Superpixel-based enhanced anchor graph for large hyperspectral image clustering

Abstract

The application discloses a large-scale hyperspectral image clustering method, system and equipment based on superpixel enhanced anchor point graph, and relates to the technical field of hyperspectral image processing. The hyperspectral image has the characteristics of large data scale and containing complex noise. In view of the problems of high calculation complexity, insufficient precision and robustness of the existing method, the application provides a large-scale hyperspectral image clustering method, system and equipment based on superpixel enhanced anchor point graph. The application improves the clustering efficiency and precision of large-scale hyperspectral image by integrating the superpixel enhanced anchor point graph of superpixel-level spatial-spectral information. The superpixel enhanced anchor point graph is decomposed under the correlation entropy criterion, and the robustness of the hyperspectral clustering task containing complex noise is improved. A large number of experiments show that the method of the application is significantly better than the prior art in clustering efficiency, precision and robustness, and provides an effective solution for large-scale hyperspectral image clustering, and can be used in the fields of precision agriculture, environmental monitoring, military reconnaissance, mineral exploration and the like.

Description

Technical Field

[0001] This invention relates to the field of hyperspectral image processing technology, and in particular to a method, system and device for large-scale hyperspectral image clustering based on superpixel-enhanced anchor point maps. Background Technology

[0002] Hyperspectral image clustering plays a crucial role in numerous fields, enabling the extraction of latent information from objects within images and providing strong support for subsequent image analysis. It is widely applied in precision agriculture, environmental monitoring, military reconnaissance, and mineral exploration.

[0003] Hyperspectral images typically contain tens of thousands or even millions of pixels. Traditional graph clustering requires calculating the similarity of all pixel pairs, resulting in a quadratic computational complexity. Furthermore, the spectral decomposition of large-scale graph structures further exacerbates the computational burden, making these methods difficult to deploy in time-sensitive applications.

[0004] Hyperspectral images possess rich spectral information, but the spatial correlation between pixels is crucial for improving clustering accuracy. Most existing methods rely solely on spectral similarity metrics, completely ignoring the inherent spatial structure of hyperspectral images. Although some recent methods have incorporated spatial constraints into graph learning, such as including neighborhood relationships or employing spatial regularization, the results remain far from satisfactory, especially for large-scale hyperspectral images with complex structural patterns.

[0005] Hyperspectral images are inevitably affected by factors such as sensor defects, atmospheric influences, and data transmission errors, resulting in complex noise such as Gaussian noise, impulse noise, stripe noise, and mixed noise distribution. Existing noise reduction methods include those based on the l1 norm and l... 2,1 Norm-based methods are only effective for sparse or simple noise and are difficult to effectively deal with nonlinear, non-Gaussian, or mixed noise distributions.

[0006] In summary, existing hyperspectral image clustering methods have significant shortcomings in improving clustering efficiency, clustering accuracy, and resistance to noise interference. They are unable to cope with the large scale and complex noise characteristics of hyperspectral images, and there is an urgent need for a new method to solve these problems. Summary of the Invention

[0007] The purpose of this invention is to provide a method, system and device for large-scale hyperspectral image clustering based on superpixel enhanced anchor point maps, so as to solve the problems of low clustering efficiency, insufficient clustering accuracy and poor clustering robustness in the existing large-scale hyperspectral image clustering.

[0008] A large-scale hyperspectral image clustering method based on superpixel-enhanced anchor point maps includes the following steps:

[0009] Superpixel segmentation: Perform superpixel segmentation on hyperspectral images to divide the hyperspectral images into a series of non-overlapping superpixel regions containing highly homogeneous pixels, generating a superpixel spectral set;

[0010] Construction of Superpixel Spatial Spectrum Enhancement Set: For the superpixel spectral set, calculate the average spectral information of the neighbors of the superpixel point, and obtain the superpixel spatial spectrum enhancement set by fusing it with the original features of the superpixel point. Each superpixel feature in the superpixel spatial spectrum enhancement set contains two parts of information: spectral information and neighborhood spatial information.

[0011] Superpixel enhancement anchor map construction: Select anchor points for the superpixel spectral set and the superpixel spatial spectral enhancement set respectively to obtain the superpixel spectral anchor point set and the superpixel spatial spectral enhancement anchor point set. Then, construct the superpixel enhancement anchor map by jointly learning the relationship between the original data and anchor points in the spectral features and spatial spectral features.

