Space-based hyperspectral point target detection method based on spatial preprocessing

Abstract

The present invention relates to a space-based hyperspectral point target detection method based on spatial preprocessing, comprising: performing spatial preprocessing based on pixel domain projection on a hyperspectral image to obtain a first potential target area; performing spatial preprocessing based on feature domain decomposition on the hyperspectral image to obtain a second potential target area; using the intersection of the first potential target area and the second potential target area as a point target to be detected area; when there is no prior target spectrum, based on the main endmember spectrum in the hyperspectral image and the point target to be detected area, obtaining the location of the point target; when there is a prior target spectrum, based on the prior target spectrum and the point target to be detected area, obtaining the type and position of the point target. The present invention can effectively improve the detection probability and accuracy of point targets in low spatial resolution hyperspectral images in space-based scenarios.

Description

Technical Field

[0001] The present invention relates to the technical field of space-based hyperspectral point target detection, and in particular to a space-based hyperspectral point target detection method based on space preprocessing. Background Art

[0002] Hyperspectral imaging has the characteristics of high spectral resolution and unified image and spectrum. Its spectral resolution is as high as nanometers (nm). It can obtain hundreds of continuous narrow band reflectance information in the area of ​​interest. All bands are arranged together so that a continuous and complete spectral curve can be formed at each pixel point, which is called the unique "spectral fingerprint" of the substance. This feature makes it possible to detect substances that are originally undetectable in full-color and multi-spectral wide bands in hyperspectral remote sensing. Hyperspectral target detection is to identify specific substances based on the spectral information in the hyperspectral image. Due to the characteristics of spectral fingerprints, hyperspectral target detection can detect point targets, even sub-pixel point targets.

[0003] Since space-based hyperspectral images are usually acquired through imaging spectrometers mounted on satellite platforms, their spatial resolution is limited. For example, targets such as civil aircraft only occupy a few pixels in a hyperspectral image with a spatial resolution of 20 meters. Traditional hyperspectral point target detection is usually regarded as a binary classification problem for each pixel. Each pixel in the hyperspectral image is classified as a target or background based on the known target spectrum. This type of method is implemented through pixel-by-pixel classification, which greatly restricts its detection efficiency. At the same time, in the process of target detection, existing traditional algorithms usually only use the prior information of the target spectrum, and do not make full use of the endmember information hidden in the hyperspectral image, which limits the improvement of hyperspectral target detection accuracy.

[0004] In recent years, with the rapid development of deep learning technology, target detection algorithms based on deep networks usually learn the mapping from images to target boxes from massive training data by designing complex network structures. For example, the two-stage RCNN series methods randomly generate candidate areas on the feature map and then classify and regress them one by one. The target detection accuracy is high but the computational complexity is large; while the single-stage YOLO series methods directly detect on the entire image in an end-to-end manner, with better real-time performance but usually slightly lower accuracy than the two-stage methods. These deep learning-based methods require sufficient training data and rich hardware resources to achieve excellent target detection performance.

[0005] Therefore, designing a space-based hyperspectral point target detection method based on spatial preprocessing is an urgent problem to be solved. Summary of the invention

[0006] In order to solve the technical problems existing in the above-mentioned prior art, the object of the present invention is to provide a space-based hyperspectral point target detection method based on spatial preprocessing, which can effectively improve the detection probability and accuracy of point targets in low spatial resolution hyperspectral images in space-based scenarios.

[0007] To achieve the above-mentioned object of the invention, the present invention provides a space-based hyperspectral point target detection method based on space preprocessing, comprising the following steps:

[0008] Step S1, performing spatial preprocessing based on pixel domain projection on the hyperspectral image to obtain a first potential target area;

[0009] Step S2, performing spatial preprocessing on the hyperspectral image based on feature domain decomposition to obtain a second potential target area;

[0010] Step S3, taking the intersection of the first potential target area and the second potential target area as a point target to be detected area;

[0011] Step S4, when there is no prior target spectrum, based on the main endmember spectrum in the hyperspectral image and the area to be detected of the point target, obtaining the location of the point target;

[0012] Step S5: when there is a priori target spectrum, based on the priori target spectrum and the point target area to be detected, obtain the point target type and position.

