System and method for detecting optical performance of camouflage material

Abstract

The invention discloses an optical performance detection system and method for a camouflage material, and relates to the technical field of camouflage material detection, and the system comprises a structure identification and illumination control module which is used for collecting a surface image of a camouflage material sample, identifying the surface structure of the sample, starting a full-spectrum light source, and simulating step-by-step illumination from weak light to strong light; carrying out RGB imaging based on the standard illumination; and the region division module is used for applying a superpixel segmentation algorithm to the image, automatically dividing sub-regions, performing hyperspectral reflectivity scanning on each region, synchronously acquiring a light intensity image in a polarization direction, calculating a Stokes vector and constructing a polarization enhanced image. According to the method, frequency domain quantitative analysis of microcosmic interference fringes and accurate mask extraction of polarization abnormal areas can be realized, optical exposure risks of camouflage materials under complex illumination and visual angle conditions are effectively quantified through multi-angle reflection sampling and scoring function modeling, and the detection precision, the area pertinence and the automation level are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of camouflage material detection, and particularly to an optical performance detection system and method for camouflage materials. Background Art

[0002] With the continuous development of modern military reconnaissance and intelligent imaging technologies, the optical camouflage ability of camouflage materials has increasingly become a core indicator in material research and development and application evaluation. Existing camouflage materials usually rely on the regulation of reflectivity in the visible light band and the near-infrared band, and achieve color fusion and reflection control of the background through surface pattern design, nanostructure construction, or multi-layer film deposition. In the context of the continuous upgrading of imaging technologies, traditional evaluation methods mainly use spectral reflection tests or image comparison technologies for preliminary judgment, but these methods often cannot accurately identify the risks of camouflage failure caused by factors such as structural color, angle-dependent reflection, or polarization characteristics, and are particularly vulnerable to exposure under complex conditions such as natural light dynamic changes, oblique-angle detection, or polarized laser irradiation. On the other hand, advanced optical detection means such as polarization imaging, hyperspectral imaging, and BRDF (Bidirectional Reflectance Distribution Function) modeling have gradually been introduced into the field of camouflage material research to capture more subtle optical response differences, but there is currently a lack of a complete detection system that can integrate multi-dimensional physical indicators and achieve accurate region division and scoring. Summary of the Invention

[0003] In view of the above existing problems, the present invention is proposed.

[0004] Therefore, the present invention provides an optical performance detection system and method for camouflage materials, which solves the problem that there is currently a lack of a complete detection system that can integrate multi-dimensional physical indicators and achieve accurate region division and scoring.

[0005] To solve the above technical problems, the present invention provides the following technical solutions:

[0006] In a first aspect, the present invention provides an optical performance detection system for camouflage materials, which includes,

[0007] A structure recognition and illumination control module, which is used to collect the surface image of the camouflage material sample, identify the surface structure of the sample, start the full-spectrum light source, simulate the step-by-step illumination from weak light to strong light, and perform RGB imaging based on the standard illuminance;

[0008] A region division module, which is used to apply the superpixel segmentation algorithm to the image, automatically divide sub-regions, perform hyperspectral reflectance scanning on each region, synchronously collect the polarization-direction light intensity image, calculate the Stokes vector, and construct a polarization-enhanced image;

[0009] The regional mask generation module is used to perform adaptive segmentation on the polarized reflection intensity map using the Otsu algorithm, modify the boundary using morphological erosion and dilation operations, and generate the target polarized reflection significant region mask IE(x,y);

[0010] Interference scoring module, which is used to select the hyperspectral reflectance curve within the IE(x,y) range, detect the periodic fringe fluctuations using fast Fourier transform and calculate the interference score Ir(i);

[0011] Multi-angle BRDF measurement module, used to change the sample observation angle, collect multi-angle spectra of the mask area, establish the BRDF response function, and calculate the directional deviation ΔR(i);

[0012] The scoring module is used to comprehensively score the camouflage performance based on the intensity index Ir(i) and the directional deviation ΔR(i). It generates a two-dimensional heat map based on the camouflage performance score, displays the camouflage exposure risk level, and automatically generates a detection report.

[0013] As a preferred solution of the optical property detection system of the camouflage material of the present invention, wherein: the adaptive segmentation of the polarized reflection intensity map using the Otsu algorithm, the boundary correction using morphological erosion and dilation operations, and the generation of the target polarized reflection significant area mask IE(x, y) include:

[0014] Traverse the entire polarization reflection intensity map and normalize it to obtain the normalized image pixel value;

[0015] Perform differential calculations on the entire row of pixels for each image row;

[0016] A standard Sobel operator is applied to the entire image to obtain horizontal and vertical gradient maps and calculate the local edge structure strength of the image. A pixel energy function is constructed, which combines brightness changes with edge suppression information to calculate the pixel-level energy term E(x,y). The energy of all pixels is accumulated row by row to obtain the row energy integration. The maximum energy row y1 is searched as the structural boundary to divide the image into two sub-images, the upper and lower sub-images. If the pixels in row y are less than y1, it is the upper half of the image; otherwise, it is the lower half of the image.

[0017] Perform median filtering on the normalized image to obtain the median map M(x,y). Perform mean filtering on the median map to obtain the local average map G(x,y). Based on M(x,y) and G(x,y), construct the joint grayscale pair (i,j) of each pixel (x,y). Assign the grayscale pairs to the upper and lower half images according to their respective regions. Count the frequencies of all (i,j) in the two regions to form a joint grayscale histogram.

[0018] Define the joint grayscale threshold combination (s, t), and in each histogram, divide (i, j) into 4 regions according to (s, t);

[0019] In calculating the joint gray histogram, calculate the sum of the relative frequencies of all gray value pairs (i, j) belonging to region k in the entire image. Using the gray value of the median filter image M(x, y) as the main variable, calculate the joint gray mean μ of each region respectively. k and the overall mean μ T , According to the two-dimensional Otsu definition, calculate the comprehensive scoring index of the between-class difference. For each (s, t), calculate the trace value of the between-class variance of the upper and lower half images of the joint gray histogram respectively, and multiply the two trace values to obtain the joint scoring function Q(s, t);

[0020] Traverse all the scoring results to find the threshold combination with the highest score;

[0021] In the upper and lower sub-images, traverse each pixel (x, y) respectively to construct the upper half sub-image and lower half sub-image masks, and merge the upper and lower region masks into the full image mask M0(x, y);

[0022] Perform connected region labeling on the mask image M0(x, y), remove the regions with an area smaller than the preset threshold, and perform morphological closing and opening operations on the masked image after removal to obtain the final polarization significant region mask image.

