A night vision imaging method and system for micro-light and infrared image fusion

Abstract

The present application relates to the technical field of image processing, and particularly relates to a night vision imaging method and system for micro-light and infrared image fusion. The method realizes multi-source data cooperation by constructing a dual-band polarization characteristic cube, accurately quantifies the degradation law of fog scattering on micro-light texture, and generates a scattering-invariant enhanced image. Based on the fog concentration gradient dynamic partition, the micro-light details and infrared outlines are fused in the low concentration area, and the polarization state is reorganized to penetrate thick fog in the high concentration area. Finally, the night vision image with high definition, strong anti-scattering property and scene adaptability is output, and the recognition ability and monitoring reliability of foggy night road targets are improved.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to a night vision imaging method and system that fuses low-light and infrared images. Background Technology

[0002] In foggy road environments, traditional night vision imaging technologies are limited by fog scattering, low illumination, and limited spectral information, making it difficult to achieve clear and reliable scene perception. While low-light imaging can enhance visible light information, it is susceptible to Mie scattering, leading to texture degradation. Infrared imaging, although capable of penetrating fog, lacks detailed texture and polarization characteristics, making it difficult to meet the monitoring needs of complex environments. Existing fusion methods are mostly based on single light intensity or radiation characteristics, failing to fully utilize the scattering invariance of polarization information and exhibiting insufficient adaptability to dynamic changes in fog concentration, resulting in problems such as loss of detail, blurred edges, or thermal radiation distortion in the fused images. Furthermore, traditional fusion algorithms often ignore the physical relationship between polarization state and optical path difference, making it difficult to effectively correct scattering degradation, especially in dense fog areas where the complementarity of infrared and low-light information is not fully explored. In view of this, this application proposes a night vision imaging method and system for fusing low-light and infrared images, achieving optimized fusion of multimodal information under fog interference, and improving imaging clarity and scene understanding capabilities. Summary of the Invention

[0003] This invention overcomes the shortcomings of the prior art and provides a night vision imaging method and system that fuses low-light and infrared images.

[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0005] The first aspect of this invention discloses a night vision imaging method that fuses low-light and infrared images, comprising the following steps:

[0006] S102: In foggy night road environments, simultaneously acquire low-light polarization images and long-wave infrared polarization images to construct a spatiotemporally aligned dual-band polarization feature cube;

[0007] S104: Based on the dual-band polarization feature cube, the fog concentration gradient tensor of the monitoring scene is extracted and the degraded texture feature map modulated by Mie scattering is separated. The optical path difference mapping relationship between the low-light degraded texture and the fog concentration field is established, and the optical path difference mapping relationship matrix is ​​generated.

[0008] S106: Generate a joint compensation operator based on the optical path difference mapping relationship matrix, and combine the compensation operator to perform polarization channel transmittance correction on the micro-light degradation texture, and output an enhanced micro-light feature map with fog scattering invariance.

[0009] S108: Analyze the fog concentration gradient tensor and dynamically divide the low-concentration region and high-concentration region based on the Laplace norm response value of the fog concentration gradient tensor.

[0010] S110: In the low-concentration region, the enhanced low-light feature map and the infrared thermal radiation intensity are fused using gradient tensor fidelity; in the high-concentration region, the polarization state tensor is recombined with the Stokes component of the enhanced low-light feature map to generate a night vision imaging map.

[0011] Preferably, 102 specifically refers to:

[0012] Simultaneously collect light intensity data in four polarization directions of the low-light band and radiation intensity data in two orthogonal polarization directions of the long-wave infrared band, and generate low-light polarization intensity image group and long-wave infrared polarization radiation intensity image group respectively.

[0013] Stokes vector calculation was performed on the low-light polarization image group to obtain the low-light intensity component and polarization characteristic component; thermal radiation polarization calculation was performed on the long-wave infrared polarization image group to obtain the infrared radiation intensity component and polarization degree component.

[0014] Based on the spatial gradient characteristics of the low-light intensity component and the thermal radiation profile characteristics of the infrared radiation intensity component, a spatial registration matrix is ​​generated through affine transformation to achieve sub-pixel-level alignment of the two band images; the polarization correlation between the bands is determined by the low-light polarization characteristic component and the infrared polarization degree component, and a fusion weight matrix is ​​generated based on the polarization correlation.

[0015] The infrared radiation intensity component is geometrically corrected using a spatial registration matrix to obtain the corrected infrared radiation intensity; the polarization direction of the infrared polarization component is compensated to eliminate the polarization angle shift caused by parallax, and the corrected infrared polarization is output.

[0016] The corrected infrared radiation intensity and the micro-light intensity component are fused according to the fusion weight matrix to generate fused light intensity features; the corrected infrared polarization degree and the micro-light polarization characteristic component are recombined in polarization state to generate fused polarization features.

[0017] By superimposing the fused light intensity characteristics, the corrected infrared radiation intensity, and the fused polarization characteristics along the spatial dimension, a six-channel dual-band polarization feature cube containing light intensity, radiation, and polarization information is constructed.

[0018] Preferably, the 104 specifically refers to:

[0019] A pixel-by-pixel differential operation is performed on the micro-light intensity component and the infrared radiation intensity component in the dual-band polarization feature cube to obtain the dual-band differential intensity parameter. The polarization difference ratio is calculated using the Stokes parameter of the micro-light polarization characteristic component to obtain the polarization modulation difference parameter.

[0020] An initial fog concentration distribution map is generated based on the polarization modulation difference parameter and the dual-band differential intensity parameter; anisotropic diffusion filtering is applied to the initial fog concentration distribution map, and combined with the thermal radiation polarization constraint condition of the infrared polarization degree component, noise is suppressed and the fog edge structure is preserved, and the optimized fog concentration gradient tensor is output.

[0021] By utilizing the statistical correlation between the Stokes vector phase angle of the micro-light polarization characteristic component and the light intensity attenuation, a mapping relationship between polarization modulation and light intensity degradation is established. Through this mapping relationship, the degraded texture feature map modulated by Mie scattering is decoupled from the micro-light intensity component.

[0022] A three-dimensional distribution model of fog concentration field is constructed based on the optimized fog concentration gradient tensor. At the same time, the optical path accumulation of the low-light degraded texture in the fog concentration field is determined according to the local contrast attenuation rate of the degraded texture feature map.

[0023] A mapping relationship between the gray-level attenuation gradient of the degraded texture feature map and the cumulative optical path length of the fog concentration field is established by nonlinear regression method, and a calibrated optical path difference mapping relationship matrix is ​​generated. Each element of the matrix represents the degree of scattering degradation of the micro-light texture under a specific fog concentration.

[0024] Preferably, a polarization modulation-light intensity degradation mapping relationship is established by utilizing the statistical correlation between the Stokes vector phase angle of the micro-light polarization characteristic component and the light intensity attenuation. This mapping relationship is then used to decouple the degraded texture feature map modulated by Mie scattering from the micro-light intensity component. Specifically:

[0025] The Stokes vectors of the micro-light polarization characteristic components in the dual-band polarization feature cube are extracted, and the phase angle distribution map of the Stokes vectors is obtained. At the same time, the local contrast attenuation rate matrix of the micro-light intensity components is obtained.

[0026] The phase angle distribution map is normalized to generate a polarization modulation phase feature map, and a polarization phase-attenuation statistical coupling relationship is established by combining the local contrast attenuation rate matrix and Pearson correlation analysis.

