a hyperspectral image classification method provided by Embodiment 1 of the present invention;

Fig. 2 shows a schematic structural diagram of a hyperspectral image classification device provided by an embodiment of the present invention.

detailed description

In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions 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 It is a part of embodiments of the present invention, but not all embodiments. The components of the embodiments of the invention generally described and illustrated in the figures herein may be arranged and designed in a variety of different configurations. Accordingly, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative efforts belong to the protection scope of the present invention.

In related technologies, when classifying hyperspectral images, spectral value features can be used as the basic basis for classification, and various other spatial features of hyperspectral images can be used to fully mine and utilize spatial information to classify hyperspectral images. Based on this, the present application provides a hyperspectral image classification method and device.

Example 1

To facilitate the understanding of this embodiment, a hyperspectral image classification method disclosed in the embodiment of the present invention is first introduced in detail.

The hyperspectral image classification method provided in this embodiment is executed by a computing device, and the computing device acquires various image features in the hyperspectral image, and classifies the hyperspectral image according to the acquired various image features.

The computing device can use any existing computer or server to classify the hyperspectral image, which will not be repeated here.

Referring to Fig. 1, the present embodiment provides a hyperspectral image classification method, comprising the following steps:

Step 100, acquiring a hyperspectral image and extracting various image features from the hyperspectral image.

Among them, a variety of image features include: spectral value features, extended morphological profile features, Gabor texture features and differential morphological profile features.

In the above step 100, the original hyperspectral data is used as the spectral value feature, expressed as x ¹ ∈ R ^200×1 ; the extended morphological filtering algorithm is adopted (algorithm parameters: the number of opening and closing operations is 4, and the step size of the filter kernel is 2 pixel units) from the first, second and third principal component analysis diagrams of HSI to extract the morphological filter profile features, expressed as x ² ∈ R ^27×1 ; then, Gabor filters with different scales and phases Kernel (scale is [1,2,3,4,5,6,7,8,9,10] pixel units, phase is [30°,60°,90°,120°,150°,180°] ) extract the Gabor texture feature from the first principal component analysis graph of HSI, expressed as x ³ ∈ R ^60×1 ; use the differential morphological filtering algorithm (parameters: the kernel radius of the opening operation and the closing operation is [3,5, 7,9,11,13,15,17] pixel units) to extract the differential morphological profile features from the first, second, and third principal component analysis graphs of HSI, expressed as x ⁴ ∈ R ^48×1 .

Among them, the first, second, and third principal component analysis graphs are extracted from the original hyperspectral data.

Step 102, respectively constructing training samples and testing samples for the various extracted image features.

In step 102, the above-mentioned process of constructing training samples for the extracted multiple image features includes the following steps:

Select a sample of a preset ratio from each image feature in a variety of image features as a training sample, expressed as {D ^k } _k=1,2,3,4 ;

Among them, D ^k represents the training dictionary of the kth feature, C represents the total number of sample categories, N ^k represents the dimension of the sample vector of the kth image feature, M represents the total number of training samples, and R represents a real number.

The above sample categories refer to the types of substances that can be determined from hyperspectral images.

The above-mentioned number of test samples = the total number of samples of the image - M.

Step 104, extract an adaptive shape neighborhood for each test sample, and construct an adaptive shape neighborhood matrix corresponding to each extracted adaptive shape neighborhood.

The above-mentioned step 104 is to estimate the similarity between the test pixel and its neighbor pixels through the estimation method of local polynomial approximation, and extract the adaptive shape neighborhood for each test sample. The self-adaptive shape neighborhood is an octagonal area centered on itself, and can be expanded in eight directions with [1,2,3,5,7,9] pixel units.

Any existing local polynomial approximation estimation method can be used to extract the adaptive shape neighborhood for each test sample, which will not be repeated here.

Step 106: Calculate each adaptive shape neighborhood matrix to obtain a multi-feature sparse coefficient matrix.

