Polarization decoding method for multi-view field aliasing infrared image based on stokes vector reconstruction

Abstract

This invention relates to a polarization decoding method for multi-field-of-view aliased infrared images based on Stokes vector reconstruction, belonging to the field of space optics technology. It addresses the problems of aliased image recognition and reconstruction schemes generated by detector multiplexing imaging relying on scalar intensity and local contrast features, distortion bias leading to decoding algorithm failure, and poor accuracy and robustness. The method includes the following steps: acquisition of multi-channel polarization measurement data; Stokes vector reconstruction; construction of a polarization reference template; pixel-level classification based on Stokes space; foreground target extraction and field-of-view determination; target reconstruction and output. This invention's polarization decoding method for multi-field-of-view aliased infrared images based on Stokes vector reconstruction introduces polarization information as an orthogonal physical dimension independent of intensity information, extending the traditional detection method that relies solely on scalar intensity features to a discrimination mechanism based on vector information, effectively solving the problem of insufficient recognition capability of traditional intensity methods in aliased scenes.

Description

Technical Field

[0001] This invention relates to the field of space optics technology, and in particular to a polarization decoding method for multi-field aliased infrared images based on Stokes vector reconstruction. Background Technology

[0002] Wide field-of-view (FOV) infrared imaging systems have important applications in space target observation and long-range, weak target detection. However, such systems are often severely limited by detector size, cooling requirements, and payload capacity. To achieve large-scale observation without increasing hardware resources, detector multiplexing has become a research hotspot, which maps and superimposes signals from multiple fields of view (FOV) onto a single detector using optical methods. However, during detector multiplexing, the sub-images from different FOVs undergo severe spatial aliasing after geometric folding, making it impossible for traditional detection methods to distinguish the target's source FOV, thus significantly reducing target localization accuracy and detection reliability. Therefore, accurately decoding the original spatial information from the aliased detection data is a key challenge in computational infrared imaging.

[0003] Currently, existing approximate solutions for the recognition and reconstruction of aliased images generated by detector multiplexing imaging mainly fall into two categories: First, traditional intensity-based detection algorithms, such as Top-hat transform filtering, Max-Median filtering, and detection techniques based on weighted local difference measurements. These methods primarily rely on scalar intensity features such as target brightness and local contrast for small target extraction. However, in scenarios with deep aliasing of multi-field signals, targets from different sources may have similar intensity and motion characteristics, and background superposition leads to a significant decrease in contrast, limiting the detection performance of these algorithms. Second, field-of-view shape modulation coding methods. This method designs the spatially varied point spread function (PSF) of the optical system to actively introduce specific geometric distortions (such as different shaped speckles) in different fields of view to mark spatial location information, and then uses shape-aware algorithms for decoding and recognition. However, spatial PSF modulation is essentially still a modulation of light intensity distribution, with limited information channel capacity and high sensitivity to manufacturing tolerances of the optical system, system alignment errors, and overlapping interference from the point spread function.

[0004] In summary, current infrared detection technologies suffer from the following shortcomings when dealing with image aliasing caused by detector multiplexing: First, traditional intensity-based detection algorithms (such as Top-hat or Max-Median filtering) rely primarily on scalar intensity and local contrast features. In scenarios where multiple field-of-view signals overlap significantly, the superposition of background noise severely reduces signal contrast. Furthermore, targets from different spatial regions often exhibit similar radiation characteristics in the intensity domain, making it difficult for the algorithm to distinguish the true origin of the target, resulting in low detection rates. Second, while existing field-of-view shape modulation coding methods introduce morphological features, they are essentially still modulations of the spatial distribution of light intensity. Their information channel capacity is limited, and they are highly sensitive to manufacturing tolerances, assembly alignment errors, and overlapping interference from the point spread function (PSF). Even minor physical distortion deviations can cause the decoding algorithm to fail. In addition, existing technologies exhibit poor accuracy and robustness in the presence of measurement noise or non-ideal polarization modulation states, making it difficult to meet the high-reliability detection requirements of compact wide-field-of-view systems in complex environments. Summary of the Invention

[0005] This invention aims to address the technical problems in existing technologies where the recognition and restoration schemes for aliased images generated by detector multiplexing imaging rely on scalar intensity and local contrast features, and small physical distortion deviations can lead to decoding algorithm failure, resulting in poor accuracy and robustness. The invention provides a polarization decoding method for multi-field aliased infrared images based on Stokes vector reconstruction.