[0027] Based on the statistical coupling relationship between polarization phase and attenuation, the mapping curve between the polarization modulation phase feature map and the local contrast attenuation rate matrix is ​​fitted using the nonlinear least squares method to obtain the coupling coefficient matrix between polarization phase and attenuation.

[0028] The polarization phase-attenuation coupling coefficient matrix is ​​applied to the micro-light intensity component, and the global attenuation substrate layer dominated by Mie scattering is separated by inverse mapping operation to obtain the residual texture component.

[0029] Anisotropic guided filtering is applied to the residual texture components, and the polarization degree component of the Stokes vector is used as the edge constraint weight to suppress noise and preserve the topology of the scattering degradation texture, outputting a degradation texture feature map modulated by Mie scattering.

[0030] By performing a differential operation between the degraded texture feature map and the global attenuation substrate, the linear independence between the light intensity attenuation gradient and the polarization modulation phase angle is verified, and finally the scattering modulation domain of the degraded texture feature map is calibrated.

[0031] Preferably, the 106 specifically refers to:

[0032] Based on the cumulative optical path of each pixel in the optical path difference mapping matrix, the polarization channel transmittance attenuation coefficient corresponding to the micro-light degradation texture feature map is determined, and a transmittance attenuation coefficient distribution map is generated.

[0033] By combining the polarization phase angle distribution of the Stokes vector, a nonlinear coupling relationship between the transmittance attenuation coefficient and the polarization phase angle is established, and a polarization-transmittance coupling tensor is generated.

[0034] By utilizing the infrared radiation intensity component in the dual-band polarization feature cube, the penetration characteristic curve of thermal radiation in the fog medium is extracted. By calibrating the inverse proportional relationship between infrared thermal radiation intensity and fog concentration, a thermal radiation penetration compensation coefficient matrix is ​​constructed.

[0035] The polarization-transmittance coupling tensor and the thermal radiation penetration compensation coefficient matrix are multiplied by tensor dot product to obtain a joint compensation operator that integrates polarization modulation characteristics and thermal radiation penetration characteristics.

[0036] The joint compensation operator is applied to the low-light degradation texture feature map. By correcting the transmittance attenuation coefficient pixel by pixel, the polarization channel energy attenuation caused by Mie scattering is eliminated, and the intermediate low-light texture after transmittance correction is output.

[0037] The polarization degree component of the Stokes vector is used to enhance the edges of the intermediate micro-light texture. The local contrast lost due to scattering is restored by polarization degree weighting, and an enhanced micro-light feature map with fog scattering invariance is generated.

[0038] Preferably, the 108 specifically refers to:

[0039] The Laplacian operator convolution operation is performed on the fog concentration gradient tensor to calculate the second spatial derivative at each pixel location, generating a Laplacian response map of fog concentration.

[0040] Based on the local extreme value distribution characteristics of the Laplace response diagram, the abrupt boundary of fog concentration change is extracted to form the concentration partition boundary;

[0041] The concentration partition boundaries are smoothed by morphological closing operations to eliminate holes caused by noise. Combined with the spatial consistency constraints of infrared polarization components, the mis-segmented regions caused by thermal radiation interference are corrected to generate a dynamic partition mask.

[0042] Based on the high and low concentration regions marked in the dynamic partition mask, the pixels in the monitored scene are classified into low-concentration areas and high-concentration areas; where the low-concentration area is the area with a mask value of 0, and the high-concentration area is the area with a mask value of 1.

[0043] Preferably, the 110 specifically refers to:

[0044] In the low-concentration region, the fog concentration gradient tensor of the enhanced low-light feature map is extracted, and the thermal radiation contour gradient field is parsed from the infrared thermal radiation intensity. The fog concentration gradient tensor and the thermal radiation contour gradient field are measured pixel by pixel gradient structure similarity to generate a gradient fidelity weight map.

[0045] Based on the gradient fidelity weight map, the low-frequency component of the enhanced low-light feature map and the high-frequency component of the infrared thermal radiation intensity are adaptively weighted and fused to obtain the primary fused feature.

[0046] The gradient field is rebalanced on the primary fusion features by utilizing the local anisotropy coefficient of the fog concentration gradient tensor to eliminate radiation distortion at the fusion boundary and output the fusion result in the low concentration region.

[0047] Also includes:

[0048] In the high-concentration region, the spatial distribution matrix of the infrared polarization vector direction angle is calculated, and the polarization phase angle feature in the Stokes component of the enhanced micro-light feature map is extracted. The spatial distribution matrix and the polarization phase angle feature are subjected to polarization state covariance analysis to establish a polarization direction-phase coupling tensor.

[0049] The infrared polarization direction angle is transformed by Stokes space projection using the coupling tensor to generate a recombinant polarization basis.

[0050] By dynamically adjusting the weight ratio of the reconstituted polarization basis to the enhanced low-light feature map using the Laplace norm response value of the fog concentration gradient tensor, tensor product operation of the polarization channels is performed to obtain the polarization recombination features in the high-concentration region.

[0051] Finally, the fusion results of the low-concentration area and the polarization recombination features of the high-concentration area are spatially stitched together using a dynamic partitioning mask, and the final night vision image is generated after a smooth transition by bilinear interpolation.

[0052] The second aspect of the present invention discloses a night vision imaging system for fusing low-light and infrared images. The night vision imaging system includes a memory and a processor. The memory stores a night vision imaging method program for fusing low-light and infrared images. When the night vision imaging method program for fusing low-light and infrared images is executed by the processor, any one of the steps of the night vision imaging method for fusing low-light and infrared images is implemented.

[0053] This invention addresses the technical deficiencies in the prior art and possesses the following beneficial effects: by constructing a dual-band polarization feature cube to achieve multi-source data collaboration, it accurately quantifies the degradation law of low-light texture caused by fog scattering, generating scattering invariant enhanced images; based on dynamic zoning of fog concentration gradient, it fuses low-light details and infrared contours in low-concentration areas, and reconstructs polarization states to penetrate dense fog in high-concentration areas, ultimately outputting night view images with high definition, strong anti-scattering properties, and scene adaptability, thereby improving the identification capability and monitoring reliability of road targets in foggy nights. Attached Figure Description

[0054] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other embodiments can be obtained from these drawings without creative effort.

[0055] Figure 1 This is a flowchart illustrating the overall method of this night vision imaging technique.

[0056] Figure 2 This is a partial flowchart of the night vision imaging method.

[0057] Figure 3 This is a system block diagram of the night vision imaging system. Detailed Implementation

[0058] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0059] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0060] like Figure 1 As shown, the first aspect of this invention discloses a night vision imaging method that fuses low-light and infrared images, comprising the following steps:

[0061] S102: In foggy night road environments, simultaneously acquire low-light polarization images and long-wave infrared polarization images to construct a spatiotemporally aligned dual-band polarization feature cube;

[0062] S104: Based on the dual-band polarization feature cube, the fog concentration gradient tensor of the monitoring scene is extracted and the degraded texture feature map modulated by Mie scattering is separated. The optical path difference mapping relationship between the low-light degraded texture and the fog concentration field is established, and the optical path difference mapping relationship matrix is ​​generated.

