A scene-adaptive polarization channel combination detection method

Abstract

This invention provides a scene-adaptive polarization channel combination detection method, belonging to the field of imaging detection technology. The method includes the following steps: S1: acquiring intensity images of multiple polarization channels, fitting the images using the least squares method to obtain a fitted image, and obtaining a Stokes parameter image of the scene target through a preset first calculation process; S2: generating intensity images corresponding to the polarization angles based on the Stokes parameter images; S3: traversing the polarization channel combinations; S4: calculating statistical parameters of the target and background through a preset second calculation process to obtain the desired polarization channel combination. This method solves the problem of low time efficiency inherent in polarization channel combination methods that require acquiring multiple polarization channel intensity images at the hardware level.

Description

Technical Field

[0001] This invention relates to the field of imaging detection technology, and more specifically, to a scene-adaptive polarization channel combination detection method. Background Technology

[0002] Polarization imaging, as a novel photoelectric detection technology, can simultaneously acquire the spatial and polarization characteristics of a target. For specific application scenarios, it offers enhanced material and contour identification capabilities, contributing to improved combat effectiveness in tactical and strategic early warning, battlefield situational awareness, and precision target strikes. Among existing imaging systems, rotary polarization imaging systems employ a time-series operation to control the rotation of polarization devices to a specific angle, achieving the acquisition of images with different polarization information for the same scene.

[0003] Rotary polarization imaging systems employ a polarization direction rotation tuning method to indirectly measure incident light with unknown polarization states, thereby acquiring the polarization information of scene targets. Rotary polarization imaging systems offer advantages such as analyzing the intrinsic polarization state of scene targets from any angle, comprehensively deconstructing target polarization information, and excellent target detection and recognition performance. However, extracting useful information from rich polarization channel data and effectively fusing it to achieve efficient and accurate differentiation between background and target remains challenging. There is an urgent need to explore new imaging modes and analytical methods to solve the problem of scene-adaptive polarization channel combination detection. Existing linear polarization imaging systems suffer from poor scene adaptability, low target detection accuracy with traditional polarization channel combination methods, and time-consuming existing angle combination optimization methods. Summary of the Invention

[0004] The problem addressed by this invention is how to solve the problem of low time efficiency in polarization channel combination methods that require obtaining multiple polarization channel intensity images from the hardware level.

[0005] To address the above problems, this invention provides a scene-adaptive polarization channel combination detection method, comprising the following steps:

[0006] S1: Obtain the intensity image of the multi-polarization channel, fit the image using the least squares method to obtain the fitted image, and obtain the Stokes parameter image of the scene target through the preset first calculation process;

[0007] S2: Generate the intensity image corresponding to the polarization angle based on the Stokes parametric image;

[0008] S3: Traverse polarization channel combinations;

[0009] S4: Calculate the target and background statistical parameters through a preset second operation process to obtain the desired polarization channel combination.

[0010] In the above method, by obtaining prior information on the polarization features of the scene target, the desired polarization channel combination, i.e. the optimal polarization channel combination configuration, is obtained, which solves the problems of low efficiency in polarization information acquisition and low target detection accuracy under the premise of time cost.

[0011] Further, step S1 includes:

[0012] S11: Adjust the polarization rotation mechanism in the rotary polarization imaging system to multiple angular positions to acquire intensity images of multiple polarization channels;

[0013] S12: Based on the multi-polarization channel intensity image obtained in step S11, the Marius equation is used as the fitting objective function, and the nonlinear least squares method is used to obtain the fitting image corresponding to the multi-polarization channel.

[0014] S13: The preset first operation process is to subtract the scene target intensity image obtained in step S11 from the least squares fitted image obtained in step S12 to obtain the error image corresponding to the multi-polarization channel.

[0015] S14: Obtain the Stokes parametric image based on the fitted image in step S12.

[0016] In the above method, the intensity image of the scene target is obtained. The pixel value distribution corresponding to the intensity image at different polarization angles follows Malus's law. According to statistical theory, when only additive Gaussian noise is considered, the gray value of the error image follows a Gaussian distribution.

[0017] Further, in step S12, after the incident light beam reaches the detector via the polarization rotation mechanism, the intensity image is represented as follows:

[0018]

[0019] Among them, I unpol Indicates the non-polarized component of the incident ray; I pol α represents the polarization component of the incident ray; α represents the polarization angle of the incident ray; θ i Indicates the rotation angle of the polarization device; A graph representing the intensity of an ideal object that obeys Malus's law;

[0020] The fitted image is represented as:

[0021]

[0022] in, This represents the fitted image corresponding to different polarization angles; This represents the multi-polarization channel intensity image obtained in step S11.