[0012] Non-negative matrix decomposition based on correlation entropy: The constructed superpixel enhancement anchor point map is decomposed into a non-negative matrix based on the correlation entropy criterion. At the same time, graph regularization constraints are applied to the decomposition process, and finally the category of superpixel spectral points is obtained directly from the non-negative matrix obtained by decomposition.

[0013] Hyperspectral point clustering result generation: Hyperspectral points are mapped by superpixel spectral points to obtain the clustering result of all pixels in the final hyperspectral image.

[0014] The large-scale hyperspectral image clustering method based on superpixel enhanced anchor point map applies an entropy rate superpixel segmentation algorithm based on graph theory to the input large-scale hyperspectral image data to divide the superpixel regions and generate a superpixel spectral set.

[0015] The large-scale hyperspectral image clustering method based on superpixel enhancement anchor point map calculates the average spectral information of its neighbors for each superpixel point in the superpixel spectral set, and concatenates the average spectral information of the neighbors with the superpixel information to obtain the superpixel spatial spectral enhancement set.

[0016] The large-scale hyperspectral image clustering method based on superpixel enhancement anchor maps describes the following steps: First, a k-means algorithm is used to obtain a superpixel spectral anchor set from the superpixel spectral set. Then, based on the obtained superpixel spectral anchor set, a mapping calculation is performed on the superpixel spatial-spectral enhancement set to obtain the superpixel spatial-spectral enhancement anchor set corresponding to the superpixel spatial-spectral enhancement set. Subsequently, Euclidean distance is used to measure the structural relationships between the superpixel spectral anchor set and the superpixel spatial-spectral enhancement anchor set, as well as between the superpixel spectral set and the superpixel spatial-spectral enhancement set. Based on these two structural relationships, a superpixel enhancement anchor map containing both spectral and spatial information is constructed.

[0017] The large-scale hyperspectral image clustering method based on superpixel-enhanced anchor maps described herein improves computational efficiency and robustness by performing non-negative matrix decomposition on the superpixel-enhanced anchor maps under the measure of correlation entropy. Simultaneously, the Laplacian matrix obtained from the superpixel-enhanced anchor maps is used for graph regularization constraints to improve clustering accuracy. A semi-quadratic technique is used to transform the decomposition process into a convex problem for solution. Finally, the categories of superpixels are directly obtained from the non-negative matrix obtained from the decomposition.

[0018] A large-scale hyperspectral image clustering system based on superpixel-enhanced anchor point maps includes:

[0019] The superpixel segmentation module is used to divide the hyperspectral image into superpixel regions and generate a superpixel set;

[0020] The superpixel spatial spectrum enhancement set construction module is used to calculate the average spectral information of the superpixel's neighbors and fuse it with the original features of the superpixel to obtain the superpixel spatial spectrum enhancement set.

[0021] The superpixel-enhanced anchor point map construction module is used to jointly learn the relationship between the original data and anchor points in spectral features and spatial spectral features to obtain the superpixel-enhanced anchor point map.

[0022] The nonnegative matrix factorization module based on correlation entropy is used to perform nonnegative matrix factorization on the superpixel enhancement anchor point map and directly obtain the superpixel category from the nonnegative matrix obtained by factorization.

[0023] The hyperspectral point clustering result generation module is used to map hyperspectral points through superpixels to obtain the clustering result of all pixels in the final hyperspectral image.

[0024] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the large-scale hyperspectral image clustering method based on superpixel-enhanced anchor maps.

[0025] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the large-scale hyperspectral image clustering method based on superpixel-enhanced anchor point maps.

[0026] Compared with the prior art, the present invention has the following advantages:

[0027] 1) This invention constructs a superpixel enhancement anchor map and performs non-negative matrix decomposition on the constructed superpixel enhancement level anchor map, which has linear complexity and significantly improves the efficiency of hyperspectral image clustering.

[0028] 2) This invention significantly improves the accuracy of hyperspectral image clustering by constructing a superpixel spatial spectrum enhancement set and jointly learning the relationship between the original data and anchor points in the spectral features and spatial spectrum features during the construction of the superpixel enhancement anchor point map.

[0029] 3) This invention reduces the impact of a large amount of nonlinear and non-Gaussian noise in the real world by introducing a relevant entropy metric into the non-negative matrix decomposition process of the anchor point graph, and significantly improves the robustness of hyperspectral image clustering in the real world.