[0063] S106: Generate a joint compensation operator based on the optical path difference mapping relationship matrix, and combine the compensation operator to perform polarization channel transmittance correction on the micro-light degradation texture, and output an enhanced micro-light feature map with fog scattering invariance.

[0064] S108: Analyze the fog concentration gradient tensor and dynamically divide the low-concentration region and high-concentration region based on the Laplace norm response value of the fog concentration gradient tensor.

[0065] S110: In the low-concentration region, the enhanced low-light feature map and the infrared thermal radiation intensity are fused using gradient tensor fidelity; in the high-concentration region, the polarization state tensor is recombined with the Stokes component of the enhanced low-light feature map to generate a night vision imaging map.

[0066] It should be noted that this invention addresses the problems of low-light image texture degradation, infrared image detail loss, and poor dual-band fusion adaptability caused by fog scattering in traditional night vision imaging in foggy road environments. It overcomes the limitations of single-mode information and imaging distortion defects under dynamic fog interference. By constructing a dual-band polarization feature cube to achieve multi-source data collaboration, it accurately quantifies the degradation law of low-light texture caused by fog scattering, generating scattering-invariant enhanced images. Based on dynamic partitioning of fog concentration gradient, it fuses low-light details and infrared contours in low-concentration areas, and reconstructs polarization states to penetrate dense fog in high-concentration areas. The final output is a night vision image with high definition, strong anti-scattering properties, and scene adaptability, improving the identification capability and monitoring reliability of targets on foggy roads.

[0067] Preferably, 102 specifically refers to:

[0068] Simultaneously acquire light intensity data in four polarization directions (0°, 45°, 90°, 135°) of the low-light band (400-700nm), and radiation intensity data in two orthogonal polarization directions (horizontal and vertical) of the long-wave infrared band (8-14μm), and generate low-light polarization intensity image group and long-wave infrared polarization radiation intensity image group respectively.

[0069] Stokes vector calculation was performed on the low-light polarization image group to obtain the low-light intensity component and polarization characteristic component; thermal radiation polarization calculation was performed on the long-wave infrared polarization image group to obtain the infrared radiation intensity component and polarization degree component.

[0070] It should be noted that Stokes vector calculations were performed on the low-light polarization image group (four directions: 0°, 45°, 90°, and 135°). By statistically analyzing the linear combination relationship of light intensity in each polarization direction, the low-light intensity component representing the total light intensity, as well as the degree of polarization and polarization angle reflecting the polarization characteristics, were obtained. At the same time, thermal radiation polarization calculations were performed on the long-wave infrared polarization image group (horizontal and vertical directions). The infrared radiation intensity component was calculated using the intensity difference and sum of two orthogonal polarization components, and the infrared polarization degree was derived by combining Malus's law, thereby separating the radiation and polarization characteristics information of the target.

[0071] Based on the spatial gradient characteristics of the low-light intensity component and the thermal radiation profile characteristics of the infrared radiation intensity component, a spatial registration matrix is ​​generated through affine transformation to achieve sub-pixel-level alignment of the two band images; the polarization correlation between the bands is determined by the low-light polarization characteristic component and the infrared polarization degree component, and a fusion weight matrix is ​​generated based on the polarization correlation.

[0072] It should be noted that by extracting the edge gradient features of the low-light intensity components (such as road markings and vehicle outlines) and the thermal distribution features of infrared radiation intensity (such as pedestrian and vehicle heat sources), affine transformation parameters are calculated through feature point matching to generate a spatial registration matrix, aligning the low-light and infrared images to sub-pixel accuracy. Simultaneously, the spatial distribution consistency of the low-light polarization angle and infrared polarization degree is analyzed, and their covariance matrix is ​​calculated to determine the polarization correlation between bands (such as the correlation between road surface polarization reflection and thermal radiation polarization). Based on this, an adaptive fusion weight matrix is ​​generated, ensuring that the fusion process retains the effective information of each band.

[0073] The infrared radiation intensity component is geometrically corrected using a spatial registration matrix to obtain the corrected infrared radiation intensity; the polarization direction of the infrared polarization component is compensated to eliminate the polarization angle shift caused by parallax, and the corrected infrared polarization is output.

[0074] It should be noted that a spatial registration matrix is ​​used to perform geometric transformations (including translation, rotation, and scaling) on ​​the infrared radiation intensity components to ensure strict alignment between the infrared image and the low-light image, outputting position-corrected infrared radiation intensity data. Simultaneously, based on the registered coordinate offset, the polarization direction angle of the infrared polarization component is compensated and calibrated to eliminate polarization angle deviations caused by dual-camera parallax (such as phase correction of horizontal / vertical polarization directions), ultimately outputting infrared polarization information with synchronized geometric and polarization direction correction.

[0075] The corrected infrared radiation intensity and the micro-light intensity component are fused according to the fusion weight matrix to generate fused light intensity features; the corrected infrared polarization degree and the micro-light polarization characteristic component are recombined in polarization state to generate fused polarization features.

[0076] By superimposing the fused light intensity characteristics, the corrected infrared radiation intensity, and the fused polarization characteristics along the spatial dimension, a six-channel dual-band polarization feature cube containing light intensity, radiation, and polarization information is constructed.

[0077] In a specific embodiment of the present invention, in a nighttime monitoring scenario of dense fog on a highway, a split-focus plane polarization low-light camera (response band 400-700nm) is used to simultaneously acquire low-light images in four polarization directions: 0°, 45°, 90°, and 135°. Simultaneously, a long-wave infrared polarization camera (response band 8-14μm) is used to acquire radiation images in the horizontal and vertical polarization directions. The linear polarization degree and polarization angle of the low-light images are extracted using Stokes calculation, and the thermal radiation intensity and polarization degree are obtained by combining them with infrared polarization calculation. Affine transformation registration is performed based on road guardrail edge feature points (spatial gradient threshold >15dB) and vehicle thermal radiation profiles (signal-to-noise ratio ≥8dB) (registration error <0.5 pixels), and fusion weights are generated using polarization correlation. Finally, a six-channel feature cube (data dimension 1280×1024×6) containing low-light intensity, infrared radiation, and polarization phase / degree is constructed, effectively overcoming the detail loss problem of traditional single-band imaging in dense fog (visibility <50m).

[0078] It should be noted that this invention achieves high-precision registration and polarization information fusion of dual-band images, effectively improving the spatial alignment accuracy and polarization information integrity of the images.

[0079] Preferably, such as Figure 2 As shown, the 104 specifically refers to:

[0080] S202: Perform pixel-by-pixel differential operation on the micro-light intensity component and infrared radiation intensity component in the dual-band polarization feature cube to obtain the dual-band differential intensity parameter. Calculate the polarization difference ratio using the Stokes parameter of the micro-light polarization characteristic component to obtain the polarization modulation difference parameter.

[0081] It should be noted that Stokes parameters are extracted from the micro-light polarization characteristic components, and the polarization intensity difference ratio in orthogonal directions (such as 0° and 90°) is calculated (i.e., the intensity of light in the 0° polarization direction is subtracted from the intensity of light in the 90° polarization direction, and then divided by the sum of the intensity of light in these two directions). This yields the difference parameters that reflect the polarization modulation characteristics of the target surface. At the same time, the phase relationship of different polarization directions is combined to eliminate environmental stray light interference, and finally, the modulation difference parameters that characterize the polarization characteristics of the scene material are output.