[0023] Further, in step S13, the error image is represented as follows:

[0024]

[0025] in, Represents θ i Error image at polarization angle; Represents θ i Intensity image obtained at the polarization angle; Represents θ i Fitted intensity image at polarization angle;

[0026] The probability density function is expressed as:

[0027]

[0028] in, Represents the variance of grayscale values ​​in the error image; This represents the mean grayscale value of the error image.

[0029] Further, in step S14, the Stokes parametric image is represented as follows:

[0030]

[0031] Among them, S fit Represents the fitted Stokes vector; This represents the elliptic polarization component, which is usually negligible in linear polarization imaging detection systems.

[0032] Furthermore, based on the Stokes parameter image obtained in step S14, the intensity image corresponding to the polarization angle is represented as follows:

[0033]

[0034] Furthermore, the preset second operation process in step S4 is as follows: by using the intensity image corresponding to the polarization angle, the target region and the background region are selected, and the statistical parameter expected μ of the target region is calculated. t and background region expectation μ b and the target region covariance Σ t Covariance Σ of the background region b ;

[0035] Wherein, the mean μ and covariance Σ are expressed as:

[0036]

[0037]

[0038] Where, xi This represents the grayscale value at different pixels in the intensity image; N represents the number of pixels.

[0039] Furthermore, based on the mean and covariance matrices of the target and background, the optimization function, using distance JM, is expressed as:

[0040]

[0041] In the above method, the larger the JM distance value between the target and the background, the smaller the difference between their corresponding feature vectors, and the greater the probability that the target will be detected from the background.

[0042] Furthermore, the intensity image corresponding to the joint polarization angle is iterated with the mean and covariance matrix of the target and background to update the images at different polarization angles. The desired polarization channel combination is expressed as:

[0043]

[0044] Furthermore, it also includes the following steps:

[0045] S5: Based on the desired polarization channel combination obtained in step S4, drive the rotation mechanism in the rotary polarization imaging system to the corresponding angle.

[0046] The present invention employing the above technical solution has the following beneficial effects:

[0047] This invention can obtain the optimal polarization channel combination configuration by acquiring prior information on the polarization characteristics of scene targets, thereby solving the problems of low efficiency in polarization information acquisition and low target detection accuracy under the premise of time cost. Attached Figure Description

[0048] Figure 1 The flowchart of the scene-adaptive polarization channel combination detection method provided in the embodiments of the present invention Figure 1 ;

[0049] Figure 2 The flowchart of the scene-adaptive polarization channel combination detection method provided in the embodiments of the present invention Figure 2 ;

[0050] Figure 3 A schematic diagram of a rotary polarization imaging system for a scene-adaptive polarization channel combination detection method provided in an embodiment of the present invention;

[0051] Figure 4 A flowchart comparing the traditional method and the method proposed in this invention for the scene-adaptive polarization channel combination detection method provided in this embodiment of the invention;

[0052] Figure 5A schematic diagram of the polarization characteristic curve of the scene-adaptive polarization channel combination detection method provided in the embodiments of the present invention, based on Malus's law.

[0053] Figure 6 The scene target and its DoLP image are simulated by the scene-adaptive polarization channel combination detection method provided in the embodiments of the present invention;

[0054] Figure 7 A schematic diagram illustrating the process of obtaining the scene-adaptive polarization channel combination configuration of the scene-adaptive polarization channel combination detection method provided in this embodiment of the invention;

[0055] Figure 8 The target detection kernel image under different polarization channel combinations of the scene-adaptive polarization channel combination detection method provided in the embodiments of the present invention;

[0056] Figure 9 ROC detection curves and box plots for different polarization channel combinations of the scene-adaptive polarization channel combination detection method provided in this embodiment of the invention. Detailed Implementation

[0057] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0058] The following are specific embodiments of the present invention, which are described in conjunction with the accompanying drawings. However, the present invention is not limited to these embodiments.

[0059] Example

[0060] This embodiment provides a scene-adaptive polarization channel combination detection method, such as... Figure 1 and Figure 2 As shown, this method includes the following steps:

[0061] S1: Obtain the intensity image of the multi-polarization channel, fit the image using the least squares method to obtain the fitted image, and obtain the Stokes parameter image of the scene target through the preset first calculation process;

[0062] S2: Generate the intensity image corresponding to the polarization angle based on the Stokes parametric image;

[0063] S3: Traverse polarization channel combinations;

[0064] S4: Calculate the target and background statistical parameters through a preset second operation process to obtain the desired polarization channel combination.