[0006] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:

[0007] A polarization decoding method for multi-field aliased infrared images based on Stokes vector reconstruction includes the following steps:

[0008] Step S1: Acquisition of multi-channel polarization measurement data;

[0009] Multi-channel polarization measurement is performed on infrared images that have been polarized and superimposed on the detector plane;

[0010] Step S2: Stokes vector reconstruction;

[0011] After obtaining multi-channel polarization measurement data, Stokes vector reconstruction is performed on each pixel position to convert the original intensity superposition signal into a vector representation with directional and structural information.

[0012] Step S3: Construction of polarization reference template;

[0013] Pre-construct polarization reference templates corresponding to each field of view region to establish a one-to-one correspondence between field of view labels and polarization features;

[0014] Step S4: Pixel-level classification based on Stokes space;

[0015] Based on the reconstructed Stokes vector of each pixel in step S2 and the polarization reference template constructed in step S3, each pixel in the aliased image is classified and discriminated based on the polarization vector.

[0016] Step S5: Foreground target extraction and field of view determination;

[0017] Foreground targets in the image are extracted, target regions with significant radiation characteristics are separated, and a statistical field of view determination is performed for each target region;

[0018] Step S6: Target Reconstruction and Output;

[0019] Based on the determined field of view information of the target, the target is mapped back to its original spatial location, realizing the spatial reconstruction and positioning of the target, and the result is output.

[0020] In the above technical solution, the multi-channel polarization measurement in step S1 specifically involves: by setting up polarization analyzers or equivalent polarization measurement devices with different orientations, the radiation intensity of the same pixel location under multiple polarization states is collected, thereby obtaining multi-channel observation data.

[0021] In the above technical solution, the multi-channel polarization measurement in step S1 includes at least six independent channels, corresponding to linear polarization components in the directions of 0°, 45°, 90° and 135°, as well as left-hand circular polarization and right-hand circular polarization components.

[0022] In the above technical solution, step S2 specifically involves: establishing a linear relationship model between polarization measurement intensity and Stokes parameters, performing inversion calculations on multi-channel observation data, and thus obtaining the Stokes vector of the corresponding pixel.

[0023] In the above technical solution, the inversion calculation process is achieved by using the least squares method or matrix inversion method in step S2.

[0024] In the above technical solution, step S3 specifically involves: according to the modulation parameters applied to different fields of view by the polarization coding module in the detector multiplexing imaging system, assigning a unique set of Stokes vectors as reference features to each field of view, and constructing a polarization reference template corresponding to each field of view region.

[0025] In the above technical solution, step S4 specifically involves comparing the reconstructed Stokes vector of each pixel with each polarization reference template, and calculating the similarity or distance metric of each pixel in Stokes space.

[0026] In the above technical solution, in step S4, when comparing the reconstructed Stokes vector of each pixel with each polarization reference template, normalized vector similarity is used as the discrimination criterion. By comparing the size of the similarity of different categories, the pixel is classified into the most matching field of view category.

[0027] In the above technical solution, step S5 specifically involves: performing threshold segmentation, local contrast enhancement, or other infrared small target detection processing on the image based on the total intensity component or its enhanced form in the Stokes vector to obtain potential target regions; performing statistical analysis on the field-of-view classification results of all pixels in the target region, and using majority voting or weighted voting strategies to determine the most likely source field of view of the target.