[0013] According to a technical solution of the present invention, the step S1 specifically includes:

[0014] Step S101, performing a sliding window operation on the hyperspectral image, traversing each pixel in turn;

[0015] Step S102: In each spatial neighborhood, a spatial projection matrix P consisting of neighborhood pixels N excluding the central pixel is used to project the central pixel Perform linear projection to obtain the corrected pixels after spatial preprocessing The calculation method is:

[0016] Where P = N(N T N) -1 N

[0017] Step S103, calculating the spectral angle distance between the corrected pixel obtained in step S102 and the original central pixel, the calculation method is:

[0018]

[0019] Step S104: if the spectral angle distance calculated in step S103 is greater than or equal to the first specific threshold, there are abnormal pixels in the current spatial neighborhood, and the current spatial neighborhood is marked as a potential target area; otherwise, the objects in the current spatial neighborhood are of similar types and there are no abnormal pixels;

[0020] Step S105, repeating steps S101 to S104 until all pixels in the hyperspectral image are traversed;

[0021] Step S106: combine all marked spatial neighborhoods into a first potential target area Θ1.

[0022] According to a technical solution of the present invention, the step S2 specifically includes:

[0023] Step S201, performing a sliding window operation on the hyperspectral image, traversing each pixel in turn;

[0024] Step S202: In each spatial neighborhood, singular value decomposition is performed on the neighborhood pixel set N including the central pixel, and the calculation method is:

[0025] N=S·V·D

[0026] Where D is N in descending order. T The unit row eigenvector of N, S is NN T Unit column eigenvector, V is defined as:

[0027]

[0028] V ii (i=1,2,…,m) represents N T The square roots of the non-zero eigenvalues ​​of N, also in descending order;

[0029] Step S203: Calculate the number of singular vectors q corresponding to the main spectral features and set the threshold T SVD , calculated as:

[0030]

[0031] Step S204: After performing spatial preprocessing of the feature domain on the spatial neighborhood using the main spectral features, a corrected reconstructed data matrix is ​​obtained. It is expressed as: Corrected pixels after spatial preprocessing The modified neighborhood matrix is The first pixel in

[0032] in,

[0033]

[0034] Step S205, calculating the spectral angle distance between the corrected pixel obtained in step S204 and the original central pixel;

[0035] Step S206: If the spectral angle distance in step S205 is greater than or equal to the second specific threshold, there are abnormal pixels in the current spatial neighborhood, and the current spatial neighborhood is marked as a potential target area; otherwise, the objects in the current spatial neighborhood are of similar types and there are no abnormal pixels;

[0036] Step S207, repeating steps S201 to S206 until all pixels in the hyperspectral image are traversed to obtain a modified hyperspectral image after spatial preprocessing;

[0037] Step S208: combine all marked spatial neighborhoods into a second potential target area Θ2.

[0038] According to a technical solution of the present invention, in step S3, the calculation method of step S3 is:

[0039] Θ=Θ1∩Θ2

[0040] Among them, Θ represents the point target area to be detected.

[0041] According to a technical solution of the present invention, the step S4 specifically includes:

[0042] Step S401, estimating the virtual dimension of the original hyperspectral image to obtain the number of end members in the corresponding scene;

[0043] Step S402, performing endmember extraction on the hyperspectral image according to the number of endmembers estimated in step S401 to obtain main endmember spectra in the image;

[0044] Step S403, performing spectral angle matching on the endmember spectrum in the corresponding scene and the pixel set in the point target detection area Θ in sequence;

[0045] Step S404: When the spectral angle distance calculated in step S403 is less than the first distance threshold, the corresponding pixel is determined to be the location of the point target.

[0046] According to a technical solution of the present invention, the step S5 specifically includes:

[0047] Step S501: when the target spectrum is a priori information, the a priori target spectrum is sequentially matched with the pixel set in the point target detection area Θ for spectral angle matching;

[0048] Step S502: When the spectral angle distance calculated in step S501 is less than the second distance threshold, the corresponding pixel is determined to be a point target to be measured, and the corresponding position is the position of the point target.

[0049] According to a technical solution of the present invention, in step S401, a VD algorithm or a HySime algorithm is used to estimate the virtual dimension of the original hyperspectral image.

[0050] According to a technical solution of the present invention, in step S402, the OSP, VCA, or N-FINDR algorithm is used to extract endmembers from the hyperspectral image.