[0065] Specifically, when using the four-polarization channel analytical model, the multi-polarization channel analytical models for 0°, 45°, 90°, and 135° can be obtained as follows:

[0066]

[0067] In the above formula Rotate the polarizer by θ i The intensity image acquired by the detector at the specified angle. That is, when acquiring a traditional four-channel polarization image, rotating the polarization rotation mechanism in the rotary polarization imaging system to 0°, 45°, 90°, and 135° respectively will yield the desired image. The image can then be used to solve for the Stokes parametric image and polarization feature information of the scene target.

[0068] When the polarization detection system adopts a three-polarization channel analytical model of 0°, 60°, and 120°, we can obtain:

[0069]

[0070] Furthermore, when the polarization detection system adopts a multi-polarization channel analytical model with N different polarization directions, and the interval angle between adjacent directions is the same, we can obtain:

[0071]

[0072] Where, θ i =θ 0° +[(i-1)×180°] / N; S represents the fitted Stokes vector; s3 represents the elliptic polarization component, which can usually be ignored in linear polarization imaging detection systems.

[0073] The rotating polarization imaging system uses a timing-based operation to control the rotation of the polarization device to a specific direction and angle. For the composition of this imaging system, please refer to [link / reference needed]. Figure 3 It mainly includes the target and background, polarization rotation mechanism, lens and detector, and its basic working mechanism is as follows:

[0074] The Stokes vector S = [s0, s1, s2, s3] can represent the polarization state of light. For polarization imaging optical systems, S is used... in S represents the Stokes vector of the incident light. out The Stokes vector representing the outgoing light beam is shown. After the beam passes through the imaging system, the system's modulation of the light can be represented by the Mueller matrix, i.e.,

[0075]

[0076] Based on the Mueller matrix of the linear polarization device, the above equation can be expressed as:

[0077]

[0078] In a linear polarization imaging system, the photosensitive pixel units of the imaging detector are only sensitive to the intensity of the incident light. Therefore, the above equation can be expressed as:

[0079]

[0080] in, The polarization direction is represented by θ. i The light intensity measured by the detector at the angle.

[0081] According to the above formula, the detector needs to acquire intensity information from at least three different polarization directions to solve for the Stokes parameters of the target scene, and then calculate the target's degree of polarization and polarization angle, etc. (degree of polarization) Polarization angle AoLP = tan -1 (s1 / s2) / 2).

[0082] For rotary polarization imaging systems, the polarization direction rotation tuning method allows for indirect measurement of incident light with unknown polarization states, thereby obtaining the polarization information of the scene target. During the imaging process, images at different polarization angles are acquired through sequential exposure of the imaging detector, as expressed below:

[0083]

[0084] In the formula: I i The light intensity detected by the photosensitive pixels of the detector; S in Let represent the polarization state of the incident light; W represent the polarization channel combination mode constructed from multiple polarization directions; and n represent the noise of the imaging detection system. Based on the inverse matrix transformation, the Stokes vector expression of the incident light is obtained as follows:

[0085]

[0086] In the formula: For the reconstructed Stokes vector, W -1 This represents the inverse of matrix W.

[0087] In actual operation, the imaging system is inevitably affected by factors such as scene environment, detector and atmospheric transmission when acquiring information, which causes the feature information obtained by different polarization channel combinations to be different. That is, in the rotary polarization imaging system, more diverse polarization channel combination configurations can effectively change the expression of the above formula W, thereby affecting the reconstruction of target polarization feature information and detection results.

[0088] See Figure 4 -(a) and Figure 4-(b) This invention proposes a scene-adaptive polarization channel combination detection method in a linear polarization imaging system. It only needs to obtain the intensity information of traditional three- or four-polarization channels or the intensity information of four-polarization channels to solve the optimal polarization channel combination configuration under different types of target scenes.

[0089] Specifically, the scene-adaptive polarization channel combination detection method in the linear polarization imaging system can achieve adaptive, efficient, and optimized polarization channel configuration search for targets in different types of scenes, significantly increasing the intelligent imaging capability of the rotary polarization imaging system. It overcomes the limitations of low target detection rates in traditional three-polarization channel (0°, 60°, 120°), (0°, 45°, 90°), and four-polarization channel (0°, 45°, 90°, 135°) analytical modes, achieving high-detection-rate target detection and recognition through optimized algorithm search. Furthermore, the adaptive polarization channel optimization method only requires acquiring intensity image information from traditional three-polarization channel (0°, 60°, 120°), (0°, 45°, 90°), or four-polarization channel (0°, 45°, 90°, 135°) images, simplifying the optimal channel combination optimization steps and offering high time efficiency.