In the above step 106, the multi-feature sparse coefficient matrix is obtained by the following formula:

$\begin{array}{ccc}\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; S</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arg</span>}\underset{\mathrm{<span\; class="notranslate">\; S</span>}}{\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; no</span>}}\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; 4</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; S</span>}\mathrm{<span\; class="notranslate">\; A</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}{\mathrm{<span\; class="notranslate">\; S</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>{<span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; f</span>}}& \mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; .</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; .</span>& <span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; S</span>}<span\; class="notranslate">\; |</span>{<span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; p</span>}\mathrm{<span\; class="notranslate">\; t</span>}\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; v</span>}\mathrm{<span\; class="notranslate">\; e</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 0</span>}}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; K</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}\end{array}$

Among them, S and Represents a multi-feature sparse coefficient matrix, S ^k represents the sub-matrix belonging to the k-th feature part in the multi-feature sparse coefficient matrix, F represents the Frobenius norm, ||S|| _{adaptive, 0} represents the adaptive matrix norm, K ₀ means sparsity. in, Indicates the final solution S.

Specifically, ||S|| _{adaptive, 0} is an adaptive matrix norm, and several adaptive sets in S can be selected. Each adaptive set represents the index of the non-zero coefficient in the multi-feature sparse coefficient, and these indices are different, but constrained to be in the same class of multi-featured sparse coefficient matrices.

Step 108: Reconstruct the test sample through the multi-feature sparse coefficient matrix and the training sample, and classify the test sample.

The above step 108 specifically includes the following steps (1) to (2):

(1) Reconstruct the test signal of multiple features of the test sample through the multi-feature sparse coefficient matrix and the training sample;

(2) Determine the category of the test sample according to the test signal of multiple features of the reconstructed test sample and the following formula:

$\mathrm{<span\; class="notranslate">\; c</span>}\mathrm{<span\; class="notranslate">\; l</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arg</span>}\underset{\mathrm{<span\; class="notranslate">\; c</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; ...</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; C</span>}}{\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; no</span>}}\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; 4</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; S</span>}\mathrm{<span\; class="notranslate">\; A</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; c</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; S</span>}}}_{\mathrm{<span\; class="notranslate">\; c</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>{<span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; f</span>}}$

Among them, class(x) represents the category of the test sample, Denotes the test signal of the adaptively shaped neighborhood of the test sample in the reconstructed kth feature image, Represents the c-th class training sample in the k-th image feature, Indicates the part of the submatrix corresponding to the kth feature and the cth class sample in the multi-feature sparse coefficient matrix.

In step (1), the part belonging to a certain class in each training sample is multiplied by the part belonging to this feature and this class in the multi-feature sparse coefficient matrix, and the test signal of this class and this feature can be reconstructed. Test signals for other traits and other classes can be refactored.

In summary, the hyperspectral image classification method provided in this embodiment extracts spectral value features, extended morphological profile features, Gabor texture features, and differential morphological profile features from the acquired hyperspectral image, Construct training samples and test samples for the various image features extracted, and on this basis, get the neighborhood moment and matrix feature sparse coefficient matrix corresponding to the adaptive shape neighborhood of each test sample, so that each self-adaptive Adapt to the neighborhood matrix corresponding to the shape neighborhood and the multi-feature sparse coefficient matrix of various image features to classify the test samples. Compared with the hyperspectral image classification method proposed in the prior art, it can adaptively select and be tested Neighborhood pixel blocks with higher pixel similarity, and taking into account the similarity and difference between multiple features at the same time, improve the classification accuracy.