[0082] S204: Generate an initial fog concentration distribution map based on the polarization modulation difference parameter and the dual-band differential intensity parameter; perform anisotropic diffusion filtering on the initial fog concentration distribution map, combine the thermal radiation polarization constraint condition of the infrared polarization degree component to suppress noise and retain the fog edge structure, and output the optimized fog concentration gradient tensor; wherein, the thermal radiation polarization constraint condition refers to the process of using the polarization characteristics generated by the object's own thermal radiation in long-wave infrared polarization imaging (such as the differential polarization response of metal / non-metal surfaces) as a physical constraint to correct and optimize fog concentration detection.

[0083] It should be noted that the polarization modulation difference parameter (reflecting the scattering characteristics of light polarized in different directions) and the dual-band differential intensity parameter (the intensity difference between low-light and infrared images) are weighted and fused. The polarization parameter primarily characterizes the degree of modulation of polarized light by fog, while the dual-band differential parameter reflects the attenuation characteristics of light intensity by fog. Then, using a preset fog concentration calibration curve, the fused feature values ​​are converted into an initial fog concentration estimate for each pixel. Finally, this calculation is performed on all pixels in the entire image, generating a two-dimensional matrix reflecting the fog concentration distribution in different areas of the scene, where higher values ​​indicate greater fog concentration at that location.

[0084] S206: By utilizing the statistical correlation between the Stokes vector phase angle of the micro-light polarization characteristic component and the light intensity attenuation, a mapping relationship between polarization modulation and light intensity degradation is established. Through the mapping relationship, the degraded texture feature map modulated by Mie scattering is decoupled from the micro-light intensity component.

[0085] S208: Construct a three-dimensional distribution model of fog concentration field based on the optimized fog concentration gradient tensor, and determine the optical path accumulation of low-light degraded texture in fog concentration field according to the local contrast attenuation rate of degraded texture feature map.

[0086] It should be noted that the optimized 2D fog concentration gradient tensor (containing the concentration change rate in both horizontal and vertical directions) is extended by 3D spatial interpolation. Combined with scene depth information (such as distance data acquired through binocular vision or LiDAR), the fog concentration gradient of each pixel is converted into a concentration distribution vector in 3D space. Then, these vector fields are integrated using the Poisson reconstruction method to generate a continuous 3D fog concentration field model, where the X / Y axes correspond to the image plane coordinates, and the Z axis represents the fog concentration value (range 0-1). The final output is a 3D concentration field model that reflects the distribution and change of fog in 3D space.

[0087] It should be noted that a sliding window (e.g., 5×5 pixels) is used on the degraded texture feature map to calculate the local contrast attenuation rate (the contrast ratio between the original and degraded images) of each region. Simultaneously, the spatial concentration values ​​at corresponding locations in the 3D fog concentration field model are combined to establish the physical relationship between light intensity attenuation and fog concentration. Then, the cumulative effect of light passing through fog layers of different concentrations is calculated by integrating along the line of sight. Finally, the cumulative optical path length (a dimensionless parameter) corresponding to each pixel is output. This parameter directly reflects the overall scattering attenuation degree experienced by the micro-light texture during propagation.

[0088] S210: The mapping relationship between the gray-level attenuation gradient of the degraded texture feature map and the cumulative optical path length of the fog concentration field is established by nonlinear regression method, and the calibrated optical path difference mapping relationship matrix is ​​generated. Each element of the matrix represents the degree of scattering degradation of the micro-light texture under a specific fog concentration.

[0089] It should be noted that low-light image samples under different fog concentrations were collected, and the gray-level attenuation gradient (such as the local contrast reduction rate) of their degraded texture feature maps was extracted and combined with the optical path accumulation data at the corresponding locations to form a training set. Then, nonlinear fitting methods such as Gaussian process regression were used to establish a mapping function from optical path accumulation to gray-level attenuation gradient. Finally, this function was applied to the entire scene to generate an optical path difference mapping value for each pixel location that reflects the degree of scattering attenuation under a specific fog concentration, forming a calibration matrix of the same size as the original image, where each element value represents the light intensity attenuation correction coefficient caused by fog at that location.

[0090] In a specific embodiment of the present invention, in a highway fog monitoring scenario, the registered low-light intensity component and infrared radiation intensity are differentially analyzed pixel by pixel to obtain a dual-band differential intensity map (dynamic range 0-255). An initial fog concentration distribution map is generated by combining the polarization difference ratio calculated using low-light Stokes parameters (0°-90° direction ratio 1.2-1.8). An optimized fog gradient tensor is output through anisotropic diffusion filtering and infrared polarization constraint (threshold 0.15-0.3). The degradation texture dominated by Mie scattering is separated using the Pearson correlation between the Stokes phase angle (0-180°) and light intensity attenuation. Finally, an optical path difference mapping matrix (256×256 quantization levels) is established to accurately reflect the scattering attenuation law of lane markings (15cm width) at different fog concentrations (0.1-0.5g / m³).

[0091] It should be noted that this method can effectively distinguish between fog interference and real object texture in a scene, and preserve fog edge details while suppressing noise. The final constructed optical path difference mapping matrix can accurately reflect the scattering effect of different fog concentrations on low-light imaging, providing a reliable basis for degradation features for subsequent image restoration, thereby improving the accuracy of target recognition in foggy environments.

[0092] Preferably, a polarization modulation-light intensity degradation mapping relationship is established by utilizing the statistical correlation between the Stokes vector phase angle of the micro-light polarization characteristic component and the light intensity attenuation. This mapping relationship is then used to decouple the degraded texture feature map modulated by Mie scattering from the micro-light intensity component. Specifically:

[0093] The Stokes vectors of the micro-light polarization characteristic components in the dual-band polarization feature cube are extracted, and the phase angle distribution map of the Stokes vectors is obtained. At the same time, the local contrast attenuation rate matrix of the micro-light intensity components is obtained.

[0094] The phase angle distribution map is normalized to generate a polarization modulation phase feature map, and a polarization phase-attenuation statistical coupling relationship is established by combining the local contrast attenuation rate matrix and Pearson correlation analysis.

[0095] It should be noted that the Stokes vector phase angle distribution map is normalized, converting the phase angle values ​​to the 0-1 range to generate a standardized polarization modulation phase feature map. Then, this phase feature map is pixel-level matched with the local contrast attenuation rate matrix of the low-light image at the same spatial location, and the linear correlation between the two is calculated using the Pearson correlation coefficient. Finally, a quantitative statistical relationship between polarization phase and contrast attenuation is established based on the correlation coefficient, forming a mapping model reflecting the coupling strength between the two.

[0096] Based on the statistical coupling relationship between polarization phase and attenuation, the mapping curve between the polarization modulation phase feature map and the local contrast attenuation rate matrix is ​​fitted using the nonlinear least squares method to obtain the coupling coefficient matrix between polarization phase and attenuation.

[0097] It should be noted that, based on the established polarization phase-attenuation statistical coupling relationship, a nonlinear least squares method is used to fit the polarization modulation phase eigenvalues ​​and the corresponding local contrast attenuation rates. Through iterative optimization, the prediction error is minimized, and the characteristic parameters of the fitted curve (such as polynomial coefficients or exponential term weights) are extracted to construct a coupling coefficient matrix reflecting the nonlinear relationship between polarization phase and light intensity attenuation. Finally, this matrix is ​​multiplied by the original phase feature map to generate a polarization phase-attenuation coupling coefficient matrix that can accurately quantify the scattering effect.

