A hyperspectral camouflage effect evaluation method, system, device and medium based on the combination of band adaptation and spatial spectrum characteristics

Abstract

The present invention discloses a hyperspectral camouflage effectiveness assessment method based on the combination of band adaptation and spatial spectrum features, comprising the following steps: Step 1: Inputting a hyperspectral image; Step 2: Adaptively reducing the band dimensions using a K-means clustering algorithm; Step 3: Acquiring a camouflaged target area; Step 4: Extracting camouflage assessment indicators and calculating spatial spectrum features; Step 5: Processing the camouflage assessment indicators obtained in Step 4 using a linear scaling method; Step 6: Determining the weights of the camouflage assessment indicators using an entropy weighting method; and Step 7: Outputting the camouflage effectiveness assessment results. The disclosed method replaces the original hyperspectral image with several key frequency bands for processing without changing the attribute characteristics represented by the original hyperspectral data. This significantly reduces hyperspectral band redundancy, improves data processing efficiency, and ensures image algorithm accuracy.

Description

Technical Field

[0001] The present invention belongs to the research field of camouflage reconnaissance and detection, and particularly relates to a hyperspectral camouflage effect evaluation method, system, equipment and medium based on the combination of band adaptation and spatial spectrum characteristics in this field. Background Art

[0002] Experts and scholars have conducted extensive research on hyperspectral camouflage effectiveness assessment methods and achieved numerous results. However, most of these methods utilize only single spectral or spatial-spectral features, and few studies address band adaptation. With the advancement of hyperspectral imaging technology, the number of hyperspectral bands has continued to increase, along with the amount of image data and computational time. This has prompted research on hyperspectral image band dimensionality reduction techniques. Camouflage effectiveness assessment methods that combine band adaptation with spatial-spectral features have gradually become a research focus.

[0003] A comprehensive hyperspectral camouflage effectiveness evaluation method based on intuitionistic fuzzy decision-making has been disclosed. This method, targeting defense engineering targets, establishes an evaluation index system based on spectral pan-similarity, brightness contrast, visual similarity, and structural similarity between the camouflaged target and background features. It then uses the gray correlation TOPSIS method based on the Hamming distance of intuitionistic fuzzy sets to determine the evaluation level and obtain hyperspectral camouflage effectiveness evaluation results. However, this method is limited to camouflaged targets against a green vegetation background, and the spectral characteristic bands used for the evaluation indicators are already selected, making it unsuitable for other backgrounds.

[0004] Prior art discloses a camouflage effectiveness assessment method based on hyperspectral image detection and perception. This method first uses local anomaly detection, spatial density clustering, and a neighborhood fusion algorithm to perform threshold segmentation on the image, thereby determining the potential target area. A finite-time search model is then established based on three evaluation metrics: maximum detection value, false alarm rate, and target area. Camouflage evaluation results are obtained from multiple perspectives, including camouflage evaluation metrics and time-based evaluation scores. However, a limitation of this method is that its feasibility depends on whether the camouflaged target is within the potential target area. Since the potential target area is determined solely based on the Mahalanobis distance between the target and background spectra, which does not include spatial spectral features, subsequent discussion becomes meaningless if the camouflaged target is not within the potential target area.

[0005] The prior art discloses a study on the evaluation of hyperspectral camouflage effects based on spectral indices. This study uses hyperspectral images of targets and backgrounds against typical woodland and snowy backgrounds as research samples. Based on the principle of "same spectrum, same color" camouflage, the study designs and calculates the spectral indices corresponding to the target and background for different backgrounds. The spectral indices are used for image segmentation, and the spectral consistency coefficient index proposed in this study is then used to quantify the differences between the target and background. This quantitative analysis of the camouflage effect yields a quantitative evaluation result. The shortcomings of this method are that the spectral indices designed in the study are derived solely from the analysis of known typical ground background spectral curves, are based on specific characteristic bands, and lack universal applicability. Furthermore, the hyperspectral data used is captured by a ground-based system, which does not conform to the actual situation where threats primarily come from aerial reconnaissance. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to provide a hyperspectral camouflage effect evaluation method, system, equipment and medium based on the combination of band adaptation and spatial spectrum characteristics, which can be used in various backgrounds.