In related technologies, in order to obtain the neighborhood matrix corresponding to each test sample, the neighborhood pixel blocks are selected across the edge or across categories through the fixed window neighborhood, but the selected neighborhood pixel blocks are located at the edge of the image or different categories The connection between the selected neighborhood pixel blocks and the test samples will lead to a low similarity, thus reducing the classification accuracy of the hyperspectral image. Therefore, in order to improve the classification accuracy of hyperspectral images, in the hyperspectral image classification method proposed in this embodiment, an adaptive shape neighborhood is extracted from each of the multiple test samples constructed, and the construction and extraction The neighborhood matrix corresponding to each test sample of , including the following steps (1) to (2):

(1) Extract adaptive shape neighborhoods from each test sample obtained by a shape adaptive algorithm;

(2) Construct a neighborhood matrix corresponding to each extracted adaptive shape neighborhood through the pixels in the extracted adaptive neighborhood

in, Represents the adaptive shape neighborhood matrix of the kth image feature, Represents the first sample vector in the adaptive shape neighborhood matrix of the kth image feature, and Γ represents the total number of pixels in the adaptive shape neighborhood.

To sum up, by using the shape adaptive algorithm to select the neighborhood pixel blocks with higher similarity with the pixels in the adaptive neighborhood, it can well avoid the crossing of the fixed window neighborhood at the edge of the image or at the junction of different categories. The lack of edge or cross-category selection of neighborhood pixel blocks improves the classification accuracy of hyperspectral images.

Example 2:

Referring to FIG. 2, this embodiment provides a hyperspectral image classification device for performing the above hyperspectral image classification method, including:

The image acquisition module 200 is configured to acquire a hyperspectral image and extract various image features from the hyperspectral image. The various image features include: spectral value features, extended morphological profile features, Gabor texture features, and differential morphological profile features;

The sample construction module 202 is used to respectively construct training samples and test samples for various image features extracted;

The image processing module 204 is used to extract adaptive shape neighborhoods for each test sample, and construct an adaptive shape neighborhood matrix corresponding to each extracted adaptive shape neighborhood;

Calculation module 206, is used for calculating each self-adaptive shape neighborhood matrix, obtains multi-feature sparse coefficient matrix;

The classification module 208 is configured to reconstruct the test samples through the multi-feature sparse coefficient matrix and the training samples, and classify the test samples.

Specifically, the sample construction module 202 includes:

The selection unit is used to select a sample of a preset ratio from each image feature in various image features as a training sample, expressed as {D ^k } _k=1,2,3,4 ;

Among them, D ^k represents the training dictionary of the kth feature, C represents the total number of sample categories, N ^k represents the dimension of the sample vector of the kth image feature, M represents the total number of training samples, and R represents a real number.

Calculation module 206, comprising: a determination unit, used to obtain the multi-feature sparse coefficient matrix by the following formula:

$\begin{array}{ccc}\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; S</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arg</span>}\underset{\mathrm{<span\; class="notranslate">\; S</span>}}{\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; no</span>}}\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; 4</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; S</span>}\mathrm{<span\; class="notranslate">\; A</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}{\mathrm{<span\; class="notranslate">\; S</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>{<span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; f</span>}}& \mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; .</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; .</span>& <span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; S</span>}<span\; class="notranslate">\; |</span>{<span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; p</span>}\mathrm{<span\; class="notranslate">\; t</span>}\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; v</span>}\mathrm{<span\; class="notranslate">\; e</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 0</span>}}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; K</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}\end{array}$

Among them, S and Represents the multi-feature sparse coefficient matrix, S ^k represents the sub-matrix belonging to the kth feature part in the multi-feature sparse coefficient matrix, F represents the Frobenius norm, ||S|| _{adaptive, 0} represents the adaptive matrix norm, K ₀ Indicates the sparsity.

The classification module 208 includes: a reconstruction unit, used to reconstruct the test signal of multiple features of the test sample through the multi-feature sparse coefficient matrix and the training sample;

The classification unit is used to determine the category of the test sample according to the test signal of multiple characteristics of the reconstructed test sample and the following formula:

$\mathrm{<span\; class="notranslate">\; c</span>}\mathrm{<span\; class="notranslate">\; l</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arg</span>}\underset{\mathrm{<span\; class="notranslate">\; c</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; ...</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; C</span>}}{\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; no</span>}}\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; 4</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; S</span>}\mathrm{<span\; class="notranslate">\; A</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; c</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; S</span>}}}_{\mathrm{<span\; class="notranslate">\; c</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>{<span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; f</span>}}$

Among them, class(x) represents the category of the test sample, Denotes the test signal of the adaptively shaped neighborhood of the test sample in the reconstructed kth feature image, Represents the c-th class training sample in the k-th image feature, Indicates the part of the submatrix corresponding to the kth feature and the cth class sample in the multi-feature sparse coefficient matrix.