[0098] The polarization phase-attenuation coupling coefficient matrix is ​​applied to the micro-light intensity component, and the global attenuation substrate layer dominated by Mie scattering is separated by inverse mapping operation to obtain the residual texture component.

[0099] It should be noted that the coupling coefficient matrix is ​​multiplied pixel-by-pixel with the low-light intensity component to obtain the predicted scattering attenuation distribution map. Then, the predicted attenuation value is subtracted from the original light intensity through an inverse operation to separate the global attenuation base layer mainly caused by Mie scattering. Finally, this base layer is subtracted from the original image to obtain the residual texture component containing details of the real scene. The residual part retains the object features that are not affected by scattering and reduces the contrast degradation effect caused by fog.

[0100] Anisotropic guided filtering is applied to the residual texture components, and the polarization degree component of the Stokes vector is used as the edge constraint weight to suppress noise and preserve the topology of the scattering degradation texture, outputting a degradation texture feature map modulated by Mie scattering.

[0101] By performing a differential operation between the degraded texture feature map and the global attenuation substrate, the linear independence between the light intensity attenuation gradient and the polarization modulation phase angle is verified, and finally the scattering modulation domain of the degraded texture feature map is calibrated.

[0102] It should be noted that this method effectively eliminates the confusion between environmental noise and actual attenuation, utilizes the physical correlation between polarization characteristics and light attenuation to extract pure scattering degradation features, and establishes a quantifiable scattering modulation domain for subsequent image restoration, thereby improving the reliability of low-light image feature extraction in dense fog environments.

[0103] Preferably, the 106 specifically refers to:

[0104] Based on the cumulative optical path of each pixel in the optical path difference mapping matrix, the polarization channel transmittance attenuation coefficient corresponding to the micro-light degradation texture feature map is determined, and a transmittance attenuation coefficient distribution map is generated.

[0105] It should be noted that the optical path accumulation value of each pixel in the optical path difference mapping matrix is ​​read, and based on a pre-calibrated optical path-transmittance conversion curve (such as the exponential decay model of Lambert-Beer's law), the optical path accumulation value of each pixel is converted into the corresponding polarization channel transmittance attenuation coefficient. Then, the attenuation coefficients of all pixels are arranged according to their spatial positions in the image to generate a transmittance attenuation coefficient distribution map with the same size as the original image, where each pixel value represents the degree of light intensity attenuation caused by fog scattering at that location.

[0106] By combining the polarization phase angle distribution of the Stokes vector, a nonlinear coupling relationship between the transmittance attenuation coefficient and the polarization phase angle is established, and a polarization-transmittance coupling tensor is generated.

[0107] It should be noted that the transmittance attenuation coefficient distribution map is spatially aligned with the Stokes vector polarization phase angle distribution map to ensure a one-to-one correspondence between the attenuation coefficient and phase angle for each pixel. A quantitative relationship model between the two is established through nonlinear regression analysis (such as polynomial fitting or neural network modeling). Finally, the established phase-attenuation relationship is parameterized to generate a three-dimensional tensor structure containing the coupling coefficients of each pixel, which includes the nonlinear response characteristics of the attenuation coefficient as a function of polarization phase.

[0108] By utilizing the infrared radiation intensity component in the dual-band polarization feature cube, the penetration characteristic curve of thermal radiation in the fog medium is extracted. By calibrating the inverse proportional relationship between infrared thermal radiation intensity and fog concentration, a thermal radiation penetration compensation coefficient matrix is ​​constructed.

[0109] It should be noted that infrared radiation intensity components are extracted from the dual-band polarization feature cube, and the corresponding thermal radiation intensity values ​​for different fog concentration regions (e.g., low concentration region 0.1-0.3 g / m³, high concentration region 0.3-0.5 g / m³) are analyzed. Then, an inverse relationship model between infrared radiation intensity and fog concentration is established (radiation intensity = reference value / (1 + k × concentration)), where k is the medium characteristic coefficient. Finally, based on this model, the fog concentration value of each pixel is converted into the corresponding thermal radiation penetration compensation coefficient (range 0.4-1.0), generating a compensation coefficient matrix of the same size as the image for subsequent joint correction processing.

[0110] The polarization-transmittance coupling tensor and the thermal radiation penetration compensation coefficient matrix are multiplied by tensor dot product to obtain a joint compensation operator that integrates polarization modulation characteristics and thermal radiation penetration characteristics.

[0111] The joint compensation operator is applied to the low-light degradation texture feature map. By correcting the transmittance attenuation coefficient pixel by pixel, the polarization channel energy attenuation caused by Mie scattering is eliminated, and the intermediate low-light texture after transmittance correction is output.

[0112] It should be noted that the joint compensation operator is multiplied pixel-by-pixel with the low-light degradation texture feature map. Using the transmittance correction coefficients stored in the compensation operator, the attenuation degree of each pixel due to Mie scattering is specifically compensated. Then, local contrast adjustment is performed on the compensated image to restore the details lost due to scattering. Finally, the intermediate low-light texture image after transmittance correction is output.

[0113] The polarization degree component of the Stokes vector is used to enhance the edges of the intermediate micro-light texture. The local contrast lost due to scattering is restored by polarization degree weighting, and an enhanced micro-light feature map with fog scattering invariance is generated.

[0114] In a specific embodiment of the present invention, for the calibrated optical path difference mapping matrix (256×256 quantization level), the transmittance attenuation coefficient (range 0.3-0.9) is determined based on the cumulative optical path of each pixel (0.1-1.2 optical path units), generating an attenuation coefficient distribution map. A polarization-transmittance coupling tensor is established by combining the Stokes vector phase angle (0-180°), and a thermal radiation compensation matrix (transmittance 0.4-0.8) is constructed using infrared radiation intensity (8-14μm band). The two are multiplied by tensor to generate a joint compensation operator, which performs pixel-by-pixel correction on the 1280×1024 pixel low-light degradation texture, thereby improving the contrast of key features such as lane markings. Finally, edge enhancement is performed using the polarization degree component, outputting an enhanced image with scattering invariance to solve the detail loss problem in dense fog areas (fog concentration > 0.3 g / m³) using traditional methods.

[0115] It should be noted that this method can improve the texture clarity and feature fidelity of low-light images in dense fog environments, and make the enhanced image have stable scattering invariant properties.

[0116] Preferably, the 108 specifically refers to:

[0117] The Laplacian operator convolution operation is performed on the fog concentration gradient tensor to calculate the second spatial derivative at each pixel location, generating a Laplacian response map of fog concentration.

[0118] It should be noted that a 3×3 Laplacian kernel is used to perform a two-dimensional convolution operation on the fog concentration gradient tensor, calculating the second-order spatial derivative of each pixel within its eight-neighborhood. The convolution result is normalized to generate a Laplacian response map reflecting the abrupt changes in fog concentration. Positive values ​​represent fog edges where concentration rises rapidly, negative values ​​correspond to transition zones where concentration decreases, and zero values ​​represent areas with uniform concentration distribution. Finally, significant extrema (Laplacian response value ≥ +0.15) in the response map are extracted to provide boundary feature basis for subsequent concentration partitioning.