[0028] In the above technical solution, the output of step S6 includes: the field of view number to which the target belongs, the spatial coordinates, and the intensity or polarization characteristic information corresponding to the target.

[0029] The present invention has the following beneficial effects:

[0030] The present invention provides a polarization decoding method for multi-field aliased infrared images based on Stokes vector reconstruction. By introducing polarization information as an orthogonal physical dimension independent of intensity information, this method extends the traditional detection method, which relies solely on scalar intensity features, to a discrimination mechanism based on vector information. In cases of spatial aliasing of multi-field signals, signals from different sources may be difficult to distinguish in the intensity domain, but they remain separable in the Stokes vector space. This effectively solves the problem of insufficient recognition capability of traditional intensity methods in aliased scenes.

[0031] The present invention discloses a polarization decoding method for multi-field aliased infrared images based on Stokes vector reconstruction. This method constructs a multi-channel polarization measurement model and utilizes the Stokes vector reconstruction method to achieve high-fidelity inversion of the polarization state. Compared to traditional methods that rely on only single-channel or few-channel information, multi-channel observation introduces information redundancy, enabling the system to maintain high reconstruction accuracy even in the presence of measurement noise or signal disturbances, thereby significantly improving the system's noise resistance and stability.

[0032] This invention presents a polarization decoding method for multi-field-of-view aliasing infrared images based on Stokes vector reconstruction. It constructs a pixel-level classification method based on Stokes vector similarity, transforming the multi-field-of-view aliasing problem into a multi-class discrimination problem in vector space. This method does not rely on the spatial morphology or point spread function features of the target. Compared to existing methods based on field-of-view shape modulation, it avoids dependence on precise optical system design and strict alignment, thereby reducing system implementation complexity and improving adaptability to optical errors. Simultaneously, this invention introduces a target-level statistical decision mechanism, performing region-level information fusion based on pixel-level classification results, and employing majority voting or weighted strategies to determine the target's field of view. This method effectively suppresses the impact of single-pixel misjudgments, reduces the false alarm rate, and improves the stability and reliability of target recognition results.

[0033] The present invention provides a polarization decoding method for multi-field aliased infrared images based on Stokes vector reconstruction. This method achieves high-precision decoding and target reconstruction of multi-field aliased signals without increasing the number of detector pixels or hardware complexity. This improves information acquisition capabilities under limited system resources and has good engineering application value and promotion potential. Attached Figure Description

[0034] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0035] Figure 1 This is a schematic diagram of six-channel polarization intensity measurement data.

[0036] Figure 2 For the reconstructed Stokes parameter map and pixel-level classification map.

[0037] Figure 3 This is a schematic diagram of the target-level field-of-view recognition results.

[0038] Figure 4 This is a schematic diagram illustrating the steps of the Stokes vector-based multi-field aliasing infrared image polarization decoding method of the present invention. Detailed Implementation

[0039] The inventive concept of this invention is as follows:

[0040] The purpose of this invention is to provide a polarization decoding method for multi-field-of-view aliased infrared images based on Stokes vector reconstruction. This invention aims to extend the modulation domain from traditional "scalar intensity encoding" to "vector polarization field encoding" by introducing polarization as an orthogonal physical information dimension, thereby achieving high-fidelity separation of deeply aliased signals without increasing the size of the detector hardware. Specifically, this invention constructs a six-channel polarization measurement model and a Stokes vector linear inversion algorithm, utilizes the redundancy of polarization information to suppress measurement noise, and combines vector similarity matching and a target-level voting decision mechanism to achieve accurate identification and localization of the target's field of view. The ultimate goal of this invention is to significantly improve the target detection probability under a detector multiplexing architecture and enhance the system's robustness to optical alignment errors and modulation deviations, providing an efficient computational imaging solution for next-generation miniaturized, large-field-of-view infrared sensing systems.