[0023] As a preferred solution of the optical performance detection system for the camouflage material described in the present invention, wherein: collect the surface image of the camouflage material sample, identify the surface structure of the sample, start the full-spectrum light source, simulate the step-by-step illumination from weak light to strong light, perform RGB imaging based on the standard illuminance, and apply the superpixel segmentation algorithm to the image to automatically divide the sub-regions, including:

[0024] Collect the high-resolution gray image of the sample surface, perform two-dimensional fast Fourier transform on the image to obtain the frequency spectrum diagram, calculate the frequency spectrum amplitude diagram and normalize it, and extract the main frequency corresponding to the direction with the largest amplitude in the frequency spectrum;

[0025] If the main frequency is greater than the preset threshold U, further evaluate the energy proportion of the main frequency. If the energy proportion of the main frequency is greater than the preset threshold u, it is considered that this region has the possibility of artificial structural color interference;

[0026] Start the controllable full-spectrum LED array light source, and set the light intensity to five fixed levels, including I1 to I5;

[0027] At each illuminance level, collect the reflection spectrum curve of the material sample surface respectively and calculate the normalized reflectance R(λ j , I i ) based on the collected data. For all wavelength reflectances R(λ, I i ) at the i-th illuminance, calculate their mean value and the reflectance standard deviation σ(I i) Compare the standard deviation results of five groups, select the illumination level number corresponding to the smallest one as the standard illumination, and based on the determined standard illumination, start the industrial camera system to collect sample images and obtain a three-channel RGB image.

[0028] As an optimal solution of the optical property detection system of the camouflage material described in the present invention, wherein: applying a superpixel segmentation algorithm to the image to automatically divide sub-regions, performing hyperspectral reflectance scanning on each region, synchronously collecting polarized light intensity images in the polarization direction, calculating the Stokes vector and constructing a polarization-enhanced image includes:

[0029] Process each component of each pixel separately according to the standard sRGB gamma inverse correction to obtain the RGB image matrix after light intensity linearization;

[0030] For each pixel, convert the linear R, G, and B values to the X, Y, and Z three-channel values, normalize the XYZ components of each pixel point, define the perceptual mapping function f(p) to establish a color-brightness non-linear response, calculate the Lab channel value of each pixel point, and output the three-channel matrix I Lab (x, y);

[0031] Calculate the clustering window spacing S according to the preset number of regions K, and construct its two-dimensional spatial coordinate matrix for each pixel (x, y) in the image;

[0032] For each pixel point (x, y) in the image, construct a five-dimensional joint feature vector V required for SLIC clustering x,y ;

[0033] Combine the V x,y features of all pixel points into a five-dimensional tensor structure T V (x, y);

[0034] Based on S, initialize K clustering center points by uniform grid division on the image. For each clustering center, calculate the composite distance between the current pixel and the center, assign the pixel (x, y) to the clustering center number with the smallest distance, recalculate the five-dimensional feature mean of each region as the new center, calculate the moving distance m between the old center and the new center. If all m are less than the preset threshold β, stop clustering to obtain the final pixel assignment mask matrix, and perform 8-neighborhood connectivity analysis on each region corresponding to the number i;

[0035] Use the contour tracing algorithm to extract the closed boundary point set of the main connected block and save it as the region vector mask, and then calculate the region centroid coordinates

[0036] Extract the centroid coordinates of each sub-region and set them as the target sampling points. Perform pixel-by-pixel scanning on each sub-region to obtain a two-dimensional spectral matrix. Start the DoFP-type polarization camera to output four polarization-direction images, and calculate the Stokes vector components for each pixel point (x, y).

[0037] Combine the obtained Stokes vector components to calculate the polarization-enhanced image IS(x, y) of this region.

[0038] As a preferred solution of the optical property detection system of the camouflage material described in the present invention, wherein: select the hyperspectral reflection curves within the range of IE(x, y), use the fast Fourier transform to detect the periodic stripe fluctuations and calculate the interference score Ir(i), which means traversing each sub-region number i. If the corresponding sample is a structural color material, then screen the pixel points marked by IE(x, y) within this region, and only retain the spectral data in the significantly polarization-enhanced region. Synthesize the hyperspectral reflectance curves at all mask points within this region by averaging to obtain a representative reflectance curve And apply the one-dimensional fast Fourier transform operation to obtain the amplitude spectrum value F i (k) at the frequency index k, and calculate the average energy of the high-frequency band as the interference score Ir(i) of the region.

[0039] As a preferred solution of the optical property detection system of the camouflage material described in the present invention, wherein: change the sample observation angle, perform multi-angle spectral acquisition on the masked region, establish a BRDF response function, and calculate the directional deviation ΔR(i), which means controlling the three-axis servo motor platform to adjust the sample attitude to fix the incident light source. After each angle setting is completed, use the hyperspectral imaging module to collect the hyperspectral reflectance data of each sub-region within the masked region IE(x, y) under the condition of maintaining the standard illuminance. For each sub-region i, at each angle θ k calculate the BRDF function value f r,i (θ k ,λ);

[0040] Perform averaging in the wavelength dimension to obtain the band-averaged BRDF response of the angle-region

[0041] For each sub-region i, take the average of the average BRDF responses at all angles

[0042] Based on and calculate the directional consistency deviation ΔR(i) of sub-region i.

[0043] As a preferred solution of the optical performance detection system of the camouflage material described in the present invention, wherein: the comprehensive evaluation score of the camouflage performance is based on the intensity index Ir(i) and the directional deviation ΔR(i), and a two-dimensional heat map is generated according to the camouflage performance score to display the camouflage exposure risk level and automatically generate a detection report, which means linearly weighting the interference score value Ir(i) and the directional deviation score ΔR(i) of each sub-region i to construct a comprehensive optical camouflage performance score Score(i), mapping the maximum score to 1 and the minimum score to 0 to form a normalized score matrix, using the region boundary as a primitive for each sub-region, filling the color with the normalized score matrix value to form a transparent layer heat map, and generating a detection report according to the detection data.