In related technologies, in order to obtain the neighborhood matrix corresponding to each test sample, the neighborhood pixel blocks are selected across the edge or across categories through the fixed window neighborhood, but the selected neighborhood pixel blocks are located at the edge of the image or different categories The connection between the selected neighborhood pixel blocks and the test samples will lead to a low similarity, thus reducing the classification accuracy of the hyperspectral image. Therefore, in order to improve the classification accuracy of hyperspectral images, in the hyperspectral image classification device proposed in this embodiment, the image processing module 204 includes:

A computing unit is used to extract an adaptive shape neighborhood for each obtained test sample through a shape adaptive algorithm;

A matrix construction unit for constructing an adaptive shape neighborhood matrix corresponding to each extracted adaptive shape neighborhood by using pixels in the extracted adaptive neighborhood

in, Represents the adaptive shape neighborhood matrix of the kth image feature, Represents the first sample vector in the adaptive shape neighborhood matrix of the kth image feature, and Γ represents the total number of pixels in the adaptive shape neighborhood.

To sum up, by using the shape adaptive algorithm to select the neighborhood pixel blocks with higher similarity with the pixels in the adaptive neighborhood, it can well avoid the crossing of the fixed window neighborhood at the edge of the image or at the junction of different categories. The lack of edge or cross-category selection of neighborhood pixel blocks improves the classification accuracy of hyperspectral images.

In summary, the hyperspectral image classification device provided in this embodiment extracts spectral value features, extended morphological profile features, Gabor texture features, and differential morphological profile features from the acquired hyperspectral image, Construct training samples and test samples for the various image features extracted, and on this basis, obtain the adaptive shape neighborhood matrix and multi-feature sparse coefficient matrix corresponding to the adaptive shape neighborhood of each test sample, so that through each An adaptive shape neighborhood matrix and a multi-feature sparse coefficient matrix of various image features are used to classify the test samples. Compared with the hyperspectral image classification method proposed in the prior art, it can adaptively select pixels similar to the tested pixels Neighborhood pixel blocks with higher degree, and consider the similarity and difference between multiple features at the same time, improve the classification accuracy.

The computer program product for performing the hyperspectral image classification method provided by the embodiments of the present invention includes a computer-readable storage medium storing program codes, and the instructions included in the program codes can be used to execute the methods described in the foregoing method embodiments, For specific implementation, reference may be made to the method embodiments, which will not be repeated here.

Those skilled in the art can clearly understand that for the convenience and brevity of the description, the specific working process of the above-described system, device and unit can refer to the corresponding process in the foregoing method embodiment, which will not be repeated here.

In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods may be implemented in other ways. The device embodiments described above are only illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or May be integrated into another system, or some features may be ignored, or not implemented. In another point, the mutual coupling or direct coupling or communication connection shown or discussed may be through some communication interfaces, and the indirect coupling or communication connection of devices or units may be in electrical, mechanical or other forms.

The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, each unit may exist separately physically, or two or more units may be integrated into one unit.

If the functions described above are realized in the form of software function units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the essence of the technical solution of the present invention or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including Several instructions are used to make a computer device (which may be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disk or optical disk and other media that can store program codes. .