[0090] This embodiment provides a background Stokes vector S. Background = [1, 0.022, -0.074], the target Stokes vector is S Target =[1,0.087,-0.21], and an example of scene adaptive acquisition of three polarization channel combination configuration with additive white Gaussian noise added. A simulation platform for generating scene target polarization features was established using MATLAB software. The designed scene target is shown in the reference. Figure 6 For the simulated scenarios, please refer to Figure 6 (a), see the corresponding DoLP image. Figure 6 (b) The corresponding DoLP image after adding additive Gaussian noise is shown in the attached image. Figure 6 (c). For the above simulation scenario, the basic process of the adaptive polarization channel combination detection method is described in [reference needed]. Figure 7 .

[0091] See Figure 2 Step S1 includes:

[0092] S11: Adjust the polarization rotation mechanism in the rotary polarization imaging system to multiple angular positions to acquire intensity images of multiple polarization channels;

[0093] S12: Based on the multi-polarization channel intensity image obtained in step S11, the Marius equation is used as the fitting objective function, and the nonlinear least squares method is used to obtain the fitting image corresponding to the multi-polarization channel.

[0094] S13: The preset first operation process is to subtract the scene target intensity image obtained in step S11 from the least squares fitted image obtained in step S12 to obtain the error image corresponding to the multi-polarization channel.

[0095] S14: Obtain the Stokes parametric image based on the fitted image in step S12.

[0096] Specifically, in step S11, the scene target intensity image is acquired by adjusting the polarization rotation mechanism in the rotary polarization imaging system to 0°, 45°, 90°, and 135° angle positions, respectively, to obtain a traditional four-polarization channel intensity image. Step S12 involves performing least squares image fitting, which uses the four-polarization channel intensity images obtained in step S11, with the Malus equation as the fitting objective function, and employs nonlinear least squares to obtain the fitted images corresponding to the four polarization channels. The optimization problem solved during the fitting process is as follows:

[0097]

[0098] In the formula, θ i The angles are 0°, 45°, 90°, and 135°.

[0099] The error probability distribution function image is solved in step S13. Taking the detection system being interfered with by additive Gaussian noise as an example, the scene target intensity image obtained in step S11 is used to solve the error probability distribution function image. The least squares fitted image obtained in step S12 By subtracting the values, we can obtain the error images corresponding to the four polarization channels. According to statistical theory, the Gaussian distribution probability density function of the gray values ​​of the error image can be solved:

[0100]

[0101] in, The variance of the grayscale values ​​in the error image; This represents the mean grayscale value of the error image.

[0102] The Stokes parameter image is solved in step S14, that is, based on the least squares fitted image obtained in step S12, the Stokes parameter image is solved.

[0103]

[0104] In the formula: The images show the fitted intensity at polarization angles of 0°, 45°, 90°, and 135°, respectively. fit The fitted Stokes vector.

[0105] In step S12, after the incident light beam passes through the polarization rotation mechanism and reaches the detector, the intensity image is represented as follows:

[0106]

[0107] See Figure 5 , among which, I unpol Indicates the non-polarized component of the incident ray; I pol α represents the polarization component of the incident ray; α represents the polarization angle of the incident ray; θ i Indicates the rotation angle of the polarization device; A graph representing the intensity of an ideal object that obeys Malus's law;

[0108] The fitted image is represented as:

[0109]

[0110] in, This represents the fitted image corresponding to different polarization angles; This represents the multi-polarization channel intensity image obtained in step S11.

[0111] In step S13, the error image is represented as follows:

[0112]

[0113] in, Represents θ i Error image at polarization angle; Represents θ i Intensity image obtained at the polarization angle; Represents θ i Fitted intensity image at polarization angle;

[0114] The probability density function is expressed as:

[0115]

[0116] in, Represents the variance of grayscale values ​​in the error image; This represents the mean grayscale value of the error image.

[0117] In step S14, the Stokes parametric image is represented as follows:

[0118]

[0119] in, Represents θ i Fitted intensity image at polarization angle, S fitThis represents the fitted Stokes vector.