[0007] The present invention adopts the following technical solutions:

[0008] A hyperspectral camouflage effect evaluation method based on the combination of band adaptation and spatial spectrum characteristics is improved in that it includes the following steps:

[0009] Step 1: Input hyperspectral image:

[0010] Input the hyperspectral image to form a hyperspectral data cube and mark the bands;

[0011] Step 2: Adopt K-means clustering algorithm to adaptively reduce the dimension of the band;

[0012] Step 21: input hyperspectral images of each band, where the number of hyperspectral bands is m′ and the number of bands to be retained is n;

[0013] Step 22: round down m′ / n to get an integer A, use the Ath band as the first cluster center band, use the 2Ath band as the second cluster center band, and so on to obtain n initial cluster center bands;

[0014] Step 23: For each band, calculate the distance to the n initial cluster center bands, find the cluster center closest to it, and assign the band to the cluster where the cluster center is located;

[0015] Step 24, update the cluster center and its class status;

[0016] Step 25: After all bands are classified according to distance, the band average of each class is calculated as the new cluster center band;

[0017] Step 26: Repeat steps 23-25 ​​until the cluster center band no longer changes;

[0018] Step 27, get the best band set {x i |x1,x2,…,x n}: For each class, if the cluster center band is an integer, the cluster center band is taken as the best band of the class; if the cluster center band is not an integer, the cluster center band is rounded and taken as the best band of the class;

[0019] Step 3: Get the camouflaged target area:

[0020] The camouflaged target area is manually selected in the hyperspectral image and marked as O, and the area nine times the area around the camouflaged target area is used as the background area and marked as B;

[0021] Step 4: Extract camouflage evaluation indicators and calculate spatial spectrum features, including target and background brightness comparison, information entropy comparison, and structural similarity comparison; calculate spectral features, including spectral angle and spectral distance:

[0022] Target and background brightness contrast:

[0023]

[0024] In the above formula, L ib is the average brightness value of the background area in the i-th band, L io is the average brightness value of the camouflaged target area in the i-th band, and n is the number of bands;

[0025] Comparison of target and background information entropy:

[0026]

[0027]

[0028] In the above formula, P k′ Indicates the proportion of pixels with grayscale value k′ in the image, h ib is the grayscale entropy of the background area of ​​the i-th band, h io is the grayscale entropy of the camouflaged target area in the i-th band, and n is the number of bands;

[0029] Comparison of target and background structure similarity:

[0030]

[0031] In the above formula, C1, C2, and C3 are positive constants, α, β, and γ are weight coefficients of brightness, contrast, and structure information respectively, and μ io 、μ ibare the grayscale mean of the camouflaged target area and background area in the i-th band, σ io , σ ib are the standard deviations of the camouflaged target area and background area in the i-th band, σ iob is σ io and σ ib The covariance of , n is the number of bands;

[0032] Spectral angle:

[0033] The average spectral vector of the background area is used as the reference standard spectral vector, and the calculation method is as follows:

[0034]

[0035] In the above formula, n is the number of bands, b is the total number of pixels in the background area, and q i (q1,q2,…,q n ) is the spectrum vector of the i-th pixel in the background area;

[0036] Calculate the spectral angle between the n-dimensional average spectral vector of each pixel in the camouflaged target area and the reference standard spectral vector:

[0037]

[0038]

[0039] In the above formula, P i (P1,P2,…,P n ) is the spectrum vector of the i-th pixel in the camouflaged target area, a is the total number of pixels in the camouflaged target area, is the n-dimensional average spectrum vector of the camouflaged target area;

[0040] Spectral distance:

[0041] Calculate the spectral distance between the n-dimensional average spectral vector of each pixel in the camouflaged target area and the reference standard spectral vector:

[0042]