[0051] According to one aspect of the present invention, an electronic device is provided, comprising: one or more processors, one or more memories, and one or more computer programs; wherein the processor is connected to the memory, and the one or more computer programs are stored in the memory, and when the electronic device is running, the processor executes the one or more computer programs stored in the memory, so that the electronic device performs a space-based hyperspectral point target detection method based on spatial preprocessing as described in any one of the above technical solutions.

[0052] According to one aspect of the present invention, a computer-readable storage medium is provided for storing computer instructions. When the computer instructions are executed by a processor, a space-based hyperspectral point target detection method based on spatial preprocessing as described in any one of the above technical solutions is implemented.

[0053] Compared with the prior art, the present invention has the following beneficial effects:

[0054] The present invention proposes a space-based hyperspectral point target detection method based on spatial preprocessing. By utilizing dual spatial preprocessing in the pixel domain and the feature domain, the influence of detector measurement error and scene noise on pixels is reduced. By selecting a subset of the target area to be detected, the range of potential areas of point targets in space-based hyperspectral images is narrowed, and the amount of calculation for subsequent spectral angle matching and target detection is reduced, thereby improving the detection accuracy and efficiency of the present invention.

[0055] When the target spectrum is unknown, the present invention proposes to use the endmember information in the hyperspectral image after spatial preprocessing for target matching, which can effectively improve the detection rate of low spatial resolution point targets in space-based scenarios and reduce the false alarm rate of target detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0057] Figure 1The overall flow chart of a space-based hyperspectral point target detection method based on space preprocessing according to one embodiment of the present invention is schematically shown;

[0058] Figure 2 The overall flow chart of a space-based hyperspectral point target detection method based on space preprocessing according to another embodiment of the present invention is schematically shown;

[0059] Figure 3 A schematic diagram schematically shows a flow chart of a spatial preprocessing method based on pixel domain projection according to an embodiment of the present invention;

[0060] Figure 4 The flowchart of the spatial preprocessing method based on feature domain decomposition according to one embodiment of the present invention is schematically shown. DETAILED DESCRIPTION

[0061] The description of the embodiments of this specification should be combined with the corresponding drawings, which should be considered as part of the complete specification. In the drawings, the shape or thickness of the embodiments may be enlarged and indicated for simplification or convenience. Furthermore, the parts of each structure in the drawings will be described separately. It is worth noting that the elements not shown in the drawings or not described in words are in a form known to ordinary technicians in the relevant technical field.

[0062] The description of the embodiments herein and any reference to directions and orientations are only for the convenience of description and are not to be construed as any limitation on the scope of protection of the present invention. The following description of the preferred embodiments may involve combinations of features, which may exist independently or in combination, and the present invention is not particularly limited to the preferred embodiments. The scope of the present invention is defined by the claims.

[0063] like Figure 1 and Figure 2 As shown, a space-based hyperspectral point target detection method based on space preprocessing of the present invention comprises the following steps:

[0064] Step S1: perform spatial preprocessing based on pixel domain projection on the hyperspectral image to obtain a first potential target area, such as Figure 3 As shown, specifically including:

[0065] Step S101, performing a sliding window operation on the hyperspectral image, traversing each pixel in turn;

[0066] Since all pixels except the central pixel in the window will participate in the reconstruction of the central pixel, the categories of all pixels in each spatial window should be as close or consistent as possible, so the size of the sliding window is selected as 3×3.

[0067] Step S102: In each spatial neighborhood, a spatial projection matrix P consisting of neighborhood pixels N (including the number of pixels n=8) other than the central pixel is used to project the central pixel Perform linear projection to obtain the corrected pixels after spatial preprocessing The calculation method is:

[0068] Where P = N(N T N) -1 N

[0069] Step S103: Calculate the spectral angle distance (SAD) between the corrected pixel obtained in step S102 and the original central pixel, and the calculation method is:

[0070]

[0071] Step S104: if the spectral angle distance calculated in step S103 is greater than or equal to the first specific threshold, there are abnormal pixels in the current spatial neighborhood, and the current spatial neighborhood is marked as a potential target area; otherwise, the objects in the current spatial neighborhood are of similar types and there are no abnormal pixels;

[0072] Step S105, repeating steps S101 to S104 until all pixels in the hyperspectral image are traversed;

[0073] Step S106: combine all marked spatial neighborhoods into a first potential target area Θ1.