[0119] Based on the local extreme value distribution characteristics of the Laplace response diagram, the abrupt boundary of fog concentration change is extracted to form the concentration partition boundary;

[0120] The concentration partition boundaries are smoothed by morphological closing operations to eliminate holes caused by noise. Combined with the spatial consistency constraints of infrared polarization components, the mis-segmented regions caused by thermal radiation interference are corrected to generate a dynamic partition mask.

[0121] Based on the high and low concentration regions marked in the dynamic partition mask, the pixels in the monitored scene are classified into low-concentration areas and high-concentration areas; where the low-concentration area is the area with a mask value of 0, and the high-concentration area is the area with a mask value of 1.

[0122] In a specific embodiment of the present invention, in a highway fog monitoring scenario (e.g., visibility classification: low concentration area >100m, high concentration area <50m), a Laplacian convolution (3×3 kernel) is performed on the optimized fog concentration gradient tensor (resolution 1280×1024) to generate a response map (extreme threshold ±0.15). Local maxima / minimum values ​​(interval >30 pixels) are detected to extract concentration abrupt change boundaries. After correction by morphological closing operations (circular structuring element radius 5 pixels) and infrared polarization constraints (threshold 0.25-0.35), a dynamic partitioning mask is generated. Finally, the scene is divided into a low concentration area (mask value 0, fog concentration <0.2g / m³) and a high concentration area (mask value 1, concentration ≥0.2g / m³), achieving accurate segmentation of lanes (low concentration) and fog patches (high concentration).

[0123] Preferably, the 110 specifically refers to:

[0124] In the low-concentration region, the fog concentration gradient tensor of the enhanced low-light feature map is extracted, and the thermal radiation contour gradient field is parsed from the infrared thermal radiation intensity. The fog concentration gradient tensor and the thermal radiation contour gradient field are measured pixel by pixel gradient structure similarity to generate a gradient fidelity weight map.

[0125] Based on the gradient fidelity weight map, the low-frequency component (0-0.5πrad / m spatial frequency) of the enhanced low-light feature map and the high-frequency component (0.5π-2πrad / m) of the infrared thermal radiation intensity are adaptively weighted and fused to obtain the primary fused feature; wherein, the weight range is 0.3 (emphasizing infrared details) to 0.7 (emphasizing the low-light substrate).

[0126] The gradient field is rebalanced on the primary fusion features by utilizing the local anisotropy coefficient of the fog concentration gradient tensor to eliminate radiation distortion at the fusion boundary and output the fusion result in the low concentration region.

[0127] In the high-concentration region, the spatial distribution matrix of the infrared polarization vector direction angle is calculated, and the polarization phase angle feature in the Stokes component of the enhanced low-light feature map is extracted. The spatial distribution matrix and the polarization phase angle feature are subjected to polarization state covariance analysis to establish a polarization direction-phase coupling tensor. The Stokes component of the enhanced low-light feature map refers to the set of feature parameters including polarization intensity, linear polarization degree and polarization angle obtained after Stokes vector calculation and scattering correction of the original low-light polarization image. These components together characterize the polarization characteristic distribution of the target scene after fog scattering compensation.

[0128] The infrared polarization direction angle is transformed by Stokes space projection using the coupling tensor to generate a recombinant polarization basis.

[0129] It should be noted that the infrared polarization direction angle data is input into a coupling tensor (which stores the mapping relationship between the infrared and micro-light polarization states), and the infrared polarization direction angle is projected onto the Stokes vector space (three-dimensional coordinates) through tensor multiplication. Then, in the Stokes space, the linear polarization components are re-determined based on the projection results, generating a reconstructed polarization basis (containing new polarization angle and degree of polarization parameters) compatible with dual-band polarization characteristics. The final output is reconstructed polarization basis data that can simultaneously express the fusion characteristics of the infrared polarization direction and the micro-light polarization phase.

[0130] By dynamically adjusting the weight ratio of the reconstituted polarization basis to the enhanced low-light feature map using the Laplace norm response value of the fog concentration gradient tensor, tensor product operation of the polarization channels is performed to obtain the polarization recombination features in the high-concentration region.

[0131] It should be noted that the Laplace norm response value of the fog concentration gradient tensor (reflecting the intensity of concentration abrupt changes) is obtained and normalized to a dynamic weighting coefficient of 0-1 (the larger the value, the denser the fog). This coefficient is used as the fusion weight of the reconstructed polarization basis (weight range 0.2-0.8), while the weight of the enhanced low-light feature map is set to a complementary value (1-weighting coefficient). Finally, channel-level tensor product operation (i.e., weighted product fusion of each polarization channel) is performed on the reconstructed polarization basis and the low-light feature map to generate high-concentration region polarization reconstruction features that simultaneously retain infrared polarization direction information and low-light texture details. Among them, the dense fog core region (weight > 0.6) focuses on polarization penetration, while the edge transition region (weight < 0.4) enhances low-light details.

[0132] Finally, the fusion results of the low-concentration area and the polarization recombination features of the high-concentration area are spatially stitched together using a dynamic partitioning mask, and the final night vision image is generated after a smooth transition by bilinear interpolation.

[0133] Among them, the "infrared thermal radiation intensity" used in the low-concentration region fusion is the corresponding corrected infrared radiation intensity (the polarization information has been stripped, and the thermal radiation scalar value is retained); the "infrared polarization vector direction angle" used in the high-concentration region is the vector information calculated from the original infrared polarization data (reconstructed with Stokes components).

[0134] In a specific embodiment of the present invention, in a highway fog night monitoring scenario, for low-concentration areas (fog concentration <0.2g / m³) marked by dynamic partitioning masking, an adaptive weight map is generated by calculating the structural similarity between the gradient tensor of the low-light feature map (gradient amplitude 15-30dB) and the infrared thermal radiation gradient field (temperature resolution 0.05K), thereby achieving a faithful fusion of lane markings (low-frequency low-light) and vehicle heat sources (high-frequency infrared). For high-concentration areas (≥0.2g / m³), a coupling tensor is constructed using the infrared polarization direction angle and the low-light Stokes phase angle (0-180°), and the polarization basis is dynamically weighted and recombined using the Laplace norm (response value 0.1-0.3) to enhance obstacle recognition in the fog. Finally, after bilinear interpolation and stitching, a fused image of 1280×1024 pixels is output.

[0135] It should be noted that, through a partitioned fusion strategy, the advantages of low-light and infrared features are complemented by gradient structure similarity measurement in low-concentration areas, preserving detailed textures and thermal radiation contours; in high-concentration areas, polarization state recombination is used to fully exploit the penetrating power of polarization information on scattering, improving the imaging quality in dense fog areas; finally, through dynamic weighted fusion and smooth stitching, a night view image with high definition, complete thermal radiation features and polarization enhancement effect is generated, effectively solving the fusion distortion problem of traditional methods in areas with abrupt changes in fog concentration.

[0136] In actual operation, the night vision imaging method further includes the following steps:

[0137] An orthogonal triaxial fluxgate sensor is deployed at the imaging equipment installation point. Based on the low-light image acquisition frame rate, the geomagnetic field vector intensity and deflection time series data are captured simultaneously to generate a dynamic geomagnetic disturbance baseline.

