Target detection method under multi-polarization channel combined dynamic background

Abstract

The invention relates to a multi-polarization-channel combined target detection method under a dynamic background, and the method comprises the steps: obtaining a multi-channel polarization information image in a target scene, and calculating a corresponding polarization degree image DoLP; performing threshold segmentation on the polarization degree image to obtain an initial target area image; sequentially selecting every three adjacent polarization information images by combining the difference among the multi-channel polarization information images, calculating a corresponding polarization degree image DoLPk, and performing threshold segmentation on the polarization degree image DoLPk to obtain an initial background region image; performing logical operation on the initial background region result to obtain a final background region image; performing background difference on the initial target area image and the final background area image to obtain a refined target area image; eliminating isolated and filled target areas by combining a filtering algorithm to obtain a final target area image; and finally, image fusion is carried out, target feature extraction is completed, and the purposes of eliminating dynamic background clutter interference and highlighting target features are achieved.

Description

Technical Field

[0001] The present invention relates to the field of imaging detection, and in particular to a target detection method under a dynamic background combined with multiple polarization channels. Background Art

[0002] With the development of camouflage technology and the increasing complexity of combat environments, traditional intensity imaging technology is difficult to achieve effective target detection in complex scenes such as shade, bushes, and real and fake mixed scenes. By utilizing the polarization feature differences of different objects or different states of the same object (roughness, water content, material properties, etc.), and combining spatial image information, radiation feature information, and polarization information that characterizes interface properties, polarization imaging technology can better distinguish between background and target, which is conducive to improving target detection accuracy. Existing polarization target detection methods mainly use polarization feature parameter images (such as Stokes vector, polarization degree, and polarization angle) and fusion algorithms to distinguish between targets and backgrounds. As the key to obtaining polarization state information, the time-sharing polarization imaging system uses a time-series working mode to switch optical elements to obtain polarization images of the same scene target at different angles. It has the advantages of simple structure, easy implementation, low cost, customizability, and richer multi-dimensional information acquisition. However, when detecting static objects, due to the acquisition of different polarization state information during the working process, there are not only changes in the polarization dimension of the target features between image sequences, but also changes in the spatial dimension of the dynamic background. For example, the swaying of leaves and grass will cause serious dynamic background clutter interference, resulting in "pseudo-polarization" in the acquired image, making it difficult to distinguish the background from the target efficiently and accurately. In addition, floating clouds, rotating sun, and moving people in the acquisition scene will also interfere with target detection.

[0003] At present, there is no public research on target detection methods under dynamic background interference based on polarization information at home and abroad. There is no relevant report on how to enhance or suppress the perception of scene information with the help of multi-polarization channel images to more effectively detect targets under dynamic backgrounds. However, there are still some research foundations for segmentation methods based on polarization features and inter-frame information, which provides research ideas for suppressing dynamic background interference: Patent CN201710351870.7 discloses a method and device for detecting moving targets under dynamic backgrounds. According to the position transformation of corresponding pixel points in the previous and next frame images, the phase correlation algorithm and the inter-frame difference method are used to jointly realize target detection under dynamic backgrounds. However, this method is only applicable to the detection of moving targets, and the polarization dimension differences between image sequences have not been considered, so it is difficult to take advantage of polarization to realize target detection under dynamic backgrounds. Patent CN115984299A discloses a polarization image segmentation method based on graph theory. The method uses the joint grayscale distance similarity of polarization feature parameter images as the edge weight between graph vertices, connects all vertices according to the principle of minimizing the sum of edge weights to obtain a minimum spanning tree, and performs regional comparison and merging through inter-class differences and intra-class differences to obtain the final segmentation result. However, the method involves distance calculations between all nodes in the graph, has high computational complexity, and ignores background position changes between polarization channels, making it difficult to eliminate dynamic background clutter interference.

[0004] Therefore, in order to overcome the shortcomings of target detection under dynamic background in polarization imaging, such as large background interference and low detection accuracy, and to better meet the increasingly stringent application requirements of scenarios, how to achieve effective extraction and efficient detection of target features under dynamic background is a technical problem that needs to be solved urgently in this field. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a target detection method under dynamic background by combining multiple polarization channels, which can effectively eliminate dynamic background clutter interference and improve detection accuracy.