[0074] The result of the spectral angular distance (SAD) is in degrees (°). The smaller the value, the more similar the two spectra are. In this embodiment, two spectra with a spectral angular distance greater than 4° are considered to belong to different categories of spectra, that is, the first specific threshold is set to 4°.

[0075] Step S2: perform spatial preprocessing on the hyperspectral image based on feature domain decomposition to obtain a second potential target area, such as Figure 4 As shown, specifically including:

[0076] Step S201, performing a sliding window operation on the hyperspectral image, traversing each pixel in turn;

[0077] Since only some pixel features in the window will be involved in the reconstruction of the central pixel, the effective pixels in each spatial window should be as large as possible, and pixels of other categories should not be introduced. Therefore, the size of the sliding window is selected to be 5×5;

[0078] Step S202: In each spatial neighborhood, singular value decomposition is performed on a neighborhood pixel set N (including n=25 pixels, and the first pixel is the center pixel) including the center pixel. The calculation method is:

[0079] N=S·V·D

[0080] Where D is N in descending order. T The unit row eigenvector of N, S is NN T Unit column eigenvector, V is defined as:

[0081]

[0082] V ii (i=1,2,…,m) represents N T The square roots of the non-zero eigenvalues ​​of N, also in descending order;

[0083] Larger singular vectors are the result of the main spectral features, while smaller singular vectors are caused by noise such as measurement errors. Therefore, if the singular vectors corresponding to the main spectral features can be distinguished and only the singular vectors of the main spectral features are used to reconstruct the pixels, the impact of detector measurement errors and scene noise on the pixels can be reduced.

[0084] Step S203: Calculate the number of singular vectors q corresponding to the main spectral features and set the threshold T SVD , calculated as:

[0085]

[0086] Among them, T SVD It is an artificially set threshold, and in this embodiment, the value is 0.9.

[0087] Step S204: After performing spatial preprocessing of the feature domain on the spatial neighborhood using the main spectral features, a corrected reconstructed data matrix is ​​obtained. It is expressed as: Corrected pixels after spatial preprocessing The modified neighborhood matrix is The first pixel in

[0088] in,

[0089]

[0090] Step S205, calculating the spectral angle distance between the corrected pixel obtained in step S204 and the original central pixel, the calculation method is the same as the formula in step S103;

[0091] Step S206: If the spectral angle distance in step S205 is greater than or equal to the second specific threshold, there are abnormal pixels in the current spatial neighborhood, and the current spatial neighborhood is marked as a potential target area; otherwise, the objects in the current spatial neighborhood are of similar types and there are no abnormal pixels;

[0092] The first specific threshold and the second specific threshold may be set to be equal or unequal.

[0093] Step S207, repeating steps S201 to S206 until all pixels in the hyperspectral image are traversed to obtain a modified hyperspectral image after spatial preprocessing;

[0094] Step S208, combining all marked spatial neighborhoods into a second potential target area θ2;

[0095] Step S3: The intersection of the first potential target area and the second potential target area is used as the point target to be detected area, and the calculation method is:

[0096] Θ=Θ1∩Θ2

[0097] Among them, Θ represents the point target to be detected area;

[0098] Step S4, when there is no prior target spectrum, based on the main endmember spectrum in the hyperspectral image and the point target to be detected area, obtain the location of the point target, specifically including:

[0099] Step S401, estimating the virtual dimension of the original hyperspectral image to obtain the number of end members in the corresponding scene;

[0100] In some embodiments, a VD algorithm or a HySime algorithm is used to estimate the virtual dimension of the original hyperspectral image.

[0101] Step S402, performing endmember extraction on the hyperspectral image according to the number of endmembers estimated in step S401 to obtain main endmember spectra in the image;

[0102] In some embodiments, the OSP, VCA, or N-FINDR algorithm is used to perform endmember extraction on the hyperspectral image.

[0103] Step S403, performing spectral angle matching on the endmember spectrum in the corresponding scene and the pixel set in the point target detection area Θ in sequence;

[0104] Step S404: When the spectral angle distance calculated in step S403 is less than the first distance threshold, the corresponding pixel is determined to be the location of the point target.