[0044] In a second aspect, the present invention provides an optical performance detection method for a camouflage material, including

[0045] Collecting the surface image of the camouflage material sample, performing RGB imaging based on the standard illuminance and automatically dividing sub-regions, performing hyperspectral reflectance scanning on each region, synchronously collecting the polarized light intensity image, calculating the Stokes vector and constructing a polarization enhanced image;

[0046] Performing adaptive segmentation and boundary correction on the polarized reflection intensity map to generate a target polarized reflection significant region mask IE(x,y);

[0047] Selecting the hyperspectral reflection curve within the range of IE(x,y), detecting the periodic stripe fluctuation and calculating the interference score Ir(i), changing the sample observation angle, performing multi-angle spectral collection on the masked region, establishing a BRDF response function, and calculating the directional deviation ΔR(i);

[0048] Based on the intensity index Ir(i) and the directional deviation ΔR(i), performing a comprehensive evaluation score of the camouflage performance, generating a two-dimensional heat map according to the camouflage performance score, displaying the camouflage exposure risk level and automatically generating a detection report.

[0049] In a third aspect, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and wherein: when the computer program is executed by the processor, it realizes any step of the optical performance detection method of the camouflage material as described in the first aspect of the present invention.

[0050] In a fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and wherein: when the computer program is executed by the processor, it realizes any step of the optical performance detection method of the camouflage material as described in the first aspect of the present invention.

[0051] The beneficial effects of the present invention are as follows: The present invention can not only automatically identify whether there is structural color interference behavior on the surface of the sample, but also synchronously acquire hyperspectral and polarization images, extract the target area by combining the Stokes vector and the polarization enhancement map, and quantitatively score the high-frequency interference fluctuations of the reflection spectrum based on the fast Fourier transform. At the same time, the platform integrates a multi-angle attitude adjustment mechanism, models the reflection consistency of the material at different viewing angles by constructing a BRDF response function, and finally fuses the interference score and the direction deviation score to generate a two-dimensional heat map and an automatic detection report. It can not only realize the frequency-domain quantitative analysis of microscopic interference fringes and the precise mask extraction of polarization anomaly regions, but also effectively quantify the optical exposure risk of camouflage materials under complex lighting and viewing angle conditions through multi-angle reflection sampling and scoring function modeling, significantly improving the detection accuracy, regional targeting, and automation level. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0053] Figure 1 It is a flowchart of the optical performance detection method for the camouflage material in Embodiment 1.

[0054] Figure 2 It is a structural diagram of the optical performance detection system for the camouflage material in Embodiment 1.

[0055] Figure 3 It is a schematic diagram of the image processing flow in Embodiment 1. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0056] In order to make the above objects, features, and advantages of the present invention more obvious and understandable, the following will provide a detailed description of the specific embodiments of the present invention with reference to the accompanying drawings of the specification.

[0057] In the following description, many specific details are set forth to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar generalizations without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0058] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that can be included in at least one implementation manner of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or selectively exclusive embodiment from other embodiments.

[0059] Example 1, referring to Figures 1 to 3 , which is the first embodiment of the present invention. This embodiment provides an optical property detection system for a camouflage material, including the following steps:

[0060] S1. Structure recognition and lighting control module, which is used to collect the surface image of the camouflage material sample, identify the surface structure of the sample, start the full-spectrum light source, simulate the step-by-step lighting from weak light to strong light, and perform RGB imaging based on the standard illuminance;

[0061] Specifically, collect the surface image of the camouflage material sample, identify the surface structure of the sample, start the full-spectrum light source, simulate the step-by-step lighting from weak light to strong light, perform RGB imaging based on the standard illuminance, and apply the superpixel segmentation algorithm to the image to automatically divide the sub-regions, including:

[0062] Place the camouflage material sample on the electric three-axis sample stage, set the platform attitude to the initial angle, select the scanning area and perform area array imaging on the scanning area, collect the high-resolution grayscale image of the sample surface, perform two-dimensional fast Fourier transform on the image to obtain the frequency spectrum diagram, calculate the frequency spectrum amplitude diagram and normalize it, and then extract the main frequency corresponding to the direction with the largest amplitude in the frequency spectrum;

[0063] If the main frequency is greater than the preset threshold U, further evaluate the energy ratio of the main frequency. If the energy ratio of the main frequency is greater than the preset threshold u, it is considered that this area has the possibility of artificial structural color interference, and the sample is marked as 1, otherwise it is marked as 0;

[0064] The thresholds U and u are set by experience;

[0065] Start the controllable full-spectrum LED area array light source. The spectral range in this embodiment covers 400 - 1000 nm;

[0066] Set the light intensity to five fixed levels, including I1 to I5, where I1 is (weak light / dawning), I2 is (cloudy day / woods), I3 is (daylight / indoor light), I4 is (sunny day with sunshine), I5 is (strong sunshine or reflective surface). In this embodiment, the light intensity of I1 is 500 lux, the light intensity of I2 is 2000 lux, the light intensity of I3 is 6000 lux, the light intensity of I4 is 12000 lux, and the light intensity of I5 is 20000 lux;

[0067] At each level of illuminance, keep the platform attitude unchanged, and respectively collect the reflection spectral curves of the material sample surface, including the sample reflection spectral intensity and the reference standard whiteboard reflection spectral intensity, and calculate the normalized reflectance R(λ j ,I i ):

[0068]

[0069] Wherein, L s (λ j , I i ) is the reflected radiation intensity of the sample at the j-th wavelength point and illuminance I i , L ref R(λ j , I i ) is the radiance of the standard whiteboard, and ρ ref is the known reflectivity of the standard whiteboard;

[0070] For the selected optimal illuminance, it is necessary to evaluate the variation of reflectivity with wavelength under different illuminance conditions. Let the number of sampling points of the reflectivity curve be N (it is recommended to be 50 - 100 equally spaced points within the wavelength range), then:

[0071] For the reflectivity R(λ, I i ) of all wavelengths under the i-th illuminance, calculate its mean value

[0072]

[0073] Calculate the standard deviation σ(I i ) of the reflectivity under this illuminance:

[0074]

[0075] Compare the results of the standard deviations of the five groups, and select the illuminance level number corresponding to the smallest one as the standard illuminance;

[0076] Based on the determined standard illuminance, start the industrial camera system. In this embodiment, a Basler acA1920-155uc camera is selected to collect the sample image and obtain a three-channel RGB image.

[0077] Through structural frequency analysis, potential structural color materials can be quickly identified, the structural adaptability of the detection process can be improved, and redundant calculations can be avoided; the illuminance response discrimination mechanism ensures that the system operates under the best optical performance conditions, enhancing the consistency and accuracy of subsequent reflectivity, polarization, and interference data; placing the image superpixel segmentation before the standard illuminance image acquisition is beneficial to constructing a real spatial partition map under natural light response conditions, which helps with subsequent regional optical property extraction and local scoring determination.