The above is only a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Anyone skilled in the art can easily think of changes or substitutions within the technical scope disclosed in the present invention. Should be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims (10)

1. A hyperspectral image classification method, characterized in that, comprising: Obtaining a hyperspectral image and extracting multiple image features from the hyperspectral image, the multiple image features include: spectral value features, extended morphological profile features, Gabor texture features, and differential morphological profile features; Constructing training samples and test samples respectively for the various image features extracted; Extract adaptive shape neighborhoods for each test sample, and construct adaptive shape neighborhood matrices corresponding to the extracted adaptive shape neighborhoods; Calculating each of the adaptive shape neighborhood matrices to obtain a multi-feature sparse coefficient matrix; The test samples are reconstructed through the multi-feature sparse coefficient matrix and the training samples, and the test samples are classified. 2. method according to claim 1, is characterized in that, described constructing training sample respectively to multiple described image features of extraction, comprising: Select a sample of a preset ratio from each of the various image features as a training sample, expressed as {D ^k } _k=1,2,3,4 ; Among them, D ^k represents the training dictionary of the kth feature, C represents the total number of sample categories, N ^k represents the dimension of the sample vector of the kth image feature, M represents the total number of training samples, and R represents a real number. 3. The method according to claim 2, wherein the adaptive shape neighborhood is extracted for each test sample, and an adaptive shape neighborhood corresponding to each of the extracted adaptive shape neighborhoods is constructed Matrix, including: Extracting an adaptive shape neighborhood for each of the obtained test samples through a shape adaptive algorithm; Constructing an adaptive shape neighborhood matrix corresponding to each of the extracted adaptive shape neighborhoods by using pixels in the extracted adaptive shape neighborhood in, Represents the adaptive shape neighborhood matrix of the kth image feature, Represents the first sample vector in the adaptive shape neighborhood matrix of the kth image feature, and Γ represents the total number of pixels in the adaptive shape neighborhood. 4. The method according to claim 3, wherein said calculating each said adaptive shape neighborhood matrix obtains a multi-feature sparse coefficient matrix, comprising: The multi-feature sparse coefficient matrix is obtained by the following formula: $\begin{array}{ccc}\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; S</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arg</span>}\underset{\mathrm{<span\; class="notranslate">\; S</span>}}{\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; no</span>}}\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; 4</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; S</span>}\mathrm{<span\; class="notranslate">\; A</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}{\mathrm{<span\; class="notranslate">\; S</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>{<span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; f</span>}}& \mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; .</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; .</span>& <span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; S</span>}<span\; class="notranslate">\; |</span>{<span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; p</span>}\mathrm{<span\; class="notranslate">\; t</span>}\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; v</span>}\mathrm{<span\; class="notranslate">\; e</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 0</span>}}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; K</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}\end{array}$ Among them, S and Represents the multi-feature sparse coefficient matrix, S ^k represents the sub-matrix belonging to the kth feature part in the multi-feature sparse coefficient matrix, F represents the Frobenius norm, ||S|| _{adaptive, 0} represents the adaptive matrix norm, K ₀ Indicates the sparsity. 5. The method according to claim 4, wherein said reconstructing test samples through said multi-feature sparse coefficient matrix and training samples, and classifying said test samples comprises: using said multi-feature sparse coefficient matrix Reconstructing a test signal of multiple features of the test sample with the matrix and the training sample; According to the test signal of multiple characteristics of the reconstructed test sample and the following formula, the category of the test sample is determined: $\mathrm{<span\; class="notranslate">\; c</span>}\mathrm{<span\; class="notranslate">\; l</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arg</span>}\underset{\mathrm{<span\; class="notranslate">\; c</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; ...</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; C</span>}}{\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; no</span>}}\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; 4</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; S</span>}\mathrm{<span\; class="notranslate">\; A</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; c</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; S</span>}}}_{\mathrm{<span\; class="notranslate">\; c</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>{<span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; f</span>}}$ Among them, class(x) represents the category of the test sample, Denotes the test signal of the adaptively shaped neighborhood of the test sample in the reconstructed kth feature image, Represents the c-th class training sample in the k-th image feature, Indicates the part of the submatrix corresponding to the kth feature and the cth class sample in the multi-feature sparse coefficient matrix. 