[0105] Step S5, when there is a priori target spectrum, based on the priori target spectrum and the point target to be detected area, obtaining the point target type and position, specifically includes:

[0106] Step S501: when the target spectrum is a priori information, the a priori target spectrum is sequentially matched with the pixel set in the point target detection area Θ for spectral angle matching;

[0107] Step S502: When the spectral angle distance calculated in step S501 is less than the second distance threshold, the corresponding pixel is determined to be a point target to be measured, and the corresponding position is the position of the point target.

[0108] The first distance threshold and the second distance threshold may be set to be equal or unequal.

[0109] According to one aspect of the present invention, an electronic device is provided, comprising: one or more processors, one or more memories, and one or more computer programs; wherein the processor is connected to the memory, and the one or more computer programs are stored in the memory, and when the electronic device is running, the processor executes the one or more computer programs stored in the memory, so that the electronic device performs a space-based hyperspectral point target detection method based on spatial preprocessing as any one of the above technical solutions.

[0110] According to one aspect of the present invention, a computer-readable storage medium is provided for storing computer instructions. When the computer instructions are executed by a processor, a space-based hyperspectral point target detection method based on spatial preprocessing as described in any one of the above technical solutions is implemented.

[0111] Computer-readable storage media may include any medium capable of storing or transmitting information. Examples of computer-readable storage media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, optical fiber media, radio frequency (RF) links, etc. Code segments may be downloaded via computer networks such as the Internet, intranets, etc.

[0112] The present invention discloses a space-based hyperspectral point target detection method based on spatial preprocessing. First, the hyperspectral image is spatially preprocessed in the pixel domain and feature domain, which effectively reduces the influence of detector error and measurement noise. At the same time, the area where the point target may exist is reduced by analyzing the difference of spectral characteristics in the local spatial neighborhood under different dimensions. While improving the image quality, the calculation burden for subsequent target spectrum matching is reduced. When the target spectrum is unknown, the present invention proposes to use the end member information in the hyperspectral image after spatial preprocessing for target matching, which can effectively improve the detection rate of low spatial resolution point targets in space-based scenes and reduce the false alarm rate of target detection.

[0113] In addition, it should be noted that the present invention can be provided as a method, an apparatus or a computer program product. Therefore, the embodiments of the present invention can take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the embodiments of the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code.

[0114] The embodiments of the present invention are described with reference to the flowcharts and / or block diagrams of the methods, terminal devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, an embedded processor, or other programmable data processing terminal device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0115] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing terminal device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device, so that a series of operation steps are executed on the computer or other programmable terminal device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable terminal device provide for implementing the process in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0116] It should also be noted that, in this article, the terms "include", "comprises" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or terminal device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal device. In the absence of further restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, article or terminal device including the elements.

[0117] Finally, it should be noted that the above is a preferred embodiment of the present invention. It should be pointed out that although the preferred embodiment of the present invention has been described, for those skilled in the art, once the basic creative concept of the present invention is known, several improvements and modifications can be made without departing from the principles of the present invention, and these improvements and modifications should also be regarded as the protection scope of the present invention. Therefore, the attached claims are intended to be interpreted as including the preferred embodiment and all changes and modifications that fall within the scope of the embodiments of the present invention.

Claims

1. A space-based hyperspectral point target detection method based on spatial preprocessing, characterized in that: The following steps are involved: Step S1, performing spatial preprocessing based on pixel domain projection on the hyperspectral image to obtain a first potential target area; Step S2, performing spatial preprocessing on the hyperspectral image based on feature domain decomposition to obtain a second potential target area; Step S3, taking the intersection of the first potential target area and the second potential target area as a point target to be detected area; Step S4, when there is no prior target spectrum, based on the main endmember spectrum in the hyperspectral image and the area to be detected of the point target, obtaining the location of the point target; Step S5: when there is a priori target spectrum, based on the priori target spectrum and the point target area to be detected, obtain the point target type and position.

2. The space-based hyperspectral point target detection method based on spatial preprocessing according to claim 1 is characterized in that: The step S1 specifically includes: Step S101, performing a sliding window operation on the hyperspectral image, traversing each pixel in turn; Step S102: In each spatial neighborhood, a spatial projection matrix P consisting of neighborhood pixels N excluding the central pixel is used to project the central pixel Perform linear projection to obtain the corrected pixels after spatial preprocessing The calculation method is: Where P = N(N T N) -1 N Step S103, calculating the spectral angle distance between the corrected pixel obtained in step S102 and the original central pixel, the calculation method is: Step S104: if the spectral angle distance calculated in step S103 is greater than or equal to the first specific threshold, there are abnormal pixels in the current spatial neighborhood, and the current spatial neighborhood is marked as a potential target area; otherwise, the objects in the current spatial neighborhood are of similar types and there are no abnormal pixels; Step S105, repeating steps S101 to S104 until all pixels in the hyperspectral image are traversed; Step S106: combine all marked spatial neighborhoods into a first potential target area Θ1.