[0138] It should be noted that an orthogonal triaxial fluxgate sensor is fixedly installed next to the imaging equipment, strictly synchronized with the frame rate (e.g., 30fps) of the low-light camera, to record the X / Y / Z triaxial geomagnetic field strength and declination data in real time. The geomagnetic data is matched with the image frame sequence through timestamp alignment, and transient noise is eliminated by sliding window averaging (1-second window width), generating a dynamic geomagnetic disturbance baseline dataset with a time resolution of 33ms, including magnetic field vector intensity curves and temporal characteristics of declination fluctuations.

[0139] The real-time phase angle matrix of the Stokes vector in the dual-band polarization feature cube is extracted and spatiotemporally registered with the deflection component of the dynamic geomagnetic disturbance baseline. A phase-deflection coupling response function is established through a long short-term memory network.

[0140] By applying the phase-offset coupling response function to the Stokes phase angle matrix, the polarization state drift component caused by electromagnetic interference is calculated, and a polarization drift vector field is constructed.

[0141] It should be noted that the pre-trained phase-offset coupled response function (LSTM network model) is used to calculate the Stokes phase angle matrix frame by frame to predict the intrinsic phase angle of each pixel in the absence of magnetic field interference. Then, the difference between the predicted value and the actual measured phase angle is used to obtain the polarization state drift caused by electromagnetic interference. The drift values ​​of all pixels are arranged according to their spatial positions in the image to construct a two-dimensional polarization drift vector field (containing both magnitude and direction information).

[0142] The intrinsic polarization characteristics of the fog medium are extracted from the spatial distribution of the infrared polarization degree components. Combined with the spectral characteristics of the dynamic geomagnetic disturbance baseline, the polarization intrinsic axis reference system is separated by Wiener filtering.

[0143] Using the polarization intrinsic axis reference frame as a reference, the polarization drift vector field is decomposed by antisymmetric tensor, and the non-conservative component strongly correlated with the geomagnetic declination is extracted to output the geomagnetic drift compensation field.

[0144] The magnetotropic drift compensation field is subtracted from the original Stokes vector, and the orthogonal projection normalization is performed using the polarization eigenaxis reference frame to reconstruct the diamagnetic Stokes vector to eliminate electromagnetic interference.

[0145] For example, in foggy night monitoring scenarios near high-voltage power lines on highways, a three-axis fluxgate sensor and a low-light camera are used to synchronously acquire data and record real-time fluctuations in the geomagnetic field declination. A pre-trained LSTM network is used to analyze the correlation between the Stokes phase angle (1280×1024 pixels) and the magnetic declination to calculate the polarization drift field caused by electromagnetic interference. The intrinsic polarization axis of the fog is extracted by combining infrared polarization, and after Wiener filtering and antisymmetric tensor decomposition, a geomagnetic compensation field is generated. Finally, an antimagnetic declination Stokes vector is output, improving the reliability of polarization imaging under electromagnetic interference environments.

[0146] It should be noted that this embodiment solves the problem of polarization imaging drift caused by geomagnetic disturbance in complex electromagnetic environments (such as near high-voltage lines on highways) by monitoring geomagnetic disturbance and modeling polarization state drift. This allows the imaging results to maintain the physical authenticity of polarization information even under strong electromagnetic interference, thereby improving the reliability and environmental adaptability of target polarization feature identification in foggy night road monitoring.

[0147] like Figure 3 As shown, the second aspect of the present invention discloses a night vision imaging system 8 that fuses low-light and infrared images. The night vision imaging system includes a memory 60 and a processor 80. The memory 60 stores a night vision imaging method program that fuses low-light and infrared images. When the night vision imaging method program that fuses low-light and infrared images is executed by the processor 80, any one of the steps of the night vision imaging method that fuses low-light and infrared images is implemented.

[0148] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0149] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0150] In addition, in the various embodiments of the present invention, each functional unit can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0151] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0152] Alternatively, if the integrated units of this invention are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.

[0153] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A night vision imaging method that fuses low-light and infrared images, characterized in that, Includes the following steps: S102: In foggy night road environments, simultaneously acquire low-light polarization images and long-wave infrared polarization images to construct a spatiotemporally aligned dual-band polarization feature cube; S104: Based on the dual-band polarization feature cube, the fog concentration gradient tensor of the monitoring scene is extracted and the degraded texture feature map modulated by Mie scattering is separated. The optical path difference mapping relationship between the low-light degraded texture and the fog concentration field is established, and the optical path difference mapping relationship matrix is ​​generated. S106: Generate a joint compensation operator based on the optical path difference mapping relationship matrix, and combine the compensation operator to perform polarization channel transmittance correction on the micro-light degradation texture, and output an enhanced micro-light feature map with fog scattering invariance. S108: Analyze the fog concentration gradient tensor and dynamically divide the low-concentration region and high-concentration region based on the Laplace norm response value of the fog concentration gradient tensor. S110: In the low-concentration region, the enhanced low-light feature map and the infrared thermal radiation intensity are fused using gradient tensor fidelity; in the high-concentration region, the polarization state tensor is recombined with the Stokes component of the enhanced low-light feature map to generate a night vision imaging map.

2. The night vision imaging method for fusing low-light and infrared images according to claim 1, characterized in that, Specifically, 102 refers to: Simultaneously collect light intensity data in four polarization directions of the low-light band and radiation intensity data in two orthogonal polarization directions of the long-wave infrared band, and generate low-light polarization intensity image group and long-wave infrared polarization radiation intensity image group respectively. Stokes vector calculation was performed on the low-light polarization image group to obtain the low-light intensity component and polarization characteristic component; thermal radiation polarization calculation was performed on the long-wave infrared polarization image group to obtain the infrared radiation intensity component and polarization degree component. Based on the spatial gradient characteristics of the low-light intensity component and the thermal radiation profile characteristics of the infrared radiation intensity component, a spatial registration matrix is ​​generated through affine transformation to achieve sub-pixel-level alignment of the two band images; the polarization correlation between the bands is determined by the low-light polarization characteristic component and the infrared polarization degree component, and a fusion weight matrix is ​​generated based on the polarization correlation. The infrared radiation intensity components are geometrically corrected using a spatial registration matrix to obtain the corrected infrared radiation intensity. The polarization direction of the infrared polarization degree component is compensated to eliminate the polarization angle shift caused by parallax, and the corrected infrared polarization degree is output. The corrected infrared radiation intensity and the micro-light intensity component are fused according to the fusion weight matrix to generate fused light intensity features; the corrected infrared polarization degree and the micro-light polarization characteristic component are recombined in polarization state to generate fused polarization features. By superimposing the fused light intensity characteristics, the corrected infrared radiation intensity, and the fused polarization characteristics along the spatial dimension, a six-channel dual-band polarization feature cube containing light intensity, radiation, and polarization information is constructed.