[0041] This invention provides a polarization decoding method for multi-field-of-view aliased infrared images based on Stokes vector reconstruction, applicable to scenarios where multi-field-of-view signals spatially aliased in the detector plane of a detector multiplexing imaging system. This method introduces polarization information as an orthogonal physical dimension independent of traditional intensity information, mapping signals from different field-of-view sources to a discriminative polarization vector space. This solves the aliasing problem that is difficult to distinguish in the intensity domain, enabling effective discrimination and precise localization of the target's spatial source.

[0042] Simultaneously, a processing flow of "multi-channel polarization measurement—Stokes vector reconstruction—polarization domain classification—target-level decision-making" is constructed. First, intensity information of the aliased image under different polarization states is obtained through multi-channel polarization measurement; then, pixel-level Stokes vectors are reconstructed using a linear inversion method; further, based on a preset polarization reference template, similarity discrimination is performed on each pixel in Stokes space to achieve dealiasing of multi-field information; finally, by combining foreground target extraction and statistical voting strategies, the field of view to which the target belongs is determined at the target region scale, thereby completing the reconstruction of the target's spatial location. This invention effectively improves the target recognition accuracy and system robustness under detector reuse conditions by transforming the spatial aliasing problem into a vector space classification problem.

[0043] The present invention will now be described in detail with reference to the accompanying drawings.

[0044] This invention provides a polarization decoding method for multi-field-of-view aliased infrared images based on Stokes vector reconstruction, primarily applied to scenarios where multi-field-of-view signals spatially aliased in the detector plane within a detector multiplexing imaging system. This method introduces polarization information as a physical dimension independent of intensity distribution, mapping aliased signals that are originally indistinguishable in the intensity domain to a separable polarization vector space. This enables effective differentiation of signals from different field-of-view sources and accurate reconstruction of the target's spatial location. See the schematic diagram of the method's steps. Figure 4 As shown, the main steps include the following:

[0045] Step S1: Acquisition of multi-channel polarization measurement data;

[0046] In a detector multiplexing imaging system, the first step is to perform multi-channel polarization measurements on the infrared images that have been polarized and superimposed on the detector plane. Specifically, by setting up polarization analyzers or equivalent polarization measurement devices with different orientations, the radiation intensity at the same pixel location under multiple polarization states is acquired, thereby obtaining multi-channel observation data. Preferably, the multi-channel polarization measurement includes at least six independent channels, corresponding to the linear polarization components at 0°, 45°, 90°, and 135°, as well as the left-hand circular polarization and right-hand circular polarization components. The results of the above multi-channel polarization measurements together constitute a redundant observation set describing the pixel polarization state, providing basic data support for subsequent polarization information inversion. Through this step, the differences in polarization characteristics of signals from different fields of view can be preserved even in cases of severe spatial aliasing.

[0047] Step S2: Stokes vector reconstruction;

[0048] After obtaining multi-channel polarization measurement data, the Stokes vector is reconstructed for each pixel location. The Stokes vector is used to describe the complete polarization state of light, including the total intensity component. And components characterizing linear and circular polarization properties , and This invention establishes a linear relationship model between polarization measurement intensity and Stokes parameters, and performs inversion calculations on multi-channel observation data to obtain the Stokes vectors of corresponding pixels. Preferably, the inversion calculation process can be implemented using the least squares method or matrix inversion method to improve reconstruction accuracy under conditions of measurement noise. Through this step, the original intensity superposition signal can be converted into a vector representation with directional and structural information, providing higher-dimensional information for subsequent classification.

[0049] Step S3: Construction of polarization reference template;

[0050] To distinguish signals from different fields of view, this invention pre-constructs polarization reference templates corresponding to each field of view region. Specifically, based on the modulation parameters applied to different fields of view by the polarization coding module in the detector multiplexing imaging system, a unique set of Stokes vectors is assigned as reference features for each field of view. The polarization reference templates can be obtained through theoretical calculations or through system calibration experiments. In practical applications, the reference Stokes vectors corresponding to each field of view have a certain interval in vector space, thus ensuring good separability between different categories. Through this step, a one-to-one correspondence between field of view labels and polarization features can be established.