[0030] Extensive experiments have shown that the method of this invention is significantly superior to existing technologies in terms of clustering efficiency, accuracy, and robustness, providing an effective solution for large-scale hyperspectral image clustering. It can be applied to fields such as precision agriculture, environmental monitoring, military reconnaissance, and mineral exploration. Attached Figure Description

[0031] Figure 1 This is a flowchart of the large-scale hyperspectral image clustering method based on superpixel enhanced anchor point map described in this invention;

[0032] Figure 2 It is a feature mosaic of the original spectral information and the average spectral information of the neighborhood;

[0033] Figure 3a and Figure 3b These are the convergence curves of the method of this invention on the Indian Pines and Pavia Centre datasets, respectively.

[0034] Figure 4a This is a diagram showing the final segmentation result of the S3AGC method on the Indian Pines dataset. Figure 4b This is a diagram showing the final segmentation result of the BGPC method on the Indian Pines dataset. Figure 4c This is a diagram showing the final segmentation result of the SDST method on the Indian Pines dataset. Figure 4d This is a diagram showing the final segmentation result of the method of this invention on the Indian Pines dataset. Detailed Implementation

[0035] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0036] In the first embodiment of the present invention, as Figure 1 The diagram illustrates the large-scale hyperspectral image clustering method based on superpixel-enhanced anchor point maps described in this invention, comprising the following steps:

[0037] Step S01: For the input hyperspectral image X, we have:

[0038]

[0039] Where X is the set of all pixels in the hyperspectral image, x i Let represent the i-th hyperspectral point, d be the number of spectral bands, n be the number of hyperspectral pixels, and i be the index. Apply an entropy rate superpixel segmentation algorithm based on graph theory to the input X to generate a superpixel set. Where m represents the number of superpixels, s i This represents the spectral point of the i-th superpixel.

[0040] Step S02: Construct the superpixel spatial spectrum enhancement set for superpixel s i The average spectral information of its direct neighbors, i.e., the characteristic average, is denoted as f. avg (s i ),satisfy:

[0041]

[0042] Where |H1(s) i )| represents s i The number of direct neighbors of the superpixel. The average feature value of the direct neighbors of the superpixel is compared with s. i The original features are spliced ​​and fused to obtain a new Notation:

[0043] These are superpixel spectral feature points that incorporate local spatial information, thus yielding a superpixel spatial-spectral enhancement set that incorporates spatial information. Each point in the superpixel spatial spectrum enhancement set contains two parts: the original spectral information and the neighborhood average spectral information.

[0044] Step S03: Constructing the superpixel enhancement anchor map, which involves obtaining the superpixel spectral anchor set by applying the k-means method to the superpixel spectral set. Where l represents the number of anchor points. This is achieved by mapping the superpixel spectral anchor point set to... The superpixel spatial spectrum enhancement anchor set for computational fusion of spatial spectrum information Where u i Represents the spectral anchor point of the i-th superpixel. This represents the i-th superpixel spatial spectrum enhancement anchor point. The superpixel enhancement anchor point graph is constructed using an efficient neighbor allocation strategy that fully utilizes spatial information while reducing manual setting of hyperparameters, as shown in the following formula:

[0045]

[0046] in, For the constructed superpixel enhanced anchor point map, b ij Let represent the element in the i-th row and j-th column of matrix B, and Represents superpixel s i With superpixel anchor point u j The square of the Euclidean distance between them. This represents superpixel spatial-spectral enhancement that incorporates spatial information. The objective function described above contains three terms: Learn B using spectral features; B is learned by combining local spatial and spectral features, where α is a weighting coefficient that controls the trade-off between the first two terms; This is the regularization term, where μ is the regularization parameter. By optimizing the above objective function, a sparse superpixel enhancement anchor point map B is obtained.

[0047] Step S04: Nonnegative Matrix Factorization Based on Correlation Entropy: This step designs a nonnegative matrix factorization method for superpixel-level anchor graphs based on the correlation entropy metric. The correlation entropy metric ensures the robustness of clustering, and directly obtaining the sample categories from the nonnegative matrix obtained by the factorization significantly improves the efficiency of clustering. Furthermore, this step applies a lightweight graph regularization constraint during the nonnegative matrix factorization process, ensuring clustering accuracy without affecting clustering efficiency. The objective function of the nonnegative matrix factorization based on correlation entropy is as follows:

[0048]

[0049] stQ≥0,H≥0.

[0050] in and It is a factor matrix of a non-negative matrix, c represents the number of sample classes, G{} represents the Gaussian function, η represents the balance parameter of the graph regularization constraint, and H T L represents the transpose of matrix H. AG Let be the Laplace matrix, which is defined as follows: Among them W AG =B T B is a similarity matrix constructed from anchor points, D AG It is W AG The corresponding degree matrix.