[0043] Step 5: Use the linear scaling method to process the camouflage evaluation index obtained in step 4:

[0044] Step 51: Establish the camouflage evaluation index information matrix X:

[0045]

[0046] In the above formula, n is the number of bands, m=5, is the number of camouflage evaluation indicators, and x ij is the j-th camouflage evaluation index value of the i-th band;

[0047] Step 52: normalize the jth camouflage evaluation index: the target and background structure similarity comparison is a positive index, the larger the better, using the maximum linear scaling transformation method x′ ij =x ij / max j Perform standardization, where i = 1, 2, ..., n; target and background brightness contrast, information entropy contrast, spectral angle and spectral distance are negative indicators, the smaller the better, use the minimization linear proportional transformation method x′ ij =min j / x ij Perform normalization processing, where i = 1, 2, ..., n;

[0048] Step 53: Obtain the camouflage evaluation index information normalization matrix X′:

[0049]

[0050] In the above formula, x′ ij ∈[0,1],x′ ij The bigger the number, the better the camouflage effect.

[0051] Step 6: Use the entropy weight method to establish the weights of the camouflage evaluation indicators:

[0052] Based on the camouflage evaluation index information standardization matrix X′, calculate the proportion y of the index value of the i-th band under the j-th camouflage evaluation index ij :

[0053]

[0054] Calculate the information entropy e of the jth camouflage evaluation index j :

[0055]

[0056] For the jth camouflage evaluation index, the more information the index contains, the better j The smaller the value;

[0057] Calculate the information utility value g of the jth camouflage evaluation index j =1-e j ;

[0058] Calculate the weight of the camouflage evaluation index:

[0059]

[0060] Step 7: Use the improved grey clustering algorithm based on white weight function to establish a comprehensive evaluation system and output the camouflage effect evaluation results:

[0061] Step 71, determining the camouflage evaluation level as I, II, and III, corresponding to excellent, good, and poor, respectively;

[0062] Step 72: construct the whiteweighting function as follows:

[0063] Gray class k=1, camouflage assessment level I, white weighting function is:

[0064]

[0065] Gray class k=2, camouflage assessment level II, white weighting function is:

[0066]

[0067] Gray class k=3, camouflage assessment level III, white weighting function is:

[0068]

[0069] Extending the white weighting function to the left and right respectively corresponds to level X, which is better than level I camouflage, and level IV, which is worse than level III camouflage. The corresponding white weighting functions are:

[0070]

[0071]

[0072] Step 73: Calculate the decision coefficient of the i-th band belonging to the gray class k based on the camouflage evaluation index information standardization matrix X′

[0073]

[0074] Step 74, for the i-th band, Determine the camouflage evaluation level of the i-th band belongs to the k-th * class;

[0075] Step 75 : For all n bands of the hyperspectral image after band adaptation, the lowest camouflage evaluation level is taken as the camouflage effect evaluation result of the hyperspectral image.

[0076] A hyperspectral camouflage effect evaluation system based on the combination of band adaptation and spatial spectrum characteristics has the following improvements:

[0077] Input hyperspectral image module, used to input hyperspectral image to form hyperspectral data cube and mark the band;

[0078] Adaptive dimensionality reduction module, which uses K-means clustering algorithm to adaptively reduce the dimensionality of the band;

[0079] A module for obtaining a camouflaged target area is used to select the camouflaged target area in the hyperspectral image and use an area nine times the area of ​​the camouflaged target area as the background area;

[0080] Camouflage evaluation index extraction module, used to calculate spatial spectrum features and spectral features;

[0081] A processing module processes the obtained camouflage evaluation index using a linear scaling method;

[0082] The module for establishing the weight of disguise evaluation indicators uses the entropy weight method to establish the weight of disguise evaluation indicators;

[0083] The result output module uses the improved grey clustering algorithm based on white weight function to establish a comprehensive evaluation system and output the camouflage effect evaluation results.