[0051] Step S4: Pixel-level classification based on Stokes space;

[0052] After obtaining the pixel-level Stokes vector and polarization reference templates, each pixel in the aliased image is classified based on its polarization vector. Specifically, the reconstructed Stokes vector of each pixel is compared with each polarization reference template, and their similarity or distance metric in Stokes space is calculated. Preferably, normalized vector similarity is used as the discrimination criterion; by comparing the magnitude of the similarity between different categories, the pixel is assigned to the most matching field-of-view category. This process essentially transforms the spatial aliasing problem into a multi-class classification problem in vector space, enabling signals that are originally inseparable in the intensity domain to be effectively distinguished in the polarization domain, thereby completing pixel-level field-of-view dealiasing.

[0053] Step S5: Foreground target extraction and field of view determination;

[0054] After completing pixel-level classification, to improve the stability and noise resistance of target detection, this invention further extracts foreground targets from the image. Specifically, this can be based on the total intensity component in the Stokes vector. Alternatively, or in an enhanced form, thresholding, local contrast enhancement, or other infrared small target detection processing can be performed on the image to obtain potential target regions. This step allows for the separation of target regions with significant radiation characteristics from complex backgrounds, providing a basis for subsequent target-level decision-making. After extracting the target regions, this invention performs a statistically based field-of-view determination for each target region. Specifically, the field-of-view classification results of all pixels within the target region are statistically analyzed, and a majority voting or weighted voting strategy is used to determine the most probable source field of view for the target. Since single-pixel classification results may be affected by noise or measurement errors, statistical fusion at the target region scale can effectively reduce the probability of misclassification and improve the robustness and reliability of the overall recognition.

[0055] Step S6: Target Reconstruction and Output;

[0056] Finally, based on the determined field-of-view information of the target, the target is mapped back to its corresponding original spatial location, achieving spatial reconstruction and localization of the target, and outputting the results. The output results may include the target's field-of-view number, spatial coordinates, and corresponding intensity or polarization characteristic information. Through the above steps, this invention achieves accurate analysis of the target's origin from multi-field-of-view aliased infrared images.

[0057] In summary, this invention transforms the traditional aliasing problem, which relies on intensity information, into a classification problem in vector space by constructing a processing framework that combines multi-channel polarization measurement, Stokes vector reconstruction, polarization domain classification, and target-level decision-making. This effectively improves the accuracy and robustness of target recognition and localization in multi-field-of-view multiplexing imaging systems.

[0058] In the simulation experiment, a multi-field-of-view detector multiplexing imaging scenario was constructed. The input infrared image was divided into multiple independent field-of-view regions, and the signals of each field of view were spatially superimposed after polarization encoding to simulate the aliased observation data in the actual system. Based on this, multi-channel polarization measurement data was generated for the aliased image, including intensity information of different linear polarization directions and circular polarization components. Subsequently, the polarization state of each pixel was inverted using the Stokes vector reconstruction method proposed in this invention, and pixel-level field-of-view classification and target-level determination were completed by combining a preset polarization reference template, thereby realizing the reconstruction of the spatial source of the target.

[0059] like Figure 1-3 As shown, Figure 1 This diagram illustrates the six-channel polarization intensity measurement data. It shows the direct input source of the decoding algorithm, i.e., after spatial folding and energy superposition, the data is analyzed by six polarization analyzers (0° / 45° / 90° / 135° linear polarization and left / right circular polarization). Figure 1 I are used in the middle respectively 0° I 45° I 90° I 135° I LHC I RHC (This represents) the original aliasing intensity frame acquired. Figure 2 This image shows the reconstructed Stokes parameter map and pixel-level classification map. It illustrates the intermediate computation results of the decoding algorithm, demonstrating how vector features are reconstructed from aliasing intensity through linear inversion, and showcasing the preliminary pixel-level classification effect based on normalized vector similarity matching. These represent the three components of the Stokes vector. Figure 3 This diagram illustrates the target-level field-of-view recognition results. It shows the final output of the invention, which combines the total intensity map, the foreground segmentation boundary, and the target space source determination results after applying the "majority voting" strategy. It is intuitive evidence to prove the core innovation of "target-level reconstruction".