[0051] Since the above problem is non-convex, it is difficult to solve directly. This invention transforms it into a convex optimization problem using a semi-quadratic technique; therefore, the objective function can be transformed into the following form:

[0052]

[0053] stQ≥0,H≥0.

[0054] Where V is a diagonal auxiliary matrix derived using semi-quadratic techniques, and its diagonal elements are... Where σ is the Gaussian kernel bandwidth. Finally, by optimizing the objective function in the above equation, the category of the superpixel spectral point can be directly obtained from Q.

[0055] Step S05: Generation of hyperspectral point clustering results: Hyperspectral points are mapped by superpixel spectral points to obtain the clustering results of all pixels in the final hyperspectral image.

[0056] By designing a spatially aware superpixel-enhanced anchor graph construction method that integrates superpixel neighborhood information, local spatial information is incorporated into the anchor graph construction process. This method not only considers spectral characteristics but also utilizes spatial neighborhood correlations to effectively acquire high-quality spatial information, thereby improving the overall clustering accuracy. Experiments on multiple mainstream datasets demonstrate that, compared to various existing clustering methods, this invention achieves best or near-best performance on several evaluation metrics, such as overall accuracy (OA), Kappa coefficient, and normalized mutual information (NMI).

[0057] This paper introduces a nonnegative matrix factorization (NMF) algorithm based on correlation entropy, combining it with semi-quadratic techniques to transform the original problem into a convex optimization problem. This optimized algorithm requires fewer iterations to complete the matrix factorization, and the clustering results can be directly obtained from the factor matrices in the NMF, avoiding computational redundancy caused by singular value decomposition and post-processing steps. Experimental results show that this invention requires less time than similar algorithms when processing large-scale hyperspectral images, and in some cases, it is only a fraction of the time required by other algorithms, significantly improving clustering efficiency.

[0058] Applying relevant entropy to the nonnegative matrix factorization process significantly enhances the ability of the method in this invention to resist noise and data anomalies, improves the robustness of clustering results, and exhibits higher noise resistance on datasets of different sizes.

[0059] like Figure 2 As shown, this invention compares the feature average of the direct neighbors of a superpixel with s i The original features are spliced ​​and fused to obtain a superpixel spatial spectrum enhancement set that incorporates local spatial information. This effectively improves the perception of local spatial information by the pixel enhancement anchor map.

[0060] like Figure 3a and Figure 3b As shown, the Indian Pines and Pavia Centre datasets both achieved fast convergence after iterative optimization using the method of this invention, proving that the method of this invention has good fast convergence characteristics for different large-scale hyperspectral image datasets.

[0061] The embodiments of the present invention can effectively improve the segmentation results of hyperspectral images while maintaining good robustness. Compared with other clustering methods for hyperspectral images in related technologies, the image segmentation effectiveness of the embodiments of the present invention is higher, such as... Figures 4a-4d As shown, Figure 4a , Figure 4b , Figure 4c , Figure 4d The final segmentation results of the three comparison methods and the method of this invention on the Indian Pines dataset are shown sequentially. The segmentation map obtained by the method of this invention is smoother than the other methods. It exhibits higher precision and naturalness in image smoothness and region transition processing, with softer boundaries, stronger internal consistency, and a more harmonious and unified overall appearance. Specifically, the algorithm proposed in this invention outperforms S3AGC, BGPC, and SDST, indicating that the superpixel segmentation and neighbor feature fusion method has a more accurate representation ability for homogeneous HRSI regions.

[0062] Therefore, this invention is a high-precision, robust, and efficient large-scale hyperspectral image clustering method based on superpixel-enhanced anchor maps. It improves the clustering efficiency of large-scale hyperspectral images by integrating superpixel-level spatial-spectral information into superpixel-enhanced anchor maps. Under the correlation entropy criterion, superpixel-enhanced anchor map decomposition is performed, which improves the robustness to hyperspectral clustering tasks containing complex noise.

[0063] The following are embodiments of the apparatus of the present invention, which can be used to execute embodiments of the method of the present invention. For details not omitted in the apparatus embodiments, please refer to the embodiments of the method of the present invention.