[0084] A hyperspectral camouflage effect evaluation device based on the combination of band adaptation and spatial spectrum characteristics is improved in that it includes: a processor; a memory in which executable instructions of the processor are stored; wherein the processor is configured to perform the steps of the above method by executing the executable instructions.

[0085] A computer-readable storage medium is used to store a program, wherein the improvement is that the steps of the above method are implemented when the program is executed.

[0086] The beneficial effects of the present invention are:

[0087] The method disclosed in the present invention uses several key frequency bands to replace the original hyperspectral image for processing without changing the attribute characteristics represented by the original hyperspectral data, which greatly reduces the hyperspectral band redundancy, improves data processing efficiency, and ensures the accuracy of image algorithm.

[0088] The method disclosed in this paper comprehensively considers image brightness, information volume, and structural information in the spatial dimension, building on spectral features. This overcomes the limitations of single hyperspectral evaluation metrics and makes the evaluation results more objective. By operating on specific parts of the background image, the method not only ensures effective camouflage assessment but also improves execution speed and efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0089] Figure 1 It is a schematic flow diagram of the method of the present invention;

[0090] Figure 2 It is a schematic diagram of the area relationship between the camouflaged target area and the 9-fold background area. DETAILED DESCRIPTION

[0091] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0092] Example 1: This example discloses a hyperspectral camouflage effect evaluation method based on the combination of band adaptation and spatial spectrum characteristics. Figure 1 As shown, the following steps are included:

[0093] Step 1: Input hyperspectral image:

[0094] Input the hyperspectral image to form a hyperspectral data cube and mark the bands;

[0095] Step 2: Adopt K-means clustering algorithm to adaptively reduce the dimension of the band;

[0096] Step 21: input hyperspectral images of each band, where the number of hyperspectral bands is m′ and the number of bands to be retained is n;

[0097] Step 22: round down m′ / n to get an integer A, use the Ath band as the first cluster center band, use the 2Ath band as the second cluster center band, and so on to obtain n initial cluster center bands;

[0098] Step 23: For each band, calculate the distance to the n initial cluster center bands, find the cluster center closest to it, and assign the band to the cluster where the cluster center is located;

[0099] Step 24, update the cluster center and its class status;

[0100] Step 25: After all bands are classified according to distance, the band average of each class is calculated as the new cluster center band;

[0101] Step 26, repeat steps 23-25 ​​until the cluster center band no longer changes. At this time, the cluster center band may not be an integer;

[0102] Step 27, get the best band set {x i |x1,x2,…,x n}: For each class, if the cluster center band is an integer, the cluster center band is taken as the best band of the class; if the cluster center band is not an integer, the cluster center band is rounded and taken as the best band of the class;

[0103] Step 3: Get the camouflaged target area:

[0104] The camouflaged target area is manually selected in the hyperspectral image and marked as O, and the maximum circumscribed rectangle of the camouflaged target area is generated. Figure 2 As shown in FIG, the area nine times the size of the camouflaged target area is taken as the background area, which is denoted as B;

[0105] Step 4: Extract camouflage evaluation indicators and calculate spatial spectrum features, including target and background brightness comparison, information entropy comparison, and structural similarity comparison; calculate spectral features, including spectral angle and spectral distance:

[0106] Target and background brightness contrast:

[0107]

[0108] In the above formula, L ib is the average brightness value of the background area in the i-th band, L io is the average brightness value of the camouflaged target area in the i-th band, and n is the number of bands;

[0109] Comparison of target and background information entropy:

[0110]

[0111]

[0112] In the above formula, P k′ Indicates the proportion of pixels with grayscale value k′ in the image, h ib is the grayscale entropy of the background area of ​​the i-th band, h io is the grayscale entropy of the camouflaged target area in the i-th band, and n is the number of bands;

[0113] Comparison of target and background structure similarity:

[0114]

[0115] In the above formula, C1, C2, and C3 are positive constants, α, β, and γ are weight coefficients of brightness, contrast, and structure information respectively, and μ io 、μ ib are the grayscale mean of the camouflaged target area and background area in the i-th band, σ io , σ ib are the standard deviations of the camouflaged target area and background area in the i-th band, σ iob is σ io and σ ib The covariance of , n is the number of bands;