[0078] S2. Region division module, which is used to apply the superpixel segmentation algorithm to the image, automatically divide sub-regions, perform hyperspectral reflectivity scanning on each region, synchronously collect the polarization direction light intensity image, calculate the Stokes vector, and construct a polarization enhanced image;

[0079] Specifically, the RGB image pixel value range is normalized from [0, 255] to [0, 1], and each component of each pixel is processed separately according to the standard sRGB gamma inverse correction to obtain the RGB image matrix after light intensity linearization;

[0080] For each pixel, the linear R, G, and B values are converted into three channel values of X, Y, and Z through standard matrix multiplication (the matrix is defined based on the sRGB and D65 standard light sources) to obtain the XYZ channel image;

[0081] Normalize the XYZ components of each pixel point based on the CIE D65 white point, and define the perceptual mapping function f(p) to establish the color-brightness non-linear response:

[0082]

[0083] In the formula, δ is the boundary value used to define the piecewise non-linear mapping of the function, and p is the input value;

[0084] According to the CIE specification, calculate the Lab channel values for each pixel point:

[0085] L(x, y) = 116 · f(y r ) - 16

[0086] a(x, y) = 500 · (f(x r ) - f(y r ))

[0087] b(x, y) = 200 · (f(y r ) - f(z r ))

[0088] In the formula, L(x, y) is the lightness channel, a(x, y) is the red-green color difference channel, b(x, y) is the yellow-blue color difference channel, x r is the normalized X value, y r is the normalized Y value, z r is the normalized Z value;

[0089] Output the three-channel matrix I Lab (x, y):

[0090] I Lab (x, y) = [L(x, y), a(x, y), b(x, y)]

[0091] Calculate the clustering window spacing S according to the preset number of regions K:

[0092]

[0093] In the formula, H and W are the height and width of the image respectively;

[0094] Construct a two-dimensional spatial coordinate matrix for each pixel (x, y) in the image:

[0095] X grid (x, y) = x

[0096] Y grid (x, y) = y

[0097] In the formula, X grid (x, y) and Y grid (x, y) are the original coordinate values of each pixel in the image in the horizontal direction (column direction) and the vertical direction (row direction), respectively;

[0098] Normalize X grid (x, y) and Y grid (x, y) to a metric standard consistent with the color channel, that is, divide by S:

[0099]

[0100] In the formula, x n (x, y) and y n (x, y) are the normalized results of the original spatial position coordinates of the pixel;

[0101] For each pixel point (x, y) in the image, construct a five-dimensional joint feature vector V required for SLIC clustering x,y :

[0102]

[0103] Combine the V x,y features of all pixel points into a five-dimensional tensor structure T V (x, y):

[0104]

[0105] Based on S, initialize K clustering center points by uniform grid division on the image, and the initial position of each center point is the average value of the five-dimensional vectors of the pixels in the corresponding grid block;

[0106] For each cluster center, perform matching only within its local window of 2S×2S, calculate the composite distance between the current pixel and the center (obtained by calculating the Euclidean distance of the pixel in the CIELab perceptual color space and the two-dimensional spatial Euclidean distance of the pixel on the image plane), assign the pixel (x, y) to the cluster center number with the minimum distance, recalculate the five-dimensional feature mean of each region as the new center, calculate the moving distance m between the old center and the new center. If all m values are less than the preset threshold β (set by experimental tuning), stop clustering to obtain the final pixel assignment mask matrix, which represents the final sub-region number that each pixel point is assigned to. For each region corresponding to the number i, perform 8-neighborhood connectivity analysis, extract its main connected component, and record its area. For sub-connected components or abnormal clusters with an area less than 20 pixels, perform the following merging:

[0107] If the adjacent region numbers are unique, merge them into the neighborhood number;

[0108] If the adjacent region numbers are not unique, merge them into the neighboring region with the largest area;

[0109] Use the contour tracing algorithm (the findContours function in OpenCV) to extract the closed boundary point set of the main connected component and save it as the region vector mask, and then calculate the region centroid coordinates

[0110]

[0111] In the formula, R i is the total number of region pixels, x j and y j are the two-dimensional positions of the j-th pixel in the region respectively;

[0112] Perform hyperspectral reflectance scanning on each region, synchronously collect the intensity images of the polarization directions, calculate the Stokes vector and construct the polarization enhanced image, that is, extract the centroid coordinates of each sub-region and set them as the target sampling points, control the three-axis electric displacement platform to align the hyperspectral imaging window with the coordinate center, synchronously lock the imaging area of the polarization camera to make it consistent with the hyperspectral sampling field of view, and ensure that two types of data are obtained synchronously for the same region;

[0113] Start the linear scanning type hyperspectral camera, perform pixel-by-pixel scanning on each sub-region to obtain a two-dimensional spectral matrix, start the DoFP type polarization camera, perform synchronous acquisition on the current region, output four polarization direction images, and calculate the Stokes vector components for each pixel point (x, y):

[0114] S0(x, y) = I0(x, y) + I 90 (x, y)

[0115] S1(x, y) = I0(x, y) - I90 (x, y)

[0116] S2(x, y) = I 45 (x, y) - I 135 (x, y)

[0117] wherein, I θ (x, y) is the pixel intensity value collected by the polarization camera, S0(x, y) represents the total light intensity (brightness) of the unpolarized light, S1(x, y) is the horizontal polarization component, and S2(x, y) is the diagonal polarization component;

[0118] Calculate the polarization enhancement image IS(x, y) of the region by combining the obtained Stokes vector components, and express the degree of polarization anomaly:

[0119]

[0120] wherein, ∈ is a small positive number to avoid the denominator being zero;

[0121] Perform linear normalization on the image to map its pixel value range to [0, 1].

[0122] Adopt sRGB gamma inverse linearization and CIE Lab perceptual space conversion to ensure that the basis for region division is close to the human eye visual response and improve the clustering perception consistency; introduce a five-dimensional feature vector joint modeling to achieve color-space coupling division, which is superior to the traditional segmentation scheme based on gray scale or RGB histogram; through the hyperspectral and polarization synchronous acquisition mechanism, the measurement accuracy of the physical properties of the region is improved, providing a data basis for subsequent interference feature analysis and direction response modeling; the system structure is clear, and the acquisition process is highly automated, facilitating actual deployment in the experimental verification platform of camouflage materials or quality inspection lines for use.