[0006] The technical solution adopted by the present invention is a method for detecting targets in a dynamic background by combining multiple polarization channels. The method comprises the following steps: S1, collecting a plurality of polarization information images of a target scene at different polarization angles, and calculating a polarization degree image of the polarization information image according to the polarization imaging principle; S2, using a threshold segmentation algorithm to perform threshold segmentation of the background area and the target area on the polarization image obtained in step S1 to obtain a binary initial target area image; S3, according to the polarization information images obtained in step S1, select three adjacent polarization information images in sequence, and calculate the polarization degree image corresponding to each of the three adjacent polarization information images; S4, using a threshold segmentation algorithm to perform threshold segmentation of the background area and the target area on the polarization image obtained in step S3 to obtain a binary initial background area image; S5, performing an OR logic operation on the initial background area image obtained in step S4 to obtain a binary final background area image; S6, performing background difference operation on the initial target area image obtained in step S2 and the final background area image obtained in step S5 to obtain a binary refined target area image; S7, filtering the refined target area image obtained in step S6 to obtain a final target area image; S8: Using the final target area image obtained in step S7 as a target mask, and fusing it with the polarization degree image obtained in step S1 to obtain a polarization degree image with dynamic background interference eliminated. DoLP t .

[0007] The beneficial effects of the present invention are as follows: the target detection method under dynamic background combined with multiple polarization channels, based on obtaining multiple polarization information images, combines the polarization principle, and uses a threshold segmentation algorithm to realize the initial target area extraction of binarization; then uses the threshold segmentation algorithm and the background difference method to complete the refined target area extraction; then uses the filtering method to eliminate isolated points to realize the final target area extraction; finally, "unraveling the silk" from the cluttered dynamic background to achieve the purpose of eliminating clutter interference and highlighting the target features; the present invention fully considers the polarization characteristics of light waves reflected by different materials and the data distribution differences between polarization channels, and proposes a dynamic background extraction method based on polarization information between adjacent frames, based on the polarization imaging principle and the characteristics of dynamic background changes, to achieve the purpose of weakening target features and highlighting the dynamic background, and finally realizes accurate and effective detection of dynamic background in complex environments; compared with traditional polarization detection algorithms and background segmentation algorithms, the method proposed in the present invention can give full play to the unique target detection advantages of polarization imaging while having stronger dynamic background adaptability, and avoids background clutter interference under the premise of ensuring that the processing process reduces the computing cost, and the target detection results in complex environments are more accurate, which improves the applicability of polarization imaging technology in target detection and other fields.

[0008] Preferably, the specific process of step S1 includes the following steps: S1.1. Collect several polarization information images using a polarization camera. The polarization camera is a time-sharing imaging system. A polarizer that can be rotated at any angle is integrated in its optical path. By rotating the polarizer to multiple angles, more than three polarization information images with equally spaced angles are obtained. S1.2. Calculate the polarization intensity value of each polarization information image based on the polarization imaging principle , specifically expressed as: ;in, represents the polarization angle; S 0 , S 1 , S 2 All represent Stokes parameter images; S1.3, according to the polarization intensity value of the polarization information image obtained in step S1.2, a linear equation system is constructed to solve the Stokes parameter image, which is specifically expressed as: ; Wherein, N represents N adjacent polarization information images with the same interval angle; S 0 , S 1 , S 2 , S 3 Both represent Stokes parameter images, S 3 Approximately equal to 0; S1.4. Determine the polarization degree image based on the Stokes parameter image obtained in step S1.3 DoLP , specifically expressed as: .

[0009] Preferably, in step S2, the expression of the binary initial target area image is: ; Wherein, Tt represents the threshold value; Represents the binary initial target area image; area as the initial target area.

[0010] Preferably, the specific process of step S3 includes the following steps: S3.1, according to the polarization information image obtained in step S1, select three adjacent polarization information images in sequence , as well as ,in, j ≤ i ≤180°-2* j , j Indicates the minimum interval angle of polarizer rotation; S3.2. Solve the Stokes parameter image corresponding to each of the three adjacent polarization information images, which is specifically expressed as: ; S3.3. Based on the Stokes parameter image obtained in S3.2, solve the corresponding polarization image DoLPk , specifically expressed as: ,in, k =1, …, [(180°-2* j ) / j +1].

[0011] Preferably, in step S4, the expression of the binary initial background area image is: ;in, represents the threshold value; k represents the polarization image DoLP k The corresponding serial number; Represents the binary initial background area image; The area is the initial background area.

[0012] Preferably, in step S5, the expression of the final background area image is: ;in, Represented as the final background area image; Indicates the union of multiple initial background area images; The area is the final background area.

[0013] Preferably, in step S6, the binary refined target area image is specifically expressed as: ;in, Represents the refined target image of binarization; The area is the refined target area.

[0014] Preferably, the specific process of step S7 includes the following steps: S7.1, dilate the refined target area image obtained in step S6 to obtain an image after dilation F 1 , specifically expressed