[0120] The intensity image corresponding to the polarization angle obtained from the Stokes parameter image obtained in step S14 is represented as follows:

[0121]

[0122] Specifically, based on the error probability distribution function image corresponding to the four polarization channels obtained in step S13 and the Stokes parameter image obtained in step S14, the different polarization directions θ in the range of 0° to 180° with a step size of 5° are solved. i Intensity image corresponding to the angle

[0123] The preset second operation process in step S4 is as follows: by selecting the target region and the background region through the intensity image corresponding to the polarization angle, and calculating the statistical parameter expected μ of the target region. t and background region expectation μ b and the target region covariance Σ t Covariance Σ of the background region b ;

[0124] Wherein, the mean μ and covariance Σ are expressed as:

[0125]

[0126]

[0127] Where, x i This represents the grayscale value at different pixels in the intensity image; N represents the number of pixels.

[0128] Specifically, the target region of interest (ROI) is selected using the generated images at different polarization angles. Target and background region ROI Background And calculate their statistical parameters, namely the expected μ of the target region. t and background region expectation μ b and the target region covariance Σ t Covariance Σ of the background region b ,Right now,

[0129]

[0130] in: and N represents the image grayscale values ​​corresponding to the target and background regions, respectively; t and N b These represent the number of image pixels corresponding to the target and background regions, respectively.

[0131] Here, based on the mean and covariance matrices of the target and background, the distance JM (Jeffries-Matusita) is used as the optimization function, expressed as:

[0132]

[0133] In this process, the intensity image corresponding to the joint polarization angle is iterated with the mean and covariance matrix of the target and background to update the image at different polarization angles. The desired polarization channel combination is expressed as:

[0134]

[0135] Specifically, to find the optimal polarization channel combination, based on the mean and covariance matrices of the target and background, the distance JM is selected as the optimization function. To obtain the optimal solution, iterative processing is performed to update images at different polarization angles. The final optimized three-polarization channel combination is:

[0136] f opt (θ1, θ2, θ3)=[55°, 60°, 145°].

[0137] This also includes the following steps:

[0138] S5: Based on the desired polarization channel combination obtained in step S4, drive the rotation mechanism in the rotary polarization imaging system to the corresponding angle.

[0139] Specifically, the target anomaly detection is completed, and the optimized polarization channel combination f is obtained. opt (θ1,θ2,θ3)=[55°,60°,145°], by driving the rotating mechanism of the rotary polarization imaging system to the corresponding polarization angle, the optimal polarization channel combination image group can be obtained. By combining the RX target anomaly detection method, scene target anomaly detection can be achieved. The RX anomaly detection operator is:

[0140] RX(x)=(x-μ1) T C -1 (x-μ1);

[0141] Where x represents the pixel to be detected; μ1 represents the background mean; and C represents the background covariance matrix.

[0142] See Figure 8The figures show anomaly detection operator images for simulated scene targets under different polarization angle combinations. As can be seen from the figures, the detection operator image corresponding to the optimized three-polarization channel combination configuration designed in this invention has richer target details and edge information and lower false alarm pixel interference, resulting in the best visual effect and target detection performance. The traditional M-Pickering (0°, 45°, 90°, and 135°) polarization angle combination configuration has the second best detection performance; although it can achieve background and target separation detection, it lacks detailed features. The traditional Fessenkov (0°, 60°, and 120°) polarization angle combination configuration suffers severe loss of target detail features and has poor detection performance. The traditional Pickering (0°, 45°, and 90°) polarization angle combination configuration struggles to effectively extract target features from background interference, resulting in the worst detection performance.

[0143] See Figure 9 To quantitatively evaluate the target detection accuracy under different polarization angle combinations and further verify the practicality, effectiveness, and advancement of the optimized polarization channel combination configuration designed using the method proposed in this invention, receiver operating characteristic (ROC) curves and box plots are used to evaluate the performance of different polarization channel combinations. Specifically, the area under the ROC curve (AUC) is used to quantitatively evaluate the target detection performance of different polarization angle combinations; a larger AUC area indicates better detection performance.