[0064] In the second embodiment of the present invention, a large-scale hyperspectral image clustering system is provided. This large-scale hyperspectral image clustering system can be used to implement the above-mentioned large-scale hyperspectral image clustering method. Specifically, the large-scale hyperspectral image clustering system includes a superpixel segmentation module, a superpixel spatial spectrum enhancement set construction module, a superpixel enhancement anchor point map construction module, a non-negative matrix factorization module based on correlation entropy, and a hyperspectral point clustering result generation module.

[0065] The superpixel segmentation module is used to divide the hyperspectral image into superpixel regions and generate a superpixel set.

[0066] The superpixel spatial spectrum enhancement set construction module is used to calculate the average spectral information of the superpixel's neighbors and fuse it with the original features of the superpixel to obtain the superpixel spatial spectrum enhancement set.

[0067] The superpixel-enhanced anchor map construction module is used to jointly learn the relationship between the original data and anchor points in spectral features and spatial spectral features to construct an anchor map containing spatial information, thus obtaining the superpixel-enhanced anchor map.

[0068] The non-negative matrix factorization module based on correlation entropy performs non-negative matrix factorization on the constructed superpixel enhancement anchor point map based on the correlation entropy criterion, and applies graph regularization constraints to the factorization process. Finally, the category of superpixel points is directly obtained from the non-negative matrix obtained by factorization.

[0069] The hyperspectral point clustering result generation module is used to map hyperspectral points through superpixels to obtain the clustering result of all pixels in the final hyperspectral image.

[0070] In a third embodiment of the present invention, a terminal device is provided, comprising a processor and a memory. The memory stores a computer program, which includes program instructions. The processor executes the program instructions stored in the computer storage medium. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions to achieve a corresponding method flow or corresponding function. The processor described in this embodiment can be used for large-scale hyperspectral image clustering methods, including: S01, superpixel segmentation, dividing the hyperspectral image into a series of non-overlapping superpixel regions containing highly homogeneous pixels to obtain a superpixel spectral set. S02, Superpixel Spatial-Spectral Enhancement Set Construction: For the superpixel spectral set, calculate the average spectral information of the neighbors of each superpixel spectral point, and fuse it with the original features of the superpixel points to obtain the superpixel spatial-spectral enhancement set. S03, Superpixel Enhancement Anchor Map Construction: Select anchor points for both the superpixel spectral set and the superpixel spatial-spectral enhancement set to obtain the superpixel spectral anchor set and the superpixel spatial-spectral enhancement anchor set, respectively. Then, construct the superpixel enhancement anchor map by jointly learning the relationship between the original data in the spectral features and spatial-spectral features and the anchor points. S04, Non-negative Matrix Factorization Based on Correlation Entropy: Perform non-negative matrix factorization on the constructed superpixel enhancement anchor map based on the correlation entropy criterion, and finally directly obtain the superpixel point categories from the non-negative matrix obtained by factorization. S05, Hyperspectral Point Clustering Result Generation: Map hyperspectral points to superpixel points to obtain the final clustering results of all pixels in the hyperspectral image.

[0071] In a fourth embodiment of the present invention, a computer-readable storage medium is also provided. This computer-readable storage medium is a memory device in a terminal device used to store programs and data. It is understood that the computer-readable storage medium here may include both the built-in storage medium in the terminal device and extended storage media supported by the terminal device. The computer-readable storage medium provides storage space that stores the terminal's operating system. Furthermore, this storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions may be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here may be high-speed RAM or non-volatile memory, such as at least one disk storage device.

[0072] One or more instructions stored in a computer-readable storage medium can be loaded and executed by a processor to implement the corresponding steps of the incomplete multi-view image segmentation method in the above embodiments.

[0073] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0074] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0075] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0076] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0077] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0078] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0079] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

[0080] It should be understood that the above description is for illustrative purposes and not for limitation. Many embodiments and applications beyond the provided examples will be apparent to those skilled in the art upon reading the above description. Therefore, the scope of this patent should not be determined by reference to the above description, but rather by reference to the foregoing claims and the full scope of their equivalents. For purposes of completeness, all articles and references, including patent applications and publications, are incorporated herein by reference. The omission of any aspect of the subject matter disclosed herein in the foregoing claims is not intended as a waiver of that subject matter, nor should it be construed as an indication that the applicant has not considered that subject matter as part of the disclosed inventive subject matter.