6. A hyperspectral image classification device, characterized in that, comprising: An image acquisition module, configured to acquire a hyperspectral image and extract a variety of image features from the hyperspectral image, the various image features include: spectral value features, extended morphological profile features, Gabor texture features and differential morphological profiles feature; A sample construction module, used to respectively construct training samples and test samples for the extracted multiple image features; An image processing module for extracting an adaptive shape neighborhood for each test sample, and constructing an adaptive shape neighborhood matrix corresponding to each of the extracted adaptive shape neighborhoods; A calculation module, configured to calculate each of the adaptive shape neighborhood matrices to obtain a multi-feature sparse coefficient matrix; The classification module is used for reconstructing the test samples from the multi-feature sparse coefficient matrix and the training samples, and classifying the test samples. 7. The device according to claim 6, wherein the sample building module comprises: A selection unit, configured to select a sample of a preset ratio from each of the various image features as a training sample, expressed as {D ^k } _k=1,2,3,4 ; Among them, D ^k represents the training dictionary of the kth feature, C represents the total number of sample categories, N ^k represents the dimension of the sample vector of the kth image feature, M represents the total number of training samples, and R represents a real number. 8. The device according to claim 7, wherein the image processing module comprises: A computing unit, configured to extract an adaptive shape neighborhood for each of the obtained test samples through a shape adaptive algorithm; A matrix construction unit, configured to construct an adaptive shape neighborhood matrix corresponding to each of the extracted adaptive shape neighborhoods by using pixels in the extracted adaptive shape neighborhood in, Represents the adaptive shape neighborhood matrix of the kth image feature, Represents the first sample vector in the adaptive shape neighborhood matrix of the kth image feature, and Γ represents the total number of pixels in the adaptive shape neighborhood. 9. The device according to claim 8, wherein the computing module comprises: The determination unit is used to obtain the multi-feature sparse coefficient matrix by the following formula: $\begin{array}{ccc}\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; S</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arg</span>}\underset{\mathrm{<span\; class="notranslate">\; S</span>}}{\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; no</span>}}\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; 4</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; S</span>}\mathrm{<span\; class="notranslate">\; A</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}{\mathrm{<span\; class="notranslate">\; S</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>{<span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; f</span>}}& \mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; .</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; .</span>& <span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; S</span>}<span\; class="notranslate">\; |</span>{<span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; p</span>}\mathrm{<span\; class="notranslate">\; t</span>}\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; v</span>}\mathrm{<span\; class="notranslate">\; e</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 0</span>}}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; K</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}\end{array}$ Among them, S and Represents the multi-feature sparse coefficient matrix, S ^k represents the sub-matrix belonging to the kth feature part in the multi-feature sparse coefficient matrix, F represents the Frobenius norm, ||S|| _{adaptive, 0} represents the adaptive matrix norm, K ₀ Indicates the sparsity. 10. The device according to claim 9, wherein the classification module comprises: a reconstruction unit, configured to reconstruct a test signal of multiple features of the test sample through the multi-feature sparse coefficient matrix and the training sample; A classification unit, configured to determine the category of the test sample according to the reconstructed test signal of multiple features of the test sample and the following formula: $\mathrm{<span\; class="notranslate">\; c</span>}\mathrm{<span\; class="notranslate">\; l</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arg</span>}\underset{\mathrm{<span\; class="notranslate">\; c</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; ...</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; C</span>}}{\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; no</span>}}\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; 4</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; S</span>}\mathrm{<span\; class="notranslate">\; A</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; c</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; S</span>}}}_{\mathrm{<span\; class="notranslate">\; c</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>{<span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; f</span>}}$ Among them, class(x) represents the category of the test sample, Denotes the test signal of the adaptively shaped neighborhood of the test sample in the reconstructed kth feature image, Represents the c-th class training sample in the k-th image feature, Indicates the part of the submatrix corresponding to the kth feature and the cth class sample in the multi-feature sparse coefficient matrix.