3. The night vision imaging method for fusing low-light and infrared images according to claim 1, characterized in that, Specifically, 104 refers to: A pixel-by-pixel differential operation is performed on the micro-light intensity component and the infrared radiation intensity component in the dual-band polarization feature cube to obtain the dual-band differential intensity parameter. The polarization difference ratio is calculated using the Stokes parameter of the micro-light polarization characteristic component to obtain the polarization modulation difference parameter. An initial fog concentration distribution map is generated based on the polarization modulation difference parameter and the dual-band differential intensity parameter; anisotropic diffusion filtering is applied to the initial fog concentration distribution map, and combined with the thermal radiation polarization constraint condition of the infrared polarization degree component, noise is suppressed and the fog edge structure is preserved, and the optimized fog concentration gradient tensor is output. By utilizing the statistical correlation between the Stokes vector phase angle of the micro-light polarization characteristic component and the light intensity attenuation, a mapping relationship between polarization modulation and light intensity degradation is established. Through this mapping relationship, the degraded texture feature map modulated by Mie scattering is decoupled from the micro-light intensity component. A three-dimensional distribution model of fog concentration field is constructed based on the optimized fog concentration gradient tensor. At the same time, the optical path accumulation of the low-light degraded texture in the fog concentration field is determined according to the local contrast attenuation rate of the degraded texture feature map. A mapping relationship between the gray-level attenuation gradient of the degraded texture feature map and the cumulative optical path length of the fog concentration field is established by nonlinear regression method, and a calibrated optical path difference mapping relationship matrix is ​​generated. Each element of the matrix represents the degree of scattering degradation of the micro-light texture under a specific fog concentration.

4. The night vision imaging method for fusing low-light and infrared images according to claim 3, characterized in that, By utilizing the statistical correlation between the Stokes vector phase angle of the micro-light polarization component and the light intensity attenuation, a mapping relationship between polarization modulation and light intensity degradation is established. Through this mapping relationship, degraded texture feature maps modulated by Mie scattering are decoupled from the micro-light intensity component. Specifically: The Stokes vectors of the micro-light polarization characteristic components in the dual-band polarization feature cube are extracted, and the phase angle distribution map of the Stokes vectors is obtained. At the same time, the local contrast attenuation rate matrix of the micro-light intensity components is obtained. The phase angle distribution map is normalized to generate a polarization modulation phase feature map, and a polarization phase-attenuation statistical coupling relationship is established by combining the local contrast attenuation rate matrix and Pearson correlation analysis. Based on the statistical coupling relationship between polarization phase and attenuation, the mapping curve between the polarization modulation phase feature map and the local contrast attenuation rate matrix is ​​fitted using the nonlinear least squares method to obtain the coupling coefficient matrix between polarization phase and attenuation. The polarization phase-attenuation coupling coefficient matrix is ​​applied to the micro-light intensity component, and the global attenuation substrate layer dominated by Mie scattering is separated by inverse mapping operation to obtain the residual texture component. Anisotropic guided filtering is applied to the residual texture components, and the polarization degree component of the Stokes vector is used as the edge constraint weight to suppress noise and preserve the topology of the scattering degradation texture, outputting a degradation texture feature map modulated by Mie scattering. By performing a differential operation between the degraded texture feature map and the global attenuation substrate, the linear independence between the light intensity attenuation gradient and the polarization modulation phase angle is verified, and finally the scattering modulation domain of the degraded texture feature map is calibrated.

5. The night vision imaging method for fusing low-light and infrared images according to claim 1, characterized in that, Specifically, 106 refers to: Based on the cumulative optical path of each pixel in the optical path difference mapping matrix, the polarization channel transmittance attenuation coefficient corresponding to the micro-light degradation texture feature map is determined, and a transmittance attenuation coefficient distribution map is generated. By combining the polarization phase angle distribution of the Stokes vector, a nonlinear coupling relationship between the transmittance attenuation coefficient and the polarization phase angle is established, and a polarization-transmittance coupling tensor is generated. By utilizing the infrared radiation intensity component in the dual-band polarization feature cube, the penetration characteristic curve of thermal radiation in the fog medium is extracted. By calibrating the inverse proportional relationship between infrared thermal radiation intensity and fog concentration, a thermal radiation penetration compensation coefficient matrix is ​​constructed. The polarization-transmittance coupling tensor and the thermal radiation penetration compensation coefficient matrix are multiplied by tensor dot product to obtain a joint compensation operator that integrates polarization modulation characteristics and thermal radiation penetration characteristics. The joint compensation operator is applied to the low-light degradation texture feature map. By correcting the transmittance attenuation coefficient pixel by pixel, the polarization channel energy attenuation caused by Mie scattering is eliminated, and the intermediate low-light texture after transmittance correction is output. The polarization degree component of the Stokes vector is used to enhance the edges of the intermediate micro-light texture. The local contrast lost due to scattering is recovered by polarization degree weighting, and an enhanced micro-light feature map with fog scattering invariance is generated.

6. The night vision imaging method for fusing low-light and infrared images according to claim 1, characterized in that, The 108 specifically refers to: The Laplacian operator convolution operation is performed on the fog concentration gradient tensor to calculate the second spatial derivative at each pixel location, generating a Laplacian response map of fog concentration. Based on the local extreme value distribution characteristics of the Laplace response diagram, the abrupt boundary of fog concentration change is extracted to form the concentration partition boundary; The concentration partition boundaries are smoothed by morphological closing operations to eliminate holes caused by noise. Combined with the spatial consistency constraints of infrared polarization components, the mis-segmented regions caused by thermal radiation interference are corrected to generate a dynamic partition mask. Based on the high and low concentration regions marked in the dynamic partition mask, the pixels in the monitored scene are classified into low-concentration areas and high-concentration areas; where the low-concentration area is the area with a mask value of 0, and the high-concentration area is the area with a mask value of 1.

7. The night vision imaging method for fusing low-light and infrared images according to claim 1, characterized in that, The 110 specifically refers to: In the low-concentration region, the fog concentration gradient tensor of the enhanced low-light feature map is extracted, and the thermal radiation contour gradient field is parsed from the infrared thermal radiation intensity. The fog concentration gradient tensor and the thermal radiation contour gradient field are measured pixel by pixel gradient structure similarity to generate a gradient fidelity weight map. Based on the gradient fidelity weight map, the low-frequency component of the enhanced low-light feature map and the high-frequency component of the infrared thermal radiation intensity are adaptively weighted and fused to obtain the primary fused feature. The gradient field is rebalanced on the primary fusion features by utilizing the local anisotropy coefficient of the fog concentration gradient tensor to eliminate radiation distortion at the fusion boundary and output the fusion result in the low concentration region.

8. A night vision imaging method for fusing low-light and infrared images according to claim 7, characterized in that, The 110 also includes: In the high-concentration region, the spatial distribution matrix of the infrared polarization vector direction angle is calculated, and the polarization phase angle feature in the Stokes component of the enhanced micro-light feature map is extracted. The spatial distribution matrix and the polarization phase angle feature are subjected to polarization state covariance analysis to establish a polarization direction-phase coupling tensor. The infrared polarization direction angle is transformed by Stokes space projection using the coupling tensor to generate a recombinant polarization basis. By dynamically adjusting the weight ratio of the reconstituted polarization basis to the enhanced low-light feature map using the Laplace norm response value of the fog concentration gradient tensor, tensor product operation of the polarization channels is performed to obtain the polarization recombination features in the high-concentration region. Finally, the fusion results of the low-concentration area and the polarization recombination features of the high-concentration area are spatially stitched together using a dynamic partitioning mask, and the final night vision image is generated after a smooth transition by bilinear interpolation.

9. A night vision imaging system that fuses low-light and infrared images, characterized in that, The night vision imaging system includes a memory and a processor. The memory stores a night vision imaging method program for fusing low-light and infrared images. When the processor executes the night vision imaging method program for fusing low-light and infrared images, it implements the steps of the night vision imaging method for fusing low-light and infrared images as described in any one of claims 1 to 8.

Images