[0123] S3, the region mask generation module, is used to perform adaptive segmentation on the polarization reflection intensity map by applying the Otsu algorithm, and use morphological erosion and dilation operations to correct the boundary to generate the target polarization reflection significant region mask IE(x, y);

[0124] Specifically, performing adaptive segmentation on the polarization reflection intensity map by applying the Otsu algorithm and using morphological erosion and dilation operations to correct the boundary to generate the target polarization reflection significant region mask IE(x, y) includes:

[0125] Traverse the entire polarization reflection intensity map, record the minimum value and the maximum value, and perform a per-pixel linear transformation on the image to map it to the [0, 255] gray scale space to obtain the normalized image pixel value;

[0126] Perform differential calculation on the entire row of pixels for each image row:

[0127] Ediff (x,y)=(IS norm (x,y)-IS norm (x,y-1)) 2

[0128] Where, E diff (x,y) is the vertical brightness change intensity, IS norm (x,y) is the normalized brightness value of the image at the coordinate point (x,y), IS norm (x,y-1) is the brightness of the pixel in the previous row in the same column;

[0129] Perform the standard Sobel operator on the entire image to obtain the horizontal gradient map and longitudinal gradient map And calculate the local edge structure strength E of the image edge (x,y):

[0130]

[0131] Construct a pixel energy function, fuse brightness changes and edge suppression information to calculate the pixel-level energy term E(x,y):

[0132] E(x,y)=w1×E diff (x,y)+w2×E edge (x,y)

[0133] Where w1 is the weight of the brightness difference term, w2 is the weight of the Sobel edge term, which are set through experimental tuning;

[0134] The energy of all pixels is accumulated row by row to obtain the row energy integration. The maximum energy row y1 is searched as the structural boundary to divide the image into two sub-images, the upper and lower sub-images. If the pixels in the y-th row are less than y1, it is the upper half of the image, otherwise it is the lower half of the image.

[0135] Perform median filtering on the normalized image to obtain the median map M(x,y). Perform mean filtering on the median map to obtain the local average map G(x,y). Based on M(x,y) and G(x,y), construct the joint grayscale pair (i,j) of each pixel (x,y). Assign the grayscale pairs to the upper and lower half images according to their respective regions. Count the frequencies of all (i,j) in the two regions to form a joint grayscale histogram.

[0136] Define the combined gray - level threshold combination \((s,t)\), where \(s\) is the gray - level threshold candidate of the combined median image \(M(x,y)\) and \(t\) is the gray - level threshold candidate of the combined mean image \(G(x,y)\). The actual range is dynamically adjusted according to the gray - level distribution of the image. Each pair \((s,t)\) represents the boundary that divides the value ranges of \(M(x,y)\) and \(G(x,y)\) into two sub - regions. In each histogram, \((i,j)\) is divided into 4 regions according to \((s,t)\) as follows:

[0137] Region A: \(i\leq s\) and \(j\leq t\); Region B: \(i > s\) and \(j\leq t\); Region C: \(i\leq s\) and \(j > t\); Region D: \(i > s\) and \(j > t\).

[0138] Calculate the sum of the relative frequencies of all gray - level pairs \((i,j)\) belonging to region \(k\) in the entire image in the combined gray - level histogram:

[0139] \(\omega\) k \(=\sum p\) ij

[0140] where \(p\) ij is the probability of the gray - level pair \((i,j)\), and \(\omega\) k is the region weight (combined probability sum);

[0141] Taking the gray - level value of the median - filtered image \(M(x,y)\) as the main variable, calculate the combined gray - level mean \(\mu\) k of each region and the overall mean \(\mu\) T as follows:

[0142]

[0143] \(\mu\) T \(=\sum i\cdot p\) ij

[0144] where \(i\) and \(j\) represent the gray - level indices of the median image and the mean image in the combined gray - level histogram respectively;

[0145] According to the two - dimensional Otsu definition, calculate the comprehensive scoring index of the between - class difference:

[0146]

[0147] where \(tr(\sigma\) B ) is the weighted mean square deviation of the between - class variance;

[0148] For each \((s,t)\), calculate the trace value of the between - class variance of the upper and lower half - images of the combined gray - level histogram respectively, and multiply the two trace values to obtain the combined scoring function \(Q(s,t)\):

[0149]

[0150] where and is are the inter-class covariance matrices obtained after using the current threshold pair in the “upper subgraph” and “lower subgraph” respectively;

[0151] This score reflects the likelihood that the segmentation is optimal in both regions simultaneously.

[0152] Traverse all the scoring results and find the threshold combination with the highest score. This threshold pair is the optimal segmentation boundary of the upper and lower images.

[0153] In the upper and lower sub-images, traverse each pixel (x, y) respectively to construct the upper and lower sub-image masks:

[0154]

[0155] Where, and is the binary mask pixel value;

[0156] Merge the upper and lower area masks into the full image mask M0(x,y):

[0157]

[0158] Use a common connected component labeling algorithm (such as OpenCV's connectedComponents) to label the connected regions of the mask map M0(x,y) (4-neighborhood or 8-neighborhood), remove areas with an area smaller than a preset threshold (set by statistical analysis), and perform morphological closing and opening operations on the mask map after elimination to obtain the final polarization-significant region mask map.

[0159] First, by normalizing the polarized reflection intensity map and calculating the luminance difference term, the rapid detection of abrupt polarized luminance boundaries can be achieved. The pixel energy term constructed by combining the Sobel gradient response avoids, to a certain extent, the over-segmentation or under-segmentation problems caused by luminance alone. Further, by calculating the peak energy row of the full image through the row energy accumulation curve, the image can be automatically divided into upper and lower parts, and joint gray histograms are established on their respective sub-images. This not only realizes region-level optimized segmentation but also significantly reduces the redundant influence caused by uneven texture in global optimization. By modeling the between-class variance of the four types of gray pairs (regions A, B, C, and D) in the joint histogram and introducing the product of the trace values of the covariance matrix as the joint scoring index, this method realizes the global search for the optimal threshold combination in the median-mean joint two-dimensional space, solving the problem of threshold offset in the traditional Otsu algorithm when dealing with low-contrast and blurred boundary images. Finally, through the construction and merging of binary masks in the upper and lower regions, the system obtains a complete region mask map. This mask can accurately label the position and shape of the high-polarized reflection region, providing an accurate spatial reference for subsequent interference fringe analysis and BRDF angular response modeling. It is particularly worth emphasizing that through subsequent connected component analysis and area filtering operations (such as using the connectedComponents function in OpenCV), noise regions and isolated pixel clusters can be effectively removed, and the geometric consistency and usability of the mask map are further improved by morphological opening and closing operations.