Claims

1. A large-scale hyperspectral image clustering method based on superpixel-enhanced anchor point maps, characterized in that, Includes the following steps: Superpixel segmentation: Perform superpixel segmentation on hyperspectral images to divide the hyperspectral images into a series of non-overlapping superpixel regions containing highly homogeneous pixels, generating a superpixel spectral set; Construction of Superpixel Spatial Spectrum Enhancement Set: For the superpixel spectral set, calculate the average spectral information of the neighbors of the superpixel point, and obtain the superpixel spatial spectrum enhancement set by fusing it with the original features of the superpixel point. Each superpixel feature in the superpixel spatial spectrum enhancement set contains two parts of information: spectral information and neighborhood spatial information. Superpixel enhancement anchor map construction: Select anchor points for the superpixel spectral set and the superpixel spatial spectral enhancement set respectively to obtain the superpixel spectral anchor point set and the superpixel spatial spectral enhancement anchor point set. Then, construct the superpixel enhancement anchor map by jointly learning the relationship between the original data and anchor points in the spectral features and spatial spectral features. Non-negative matrix decomposition based on correlation entropy: The constructed superpixel enhancement anchor point map is decomposed into a non-negative matrix based on the correlation entropy criterion. At the same time, graph regularization constraints are applied to the decomposition process, and finally the category of superpixel spectral points is obtained directly from the non-negative matrix obtained by decomposition. Hyperspectral point clustering result generation: Hyperspectral points are mapped by superpixel spectral points to obtain the clustering result of all pixels in the final hyperspectral image.

2. The large-scale hyperspectral image clustering method based on superpixel-enhanced anchor point maps according to claim 1, characterized in that, The input large-scale hyperspectral image data is divided into superpixel regions by an entropy rate superpixel segmentation algorithm based on graph theory, generating a superpixel spectral set.

3. The large-scale hyperspectral image clustering method based on superpixel-enhanced anchor point maps according to claim 1, characterized in that, For each superpixel in the superpixel spectral set, calculate the average spectral information of its neighbors, and concatenate the average spectral information of the neighbors with the superpixel information to obtain the superpixel spatial spectral enhancement set.

4. The large-scale hyperspectral image clustering method based on superpixel-enhanced anchor point maps according to claim 1, characterized in that, The k-means method is used to obtain the superpixel spectral anchor point set; based on the obtained superpixel spectral anchor point set, the mapping calculation is performed on the superpixel spatial spectrum enhancement set to obtain the superpixel spatial spectrum enhancement anchor point set corresponding to the superpixel spatial spectrum enhancement set. Subsequently, Euclidean distance was used to measure the structural relationships between the superpixel spectral anchor set and the superpixel spatial spectral enhancement anchor set, as well as between the superpixel spectral set and the superpixel spatial spectral enhancement set. Based on the two structural relationships, a superpixel enhancement anchor map containing both spectral and spatial information was constructed.

5. The large-scale hyperspectral image clustering method based on superpixel-enhanced anchor point maps according to claim 1, characterized in that, Under the metric of relevant entropy, nonnegative matrix decomposition is performed on the superpixel-enhanced anchor graph to improve computational efficiency and robustness; meanwhile, the Laplacian matrix obtained from the superpixel-enhanced anchor graph is used for graph regularization constraints to improve clustering accuracy. The decomposition process is transformed into a convex problem using a semi-quadratic technique for solution; finally, the superpixel category is obtained directly from the non-negative matrix obtained from the decomposition.

6. A large-scale hyperspectral image clustering system based on superpixel-enhanced anchor point maps, characterized in that, include: The superpixel segmentation module is used to divide the hyperspectral image into superpixel regions and generate a superpixel set; The superpixel spatial spectrum enhancement set construction module is used to calculate the average spectral information of the superpixel's neighbors and fuse it with the original features of the superpixel to obtain the superpixel spatial spectrum enhancement set. The superpixel-enhanced anchor point map construction module is used to jointly learn the relationship between the original data and anchor points in spectral features and spatial spectral features to obtain the superpixel-enhanced anchor point map. The non-negative matrix factorization module based on correlation entropy is used to perform non-negative matrix factorization on the superpixel enhancement anchor point map and directly obtain the superpixel category from the non-negative matrix obtained by factorization. The hyperspectral point clustering result generation module is used to map hyperspectral points through superpixels to obtain the clustering result of all pixels in the final hyperspectral image.

7. A computer device, characterized in that, The method includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the large-scale hyperspectral image clustering method based on superpixel-enhanced anchor maps as described in any one of claims 1 to 6.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the large-scale hyperspectral image clustering method based on superpixel-enhanced anchor maps as described in any one of claims 1 to 6.

Images