[0144] See Figure 9 (a) It can be seen that, under a certain detection rate, the optimized three-polarization channel combination configuration designed in the method proposed in this invention has a lower false alarm rate. Furthermore, the AUC clearly shows that the AUC value of the optimized three-polarization channel combination configuration designed in the method proposed in this invention is greater than that of the other three algorithms, indicating the best detection performance. In the box plot, the height of the boxes represents the suppression of background interference by different polarization channel configurations; the spacing between the boxes represents the algorithm's ability to separate the background and anomalies, with a larger spacing being more conducive to the separation of the target or anomaly. See also... Figure 9 (b) It can be seen that the optimal three-polarization channel combination configuration designed by the method proposed in this invention corresponds to the greatest degree of target box compression and the best background suppression. Simultaneously, the height difference between the boxes is the largest, resulting in good background anomaly separation. The optimal three-polarization channel combination configuration designed by the method proposed in this invention exhibits excellent target anomaly detection performance, effectively separating the background from the target while ensuring the detection of target details and edge information, demonstrating the best overall performance.

[0145] While the disclosure is as stated above, its scope of protection is not limited thereto. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of this disclosure, and all such changes and modifications will fall within the protection scope of this invention.

Claims

1. A scene-adaptive polarization channel combination sounding method, characterized in that, The method comprises the steps of: S1: acquiring intensity images of multiple polarization channels, fitting the images by least squares to obtain fitting images, and obtaining a Stokes parameter image of a scene target through a preset first operation process; S2: generating corresponding intensity images under a polarization angle according to the Stokes parameter image; S3: traversing polarization channel combinations; S4: calculating statistical parameters of the target and the background through a preset second operation process to obtain an expected polarization channel combination.

2. The scene-adaptive polarization channel combination sounding method of claim 1, wherein, The step S1 comprises: S11: adjusting a polarization rotation mechanism in a rotating wheel type polarization imaging system to multiple angular positions respectively to acquire multiple polarization channel intensity images; S12: acquiring the multiple polarization channel intensity images according to step S11, taking Malus equation as a fitting target function, and acquiring fitting images corresponding to the multiple polarization channels by using a nonlinear least squares method; S13: the preset first operation process is to subtract the scene target intensity image acquired in step S11 from the least squares fitting image acquired in step S12 to obtain an error image corresponding to the multiple polarization channels; S14: obtaining a Stokes parameter image according to the fitting image in step S12.

3. The scene-adaptive polarization channel combination sounding method of claim 2, wherein, In the step S12, when incident light passes through the polarization rotation mechanism to reach the detector, the intensity image is represented as: where I unpol represents the non-polarized component of the incident light; I pol represents the polarized component of the incident light; a represents the polarization angle of the incident light; θ i represents the rotation angle of the polarization device; represents the ideal intensity image subject to Malus' law; The fitting image is represented as: wherein, represent the fitting images corresponding to different polarization angles; represent the multi-polarization channel intensity images acquired in step S11.

4. The scene-adaptive polarization channel combination sounding method of claim 3, wherein, In the step S13, the error image is represented as: wherein represents θ i error images at polarization angles; represents θ i intensity images acquired at polarization angles; represents θ i fitted intensity images at polarization angles; The probability density function is represented as: wherein denotes the variance of the grey values of the error image; denotes the mean of the grey values of the error image.

5. The scene-adaptive polarization channel combination sounding method of claim 4, wherein, In the step S14, the Stokes parameter image is represented as: where S fit represents the fitted Stokes vector; represents the elliptical polarization component, which can be usually neglected in a linear polarization imaging detection system.

6. The scene-adaptive polarization channel combination sounding method of claim 5, wherein, According to the Stokes parameter image obtained in step S14, the corresponding intensity image under the polarization angle is represented as:

7. The scene-adaptive polarization channel combination sounding method of claim 6, wherein, The preset second operation procedure in the step S4 is: selecting a detection target region and a background region through the corresponding intensity image under the polarization angle, and calculating statistical parameters target region expectation μ t and background region expectation μ b , target region covariance Σ t , and background region covariance Σ b ; Wherein, the mean μ and the covariance Σ are represented as: where x i represents the gray value at different pixel points in the intensity image; N represents the number of pixels.

8. The scene-adaptive polarization channel combination sounding method of claim 7, wherein, According to the mean and the covariance matrix of the target and the background, the distance JM is taken as an optimization function, which is represented as:

9. The scene-adaptive polarization channel combination sounding method of claim 8, wherein, The joint polarization angle corresponding intensity image and the mean and the covariance matrix of the target and the background are iterated to update different polarization angle images, and the expected polarization channel combination is represented as:

10. The scene-adaptive polarization channel combination sounding method of claim 1, wherein, The method further comprises the step of: S5: driving the rotation mechanism in the rotating wheel type polarization imaging system to the corresponding angle according to the expected polarization channel combination obtained in step S4.

Images