[0116] Spectral angle:

[0117] The average spectral vector of the background area is used as the reference standard spectral vector, and the calculation method is as follows:

[0118]

[0119] In the above formula, n is the number of bands, b is the total number of pixels in the background area, and q i (q1,q2,…,q n ) is the spectrum vector of the i-th pixel in the background area;

[0120] Calculate the spectral angle between the n-dimensional average spectral vector of each pixel in the camouflaged target area and the reference standard spectral vector:

[0121]

[0122]

[0123] In the above formula, P i (P1,P2,…,P n ) is the spectrum vector of the i-th pixel in the camouflaged target area, a is the total number of pixels in the camouflaged target area, is the n-dimensional average spectrum vector of the camouflaged target area;

[0124] Spectral distance:

[0125] Calculate the spectral distance between the n-dimensional average spectral vector of each pixel in the camouflaged target area and the reference standard spectral vector:

[0126]

[0127] Step 5: Use the linear scaling method to process the camouflage evaluation index obtained in step 4:

[0128] The main purpose is to standardize the hyperspectral spatial spectrum features and spectral feature camouflage evaluation indicators extracted in step 4. According to different indicator requirements, corresponding processing methods are adopted for the dimensionless processing problems of the two types of traits, "the bigger the better" and "the smaller the better".

[0129] Step 51: Establish the camouflage evaluation index information matrix X:

[0130]

[0131] In the above formula, n is the number of bands, m=5, is the number of camouflage evaluation indicators, and x ij is the j-th camouflage evaluation index value of the i-th band;

[0132] Step 52: normalize the jth camouflage evaluation index: the target and background structure similarity comparison is a positive index, the larger the better, using the maximum linear scaling transformation method x′ ij =x ij / max jPerform standardization, where i = 1, 2, ..., n; target and background brightness contrast, information entropy contrast, spectral angle and spectral distance are negative indicators, the smaller the better, use the minimization linear proportional transformation method x′ ij =min j / x ij Perform normalization processing, where i = 1, 2, ..., n;

[0133] Step 53: Obtain the camouflage evaluation index information normalization matrix X′:

[0134]

[0135] In the above formula, x′ ij ∈[0,1],x′ ij The bigger the number, the better the camouflage effect.

[0136] Step 6: Use the entropy weight method to establish the weights of the camouflage evaluation indicators:

[0137] Based on the camouflage evaluation index information standardization matrix X′, calculate the proportion y of the index value of the i-th band under the j-th camouflage evaluation index ij :

[0138]

[0139] Calculate the information entropy e of the jth camouflage evaluation index j :

[0140]

[0141] For the jth camouflage evaluation index, the more information the index contains, the better j The smaller the value;

[0142] Calculate the information utility value g of the jth camouflage evaluation index j =1-e j ;

[0143] Calculate the weight of the camouflage evaluation index:

[0144]

[0145] Step 7: Use the improved grey clustering algorithm based on white weight function to establish a comprehensive evaluation system and output the camouflage effect evaluation results:

[0146] Step 71: According to the relevant requirements of camouflage assessment, the camouflage assessment level is determined as I, II, and III, corresponding to excellent, good, and poor respectively;

[0147] Step 72: construct the whiteweighting function as follows:

[0148] Gray class k=1, camouflage assessment level I, white weighting function is:

[0149]

[0150] Gray class k=2, camouflage assessment level II, white weighting function is:

[0151]

[0152] Gray class k=3, camouflage assessment level III, white weighting function is:

[0153]

[0154] Extending the white weighting function to the left and right respectively corresponds to level X, which is better than level I camouflage, and level IV, which is worse than level III camouflage. The corresponding white weighting functions are:

[0155]

[0156]

[0157] Step 73: Calculate the decision coefficient of the i-th band belonging to the gray class k based on the camouflage evaluation index information standardization matrix X′

[0158]

[0159] Step 74, for the i-th band, Determine the camouflage evaluation level of the i-th band belongs to the k-th * class;

[0160] Step 75 , according to relevant requirements for camouflage evaluation, for all n bands of the hyperspectral image after band adaptation, the lowest camouflage evaluation level is taken as the camouflage effect evaluation result of the hyperspectral image.