[0160] S4. An interference scoring module, configured to select hyperspectral reflection curves within the range of IE(x, y), detect periodic fringe fluctuations using fast Fourier transform, and calculate the interference score Ir(i);

[0161] Specifically, selecting hyperspectral reflection curves within the range of IE(x, y) and detecting periodic fringe fluctuations using fast Fourier transform and calculating the interference score Ir(i) means traversing each sub-region number i. If the corresponding sample is a structural color material, then the pixel points marked by the IE(x, y) mask within this region are screened, and only the spectral data in the "significantly polarized enhanced region" are retained. The hyperspectral reflectance curves at all mask points within this region are averaged to obtain a representative reflection curve. For the curve of the i-th region Perform normalization processing, and apply one-dimensional fast Fourier transform operation to the normalized curve to obtain the amplitude spectrum value F i (k) at the frequency index k. To quantify the intensity of the interference fringes, calculate the average energy of the high-frequency band (removing DC and low-frequency components) as the interference score Ir(i) of this region:

[0162]

[0163] Where H is the total number of frequency points within the integration interval, k is the frequency index in the Fourier transform, k1 is the starting point of frequency integration, and k2 is the ending point of frequency integration;

[0164] The larger the evaluation score value, the stronger the high-frequency interference component in the reflection curve and the more significant the potential non-natural structure;

[0165] The Ir(i) of all structural color regions forms an interference scoring matrix.

[0166] This solution introduces FFT spectral analysis to directly quantify the high-frequency energy term, effectively distinguishing the camouflage materials with "interference characteristics but not obvious spectral main types". The determination of existing microscopic interference structures usually relies on contact devices such as scanning electron microscopes or white light interferometers, which are not suitable for rapid screening. Through the design of this system, only by collecting the reflectivity curve on the material surface can the interference behavior evaluation based on spectral data be realized, which has extremely strong practicability and engineering deployment value. As a quantitative output index of the high-frequency fringe intensity, Ir(i) can directly participate in the calculation of the comprehensive scoring function of the camouflage performance. The higher its value, the stronger the interference and the signs of artificial structure in the reflectivity curve of this area, which can be used as an important quantitative basis for identifying "optical exposure hotspots".

[0167] S5. The multi-angle BRDF measurement module is used to change the observation angle of the sample, perform multi-angle spectral acquisition on the masked area, establish the BRDF response function, and calculate the directional deviation ΔR(i);

[0168] Specifically, changing the observation angle of the sample, performing multi-angle spectral acquisition on the masked area, establishing the BRDF response function, and calculating the directional deviation ΔR(i) means controlling the three-axis servo motor platform to adjust the sample posture to fix the incident light source. In this embodiment, the receiving angles of the detector are set to 0°, 30°, 60°, and 80° respectively. After each angle setting is completed, using the hyperspectral imaging module, under the condition of maintaining the standard illuminance, collect the hyperspectral reflectivity data of each sub-region within the masked area IE(x,y). For each sub-region i, at each angle θ k calculate its BRDF function value f r,i (θ k ,λ):

[0169]

[0170] [[ID=epub:type]] Where R(θ k is the reflectivity of the material at the angle θ k , and cos(θ k ) is the angle projection coefficient;

[0171] Then, perform averaging in the wavelength dimension to obtain the band-averaged BRDF response of the angle-region

[0172]

[0173] where M is the number of sampling bands, and λ j is the j-th band;

[0174] For each sub-region i, the average BRDF response at all angles is averaged

[0175]

[0176] Calculate the directional consistency deviation ΔR(i) of sub-region i:

[0177]

[0178] where is the average BRDF response of the i-th sub-region at the angle θ k under, is the average value of BRDF at all angles. ΔR(i) is the maximum deviation degree of this region at different angles, which is the inverse index of directional camouflage consistency. The larger the value, the worse the specular reflection consistency and the weaker the camouflage effect.

[0179] Traditional reflection detection is usually carried out at normal or near-normal angles and cannot capture the high-contrast reflection problem of materials at oblique viewing angles. The present invention introduces a multi-angle difference analysis mechanism through ΔR(i) and for the first time incorporates the "directional exposure risk" as a quantitative index into the camouflage performance evaluation system. There are non-ideal microstructures (such as semi-regular particles, micro-textured coatings) on the surfaces of many new camouflage materials. Such materials may reflect well when observed at small angles, but are prone to specular reflection effects under large-angle conditions. ΔR(i) can effectively capture such directional specular reflection risks. The ΔR(i) index can not only be used for material quality determination, but also assist in battlefield deployment strategies. For example, it is recommended that a certain material be used only in specific angle shielding areas under a specific background to improve the camouflage efficiency.

[0180] S6. A scoring module, which is used to comprehensively score the camouflage performance based on the intensity index Ir(i) and the directional deviation ΔR(i), generate a two-dimensional heat map according to the camouflage performance score, display the camouflage exposure risk level and automatically generate a detection report;

[0181] Specifically, a comprehensive score of the camouflage performance is carried out based on the intensity index Ir(i) and the directional deviation ΔR(i). A two-dimensional heat map is generated according to the camouflage performance score to display the camouflage exposure risk level and automatically generate a detection report. That is, a linear weighting is performed on the interference score value Ir(i) and the directional deviation score ΔR(i) of each sub-region i to construct a comprehensive optical camouflage performance score Score(i). The scoring results of all sub-regions are normalized. The maximum score is mapped to 1, and the minimum score is mapped to 0 to form a normalized scoring matrix. Each sub-region uses its regional boundary as a primitive, and the normalized scoring matrix value controls its filling color (0: green, 1: red). OpenCV is used to automatically draw the final graph, and the original material image is overlaid to form a transparent layer heat map, and a detection report is generated according to the detection data, including the basic information of the sample, the image of the material surface structure and the judgment result of the structural color, the standard illumination parameters and the reflectance response curve, the polarization map (IS map) and the extraction area mask map, the interference score corresponding to each area, the directional deviation and the comprehensive score.

[0182] The combination of the interference score and the directional deviation makes the scoring system span the dimensions of the microscopic structure optical effect and the macroscopic reflection consistency, with high complementarity. The structural color interference reflects the manufacturing defects or design weaknesses at the microscopic level of the material, while the multi-angle consistency reflects the stability of the macroscopic morphology and the coating from the perspective of use. Together, they determine the identifiability of the material under complex natural conditions. The scoring mechanism of the present invention is calculated based on clear physical quantities, and the linear weighting structure avoids overfitting or fuzzy judgment of the model to adapt to the differences in the emphasis on structural interference or direction consistency in different scenarios.