[0060] Experimental results show that, under ideal noise-free conditions, the method of this invention can achieve high-precision Stokes vector reconstruction, with the normalized reconstruction error of each component controlled at a low level (within approximately 0.02), indicating that polarization information can be effectively preserved during aliasing. Based on this, a classification method based on Stokes spatial similarity can achieve pixel-level field-of-view discrimination with a classification accuracy of over 97%. Further combining this with a statistical voting strategy for the target region, the target-level field-of-view recognition accuracy can reach over 97%, verifying the effectiveness of the method in aliasing scenarios. To evaluate the method's noise resistance, Gaussian noise of varying intensities was introduced into the multi-channel polarization measurement data for simulation analysis. Experimental results show that, under the condition of noise standard deviation σ≤0.02, the target-level recognition accuracy can still be maintained above 95%, indicating that this invention, through multi-channel polarization redundancy information and the Stokes inversion mechanism, can effectively suppress the influence of measurement noise on the results and has good robustness. Furthermore, to verify the method's adaptability to non-ideal system factors, polarization modulation parameter errors were introduced into the simulation to perturb the polarization state within a certain range. The results show that when the polarization modulation angle error is within ±10°, the pixel-level classification accuracy remains at a high level (more than 92%), indicating that the method of the present invention has a strong tolerance for modulation errors in the actual system.

[0061] In summary, through numerical simulation experiments, the polarization decoding method for multi-field aliased infrared images based on Stokes vector reconstruction of the present invention can accurately determine the source of the target under multi-field aliasing conditions, and maintains high recognition accuracy and stability even in the presence of measurement noise and modulation errors, proving the feasibility and application value of this method in detector multiplexing infrared imaging systems.

[0062] The present invention also compares the performance of the Stokes vector-based multi-field aliasing infrared image polarization decoding method of the present invention with existing detection methods. The comparison results are shown in Table 1.

[0063] Table 1: Performance Comparison Results of Different Detection Methods

[0064]

[0065] As shown in Table 1, the performance comparison results of different detection methods indicate that the proposed polarization coding method has a higher detection probability ( The false alarm rate reached 95.90%, significantly outperforming existing technologies such as Top-hat transform filtering (74.82%), Max-Median filtering (68.59%), MDPS-LGD (corrected density peak search and local gray-scale difference, 89.63%), and FOV shape modulation (94.04%). Meanwhile, its false alarm rate (…) The efficiency was 1.25%, the lowest among all comparison methods. The technical solution of this invention exhibits significant beneficial effects: First, by introducing the orthogonal physical dimension of polarization, the modulation domain is extended from scalar intensity to vector polarization field, effectively solving the deep aliasing problem caused by detector multiplexing. Even if multiple targets completely overlap in the intensity domain, high-fidelity separation can still be achieved using a unique Stokes vector feature. Second, by utilizing the information redundancy introduced by six-channel polarization measurement, random measurement noise can be effectively suppressed through a linear inversion mechanism, and it exhibits extremely strong engineering tolerance to optical alignment errors (such as modulation deviations within ±10°). Thus, without increasing the detector size and power consumption, it achieves a significant expansion of the detection field of view and a substantial improvement in system detection robustness.