[0161] This embodiment also discloses a hyperspectral camouflage effect evaluation system based on the combination of band adaptation and spatial spectrum characteristics, including:

[0162] Input hyperspectral image module, used to input hyperspectral image to form hyperspectral data cube and mark the band;

[0163] Adaptive dimensionality reduction module, which uses K-means clustering algorithm to adaptively reduce the dimensionality of the band;

[0164] A module for obtaining a camouflaged target area is used to select the camouflaged target area in the hyperspectral image and use an area nine times the area of ​​the camouflaged target area as the background area;

[0165] Camouflage evaluation index extraction module, used to calculate spatial spectrum features and spectral features;

[0166] A processing module processes the obtained camouflage evaluation index using a linear scaling method;

[0167] The module for establishing the weight of disguise evaluation indicators uses the entropy weight method to establish the weight of disguise evaluation indicators;

[0168] The result output module uses the improved grey clustering algorithm based on white weight function to establish a comprehensive evaluation system and output the camouflage effect evaluation results.

[0169] This embodiment also discloses a hyperspectral camouflage effect evaluation device based on the combination of band adaptation and spatial spectrum characteristics, including: a processor; a memory storing executable instructions of the processor; wherein the processor is configured to perform the steps of the above method by executing the executable instructions.

[0170] This embodiment also discloses a computer-readable storage medium for storing a program, which implements the steps of the above method when executed.

Claims

1. A hyperspectral camouflage effect evaluation method based on the combination of band adaptation and spatial spectrum characteristics, characterized in that: The steps include: Step 1: Input hyperspectral image: Input the hyperspectral image to form a hyperspectral data cube and mark the bands; Step 2: Adopt K-means clustering algorithm to adaptively reduce the dimension of the band; Step 21: input hyperspectral images of each band, where the number of hyperspectral bands is m′ and the number of bands to be retained is n; Step 22: round down m′ / n to get an integer A, use the Ath band as the first cluster center band, use the 2Ath band as the second cluster center band, and so on to obtain n initial cluster center bands; Step 23: For each band, calculate the distance to the n initial cluster center bands, find the cluster center closest to it, and assign the band to the cluster where the cluster center is located; Step 24, update the cluster center and its class status; Step 25: After all bands are classified according to distance, the band average of each class is calculated as the new cluster center band; Step 26: Repeat steps 23-25 ​​until the cluster center band no longer changes; Step 27, get the best band set {x i |x1,x2,…,x n }: For each class, if the cluster center band is an integer, the cluster center band is taken as the best band of the class; if the cluster center band is not an integer, the cluster center band is rounded and taken as the best band of the class; Step 3: Get the camouflaged target area: The camouflaged target area is manually selected in the hyperspectral image and marked as O, and the area nine times the area around the camouflaged target area is used as the background area and marked as B; Step 4: Extract camouflage evaluation indicators and calculate spatial spectrum features, including target and background brightness comparison, information entropy comparison, and structural similarity comparison; calculate spectral features, including spectral angle and spectral distance: Target and background brightness contrast: In the above formula, L ib is the average brightness value of the background area in the i-th band, L io is the average brightness value of the camouflaged target area in the i-th band, and n is the number of bands; Comparison of target and background information entropy: In the above formula, P k′ Indicates the proportion of pixels with grayscale value k′ in the image, h ib is the grayscale entropy of the background area of ​​the i-th band, h io is the grayscale entropy of the camouflaged target area in the i-th band, and n is the number of bands; Comparison of target and background structure similarity: In the above formula, C1, C2, and C3 are positive constants, α, β, and γ are weight coefficients of brightness, contrast, and structure information respectively, and μ io 、μ ib are the grayscale mean of the camouflaged target area and background area in the i-th band, σ io , σ ib are the standard deviations of the camouflaged target area and background area in the i-th band, σ iob is σ io and σ ib The covariance of , n is the number of bands; Spectral angle: The average spectral vector of the background area is used as the reference standard spectral vector, and the calculation method is as follows: In the above formula, n is the number of bands, b is the total number of pixels in the background area, and q i (q1,q2,…,q n ) is the spectrum vector of the i-th pixel in the background area; Calculate the spectral angle between the n-dimensional average spectral vector of each pixel in the camouflaged target area and the reference standard spectral vector: In the above formula, P i (P1,P2,…,P n ) is the spectral vector of the i-th pixel in the camouflaged target area, a is the total number of pixels in the camouflaged target area, is the n-dimensional average spectrum vector of the camouflaged target area; Spectral distance: Calculate the spectral distance between the n-dimensional average spectral vector of each pixel in the camouflaged target area and the reference standard spectral vector: Step 5: Use the linear scaling method to process the camouflage evaluation index obtained in step 4: Step 51: Establish the camouflage evaluation index information matrix X: In the above formula, n is the number of bands, m=5, is the number of camouflage evaluation indicators, and x ij is the j-th camouflage evaluation index value of the i-th band; Step 52: normalize the jth camouflage evaluation index: the target and background structure similarity comparison is a positive index, the larger the better, using the maximum linear scaling transformation method x′ ij =x ij / max j Perform standardization, where i = 1, 2, ..., n; target and background brightness contrast, information entropy contrast, spectral angle and spectral distance are negative indicators, the smaller the better, use the minimization linear proportional transformation method x′ ij =min j / x ij Perform normalization processing, where i = 1, 2, ..., n; Step 53: Obtain the camouflage evaluation index information normalization matrix X′: In the above formula, x′ ij ∈[0,1],x′ ij The bigger the number, the better the camouflage effect. Step 6: Use the entropy weight method to establish the weights of the camouflage evaluation indicators: Based on the camouflage evaluation index information standardization matrix X′, calculate the proportion y of the index value of the i-th band under the j-th camouflage evaluation index ij : Calculate the information entropy e of the jth camouflage evaluation index j : For the jth camouflage evaluation index, the more information the index contains, the better j The smaller the value; Calculate the information utility value g of the jth camouflage evaluation index j =1-e j ; Calculate the weight of the camouflage evaluation index: Step 7: Use the improved grey clustering algorithm based on white weight function to establish a comprehensive evaluation system and output the camouflage effect evaluation results: Step 71, determining the camouflage evaluation level as I, II, and III, corresponding to excellent, good, and poor, respectively; Step 72: construct the whiteweighting function as follows: Gray class k=1, camouflage assessment level I, white weighting function is: Gray class k=2, camouflage assessment level II, white weighting function is: Gray class k=3, camouflage assessment level III, white weighting function is: Extending the white weighting function to the left and right respectively corresponds to level X, which is better than level I camouflage, and level IV, which is worse than level III camouflage. The corresponding white weighting functions are: Step 73: Calculate the decision coefficient of the i-th band belonging to the gray class k based on the camouflage evaluation index information standardization matrix X′ Step 74, for the i-th band, Determine the camouflage evaluation level of the i-th band belongs to the k-th * class; Step 75 : For all n bands of the hyperspectral image after band adaptation, the lowest camouflage evaluation level is taken as the camouflage effect evaluation result of the hyperspectral image.

2. A hyperspectral camouflage effect evaluation device based on the combination of band adaptation and spatial spectrum characteristics, characterized in that: include: A processor; a memory having executable instructions for the processor stored therein; wherein the processor is configured to perform the steps of the method of claim 1 by executing the executable instructions.

3. A computer-readable storage medium for storing a program, characterized in that: When the program is executed, the steps of the method according to claim 1 are implemented.

Images