3. The space-based hyperspectral point target detection method based on spatial preprocessing according to claim 2 is characterized in that: The step S2 specifically includes: Step S201, performing a sliding window operation on the hyperspectral image, traversing each pixel in turn; Step S202: In each spatial neighborhood, singular value decomposition is performed on the neighborhood pixel set N including the central pixel, and the calculation method is: N=S·V·D Where D is N in descending order. T The unit row eigenvector of N, S is NN T Unit column eigenvector, V is defined as: V ii (i=1,2,…,m) represents N T The square roots of the non-zero eigenvalues ​​of N, also in descending order; Step S203: Calculate the number of singular vectors q corresponding to the main spectral features and set the threshold T SVD , calculated as: Step S204: After performing spatial preprocessing of the feature domain on the spatial neighborhood using the main spectral features, a corrected reconstructed data matrix is ​​obtained. It is expressed as: Corrected pixels after spatial preprocessing The modified neighborhood matrix is The first pixel in in, Step S205, calculating the spectral angle distance between the corrected pixel obtained in step S204 and the original central pixel; Step S206: If the spectral angle distance in step S205 is greater than or equal to the second specific threshold, there are abnormal pixels in the current spatial neighborhood, and the current spatial neighborhood is marked as a potential target area; otherwise, the objects in the current spatial neighborhood are of similar types and there are no abnormal pixels; Step S207, repeating steps S201 to S206 until all pixels in the hyperspectral image are traversed to obtain a modified hyperspectral image after spatial preprocessing; Step S208: combine all marked spatial neighborhoods into a second potential target area Θ2.

4. The space-based hyperspectral point target detection method based on spatial preprocessing according to claim 3 is characterized in that: In step S3, the calculation method of step S3 is: Θ=Θ1∩Θ2 Among them, Θ represents the point target area to be detected.

5. The space-based hyperspectral point target detection method based on spatial preprocessing according to claim 1 is characterized in that: The step S4 specifically includes: Step S401, estimating the virtual dimension of the original hyperspectral image to obtain the number of end members in the corresponding scene; Step S402, performing endmember extraction on the hyperspectral image according to the number of endmembers estimated in step S401 to obtain main endmember spectra in the image; Step S403, performing spectral angle matching on the endmember spectrum in the corresponding scene and the pixel set in the point target detection area Θ in sequence; Step S404: When the spectral angle distance calculated in step S403 is less than the first distance threshold, the corresponding pixel is determined to be the location of the point target.

6. The space-based hyperspectral point target detection method based on spatial preprocessing according to claim 1 is characterized in that: The step S5 specifically includes: Step S501: when the target spectrum is a priori information, the a priori target spectrum is sequentially matched with the pixel set in the point target detection area Θ for spectral angle matching; Step S502: When the spectral angle distance calculated in step S501 is less than the second distance threshold, the corresponding pixel is determined to be a point target to be measured, and the corresponding position is the position of the point target.

7. The space-based hyperspectral point target detection method based on spatial preprocessing according to claim 1 is characterized in that: In step S401, the virtual dimension of the original hyperspectral image is estimated using the VD algorithm or the HySime algorithm.

8. The space-based hyperspectral point target detection method based on spatial preprocessing according to claim 1 is characterized in that: In step S402, the OSP, VCA, or N-FINDR algorithm is used to extract endmembers from the hyperspectral image.

9. An electronic device, characterized in that: include: One or more processors, one or more memories, and one or more computer programs; wherein the processor is connected to the memory, and the one or more computer programs are stored in the memory. When the electronic device is running, the processor executes the one or more computer programs stored in the memory, so that the electronic device performs the space-based hyperspectral point target detection method based on space preprocessing as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that: Used to store computer instructions, which, when executed by a processor, implement the space-based hyperspectral point target detection method based on spatial preprocessing as described in any one of claims 1 to 8.

Images