[0066] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A polarization decoding method for multi-field aliased infrared images based on Stokes vector reconstruction, characterized in that, Includes the following steps: Step S1: Acquisition of multi-channel polarization measurement data; Multi-channel polarization measurement is performed on infrared images that have been polarized and superimposed on the detector plane; Step S2: Stokes vector reconstruction; After obtaining multi-channel polarization measurement data, Stokes vector reconstruction is performed on each pixel position to convert the original intensity superposition signal into a vector representation with directional and structural information. Step S3: Construction of polarization reference template; Pre-construct polarization reference templates corresponding to each field of view region to establish a one-to-one correspondence between field of view labels and polarization features; Step S4: Pixel-level classification based on Stokes space; Based on the reconstructed Stokes vector of each pixel in step S2 and the polarization reference template constructed in step S3, each pixel in the aliased image is classified and discriminated based on the polarization vector. Step S5: Foreground target extraction and field of view determination; Foreground targets in the image are extracted, target regions with significant radiation characteristics are separated, and a statistical field of view determination is performed for each target region; Step S6: Target Reconstruction and Output; Based on the determined field of view information of the target, the target is mapped back to its original spatial location, realizing the spatial reconstruction and positioning of the target, and the result is output.

2. The polarization decoding method for multi-field aliased infrared images based on Stokes vector reconstruction according to claim 1, characterized in that, The multi-channel polarization measurement in step S1 specifically involves: by setting up polarization analyzers or equivalent polarization measurement devices with different orientations, the radiation intensity of the same pixel location under multiple polarization states is collected, thereby obtaining multi-channel observation data.

3. The polarization decoding method for multi-field aliased infrared images based on Stokes vector reconstruction according to claim 1, characterized in that, The multi-channel polarization measurement in step S1 includes at least six independent channels, corresponding to linear polarization components in the 0°, 45°, 90° and 135° directions, as well as left-hand circular polarization and right-hand circular polarization components.

4. The polarization decoding method for multi-field aliased infrared images based on Stokes vector reconstruction according to claim 2, characterized in that, Step S2 specifically involves: establishing a linear relationship model between polarization measurement intensity and Stokes parameters, performing inversion calculations on multi-channel observation data, and thus obtaining the Stokes vector of the corresponding pixel.

5. The polarization decoding method for multi-field aliased infrared images based on Stokes vector reconstruction according to claim 4, characterized in that, In step S2, the least squares method or matrix inversion method is used to realize the inversion calculation process.

6. The polarization decoding method for multi-field aliased infrared images based on Stokes vector reconstruction according to claim 1, characterized in that, Step S3 specifically involves: based on the modulation parameters applied to different fields of view by the polarization coding module in the detector multiplexing imaging system, assigning a unique set of Stokes vectors as reference features to each field of view, and constructing a polarization reference template corresponding to each field of view region.

7. The polarization decoding method for multi-field aliased infrared images based on Stokes vector reconstruction according to claim 1, characterized in that, Step S4 specifically involves comparing the reconstructed Stokes vector of each pixel with each polarization reference template, and calculating the similarity or distance metric of each pixel in Stokes space.

8. The polarization decoding method for multi-field aliased infrared images based on Stokes vector reconstruction according to claim 7, characterized in that, In step S4, when comparing the reconstructed Stokes vector of each pixel with each polarization reference template, normalized vector similarity is used as the discrimination criterion. By comparing the magnitude of the similarity of different categories, the pixel is classified into the most matching field of view category.

9. The polarization decoding method for multi-field aliased infrared images based on Stokes vector reconstruction according to claim 1, characterized in that, Step S5 specifically involves: performing thresholding, local contrast enhancement, or other infrared small target detection processing on the image based on the total intensity component or its enhanced form in the Stokes vector, in order to obtain potential target regions; Statistical analysis is performed on the field-of-view classification results of all pixels within the target area, and the most likely source field of view of the target is determined by majority voting or weighted voting strategies.

10. The polarization decoding method for multi-field aliased infrared images based on Stokes vector reconstruction according to claim 1, characterized in that, The output of step S6 includes: the field of view number to which the target belongs, the spatial coordinates, and the intensity or polarization characteristic information corresponding to the target.

Images