[0183] This embodiment also provides an optical performance detection method for a camouflage material, including:

[0184] Collect the surface image of the camouflage material sample, perform RGB imaging based on the standard illumination and automatically divide the sub-regions, perform hyperspectral reflectance scanning on each region, synchronously collect the polarization direction light intensity image, calculate the Stokes vector and construct a polarization enhanced image;

[0185] Perform adaptive segmentation on the polarization reflection intensity map and correct the boundary to generate a target polarization reflection significant region mask IE(x, y);

[0186] Select the hyperspectral reflection curve within the range of IE(x, y), detect the periodic stripe fluctuation and calculate the interference score Ir(i). Change the observation angle of the sample, perform multi-angle spectral acquisition on the masked area, establish a BRDF response function, and calculate the directional deviation ΔR(i);

[0187] Based on the intensity index Ir(i) and the directional deviation ΔR(i), a comprehensive score of the camouflage performance is carried out. According to the camouflage performance score, a two-dimensional heat map is generated to display the camouflage exposure risk level and automatically generate a detection report.

[0188] This embodiment also provides a computer device, which is applicable to the case of the optical performance detection method of the camouflage material, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the optical performance detection method of the camouflage material as proposed in the above embodiment.

[0189] This computer device can be a terminal. The computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a carrier network, NFC (Near Field Communication), or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covered on the display screen, or a button, a trackball, or a touchpad set on the shell of the computer device, or an external keyboard, a touchpad, or a mouse, etc.

[0190] This embodiment also provides a storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the optical performance detection method of the camouflage material as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as a static random access memory (Static Random Access Memory, abbreviated as SRAM), an electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, abbreviated as EEPROM), an erasable programmable read-only memory (Erasable Programmable Read Only Memory, abbreviated as EPROM), a programmable read-only memory (Programmable Red-Only Memory, abbreviated as PROM), a read-only memory (Read-Only Memory, abbreviated as ROM), a magnetic memory, a flash memory, a magnetic disk, or an optical disk.

[0191] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. An optical property detection system for a camouflage material, characterized in that: include, The structure recognition and lighting control module is used to collect surface images of camouflage material samples, identify the sample surface structure, activate the full-spectrum light source, simulate gradual lighting from weak light to strong light, and perform RGB imaging based on standard illumination; The region segmentation module is used to apply the superpixel segmentation algorithm to the image, automatically divide the subregions, perform hyperspectral reflectance scanning on each region, synchronously collect polarization direction light intensity images, calculate the Stokes vector and construct the polarization enhanced image; The regional mask generation module is used to perform adaptive segmentation on the polarized reflection intensity map using the Otsu algorithm, modify the boundary using morphological erosion and dilation operations, and generate the target polarized reflection significant region mask IE(x,y); Interference scoring module, which is used to select the hyperspectral reflectance curve within the IE(x,y) range, detect the periodic fringe fluctuations using fast Fourier transform and calculate the interference score Ir(i); Multi-angle BRDF measurement module, used to change the sample observation angle, collect multi-angle spectra of the mask area, establish the BRDF response function, and calculate the directional deviation ΔR(i); The scoring module is used to comprehensively score the camouflage performance based on the intensity index Ir(i) and the directional deviation ΔR(i). It generates a two-dimensional heat map based on the camouflage performance score, displays the camouflage exposure risk level, and automatically generates a detection report.

2. The optical property detection system of the camouflage material according to claim 1, characterized in that: The method of applying the Otsu algorithm to adaptively segment the polarized reflection intensity map, modifying the boundary using morphological corrosion and dilation operations, and generating the target polarized reflection significant area mask IE(x, y) includes: Traverse the entire polarization reflection intensity map and normalize it to obtain the normalized image pixel value; Perform differential calculations on the entire row of pixels for each image row; A standard Sobel operator is applied to the entire image to obtain horizontal and vertical gradient maps and calculate the local edge structure strength of the image. A pixel energy function is constructed, which combines brightness changes with edge suppression information to calculate the pixel-level energy term E(x,y). The energy of all pixels is accumulated row by row to obtain the row energy integration. The maximum energy row y1 is searched as the structural boundary to divide the image into two sub-images, the upper and lower sub-images. If the pixels in row y are less than y1, it is the upper half of the image; otherwise, it is the lower half of the image. Perform median filtering on the normalized image to obtain the median map M(x,y). Perform mean filtering on the median map to obtain the local average map G(x,y). Based on M(x,y) and G(x,y), construct the joint grayscale pair (i,j) of each pixel (x,y). Assign the grayscale pairs to the upper and lower half images according to their respective regions. Count the frequencies of all (i,j) in the two regions to form a joint grayscale histogram. Define the joint grayscale threshold combination (s, t), and in each histogram, divide (i, j) into 4 regions according to (s, t); Calculate the sum of the relative frequencies of all gray - level pairs (i, j) belonging to region k in the entire image in the joint gray - level histogram. Using the gray - level value of the median - filtered image M(x, y) as the main variable, calculate the joint gray - level mean μ k of each region and the overall mean μ T . According to the two - dimensional Otsu definition, calculate the comprehensive scoring index of the between - class difference. For each (s, t), calculate the trace value of the between - class variance of the upper and lower half - images of the joint gray - level histogram respectively, and multiply the two trace values to obtain the joint scoring function Q(s, t); Traverse all scoring results and find the threshold combination with the highest score; In the upper and lower sub-images, traverse each pixel (x, y) separately, construct the upper and lower sub-image masks, and merge the upper and lower area masks into the full image mask M0(x, y); Perform connected component labeling on the mask image M0(x, y), remove regions with an area smaller than a preset threshold, and perform morphological closing and opening operations on the masked image after removal to obtain the final polarization significant region mask image.

3. The optical property detection system of the camouflage material according to claim 2, wherein: Collect the surface image of the camouflage material sample, identify the surface structure of the sample, activate the full-spectrum light source, simulate step-by-step illumination from weak light to strong light, perform RGB imaging based on the standard illuminance, and apply the superpixel segmentation algorithm to the image to automatically divide sub-regions including: Collect the high-resolution grayscale image of the sample surface, perform a two-dimensional fast Fourier transform on the image to obtain a frequency spectrum image, calculate the frequency spectrum amplitude image, normalize it, and extract the main frequency corresponding to the direction with the largest amplitude in the frequency spectrum. If the main frequency is greater than the preset threshold U, further evaluate the energy proportion of the main frequency. If the energy proportion of the main frequency is greater than the preset threshold u, it is considered that this region has the possibility of artificial structural color interference. Activate the controllable full-spectrum LED array light source and set the light intensity to five fixed levels, including I1 to I5. At each level of illuminance, the reflectance spectral curve of the surface of the material sample is collected respectively, and the normalized reflectance R(λ j ,I i ) is calculated based on the collected data. For all wavelength reflectances R(λ, I i ) at the i-th illuminance, its mean value and the standard deviation σ(I i ) of the reflectance at this illuminance are calculated. The five groups of standard deviation results are compared, and the illuminance level number corresponding to the smallest one is selected as the standard illuminance. Based on the determined standard illuminance, the industrial camera system is started to collect the sample images, and a three-channel RGB image is obtained.

4. The optical property detection system for the camouflage material according to claim 3, characterized in that: Apply the superpixel segmentation algorithm to the image to automatically divide sub-regions, perform hyperspectral reflectance scanning on each region, synchronously collect the polarization direction light intensity image, calculate the Stokes vector, and construct the polarization enhanced image including: Individually process each component of each pixel according to the standard sRGB gamma inverse correction to obtain the RGB image matrix after light intensity linearization. For each pixel, convert the linear R, G, and B values into X, Y, and Z channel values, normalize the XYZ components of each pixel point, define the perceptual mapping function f(p) to establish a non-linear color-brightness response, calculate the Lab channel values of each pixel point, and output the three-channel matrix I in the Lab space Lab (x,y); Calculate the clustering window spacing S according to the preset number of regions K, and construct a two-dimensional spatial coordinate matrix for each pixel (x, y) in the image. For each pixel point (x, y) in the image, construct a five-dimensional joint feature vector V required for SLIC clustering x,y ; Combine the V x,y features of all pixels into a five-dimensional tensor structure T V (x, y); Based on S, initialize K clustering center points by uniform grid division on the image. For each clustering center, calculate the composite distance between the current pixel and the center, assign the pixel (x, y) to the clustering center number with the smallest distance, recalculate the five-dimensional feature mean of each region as the new center, calculate the moving distance m between the old center and the new center. If all m are less than the preset threshold β, stop clustering to obtain the final pixel assignment mask matrix. For each region corresponding to the number i, perform an 8-neighborhood connectivity analysis. After using the contour tracking algorithm to extract the set of closed boundary points of the main connected component and saving it as a regional vector mask, calculate the regional centroid coordinates Extract the centroid coordinates of each sub-region and set them as the target sampling points. Perform pixel-by-pixel scanning on each sub-region to obtain a two-dimensional spectral matrix. Activate the DoFP type polarization camera to output four polarization direction images, and calculate the Stokes vector components for each pixel point (x, y). Calculate the polarization enhanced image IS(x, y) of this region by combining the obtained Stokes vector components.

5. The optical property detection system of the camouflage material according to claim 4, characterized in that: Select the hyperspectral reflection curves within IE(x,y), and use the fast Fourier transform to detect periodic fringe fluctuations and calculate the interference score Ir(i). That is, traverse each sub-region number i. If the corresponding sample is a structural color material, screen the pixel points marked by the IE(x,y) mask in this region, and only retain the spectral data in the significantly polarization-enhanced region. Then, perform mean synthesis on the hyperspectral reflectance curves at all mask points in this region to obtain a representative reflectance curve. And apply the one-dimensional fast Fourier transform operation to obtain the amplitude spectrum value F(k) at the frequency index k. i (k), and calculate the average energy of the high-frequency band as the interference score Ir(i) of the region.

6. The optical property detection system of the camouflage material according to claim 5, characterized in that: Changing the observation angle of the sample, collecting multi-angle spectra of the mask area, establishing a BRDF response function, and calculating the directional deviation ΔR(i) means controlling the three-axis servo motor platform to adjust the sample posture to fix the incident light source. After each angle setting is completed, using the hyperspectral imaging module, under the condition of maintaining the standard illuminance, collect the hyperspectral reflectance data of each sub-region within the mask area IE(x, y). For each sub-region i, at each angle θ k , calculate the BRDF function value f r,i (θ k , λ); Average in the wavelength dimension to obtain the band-averaged BRDF response of the angle-region For each sub-region i, the average BRDF response at all angles is averaged Based on and Calculate the directional consistency deviation ΔR(i) of sub-region i.

7. The optical property detection system of the camouflage material according to claim 6, wherein: Perform a comprehensive evaluation of the camouflage performance based on the intensity index Ir(i) and the directional deviation ΔR(i). Generate a two-dimensional heat map according to the camouflage performance score to display the camouflage exposure risk level and automatically generate a detection report. That is, perform linear weighting on the interference score value Ir(i) and the direction deviation score ΔR(i) of each sub-region i to construct a comprehensive optical camouflage performance score Score(i). Map the maximum score to 1 and the minimum score to 0 to form a normalized score matrix. Each sub-region uses the region boundary as the primitive and fills the color with the normalized score matrix value to form a transparent layer heat map, and generate a detection report according to the detection data.

8. A method for detecting the optical properties of a camouflage material, based on the optical property detection system of the camouflage material according to any one of claims 1 to 7, characterized in that: Including, Collect the surface images of the camouflage material samples, perform RGB imaging based on the standard illuminance and automatically divide the sub-regions, conduct hyperspectral reflectance scanning on each region, synchronously collect the polarization-direction light intensity images, calculate the Stokes vector and construct the polarization-enhanced image; Perform adaptive segmentation on the polarization reflection intensity map and correct the boundaries to generate the target polarization reflection significant region mask IE(x,y); Select the hyperspectral reflection curves within the range of IE(x,y), detect the periodic stripe fluctuations and calculate the interference score Ir(i), change the sample observation angle, perform multi-angle spectral acquisition on the masked region, establish the BRDF response function, and calculate the directional deviation ΔR(i); Conduct a comprehensive evaluation of the camouflage performance based on the intensity index Ir(i) and the directional deviation ΔR(i), generate a two-dimensional heat map according to the camouflage performance score, display the camouflage exposure risk level and automatically generate a detection report.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the optical performance detection method of the camouflage material according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the optical performance detection method of the camouflage material according to any one of claims 1 to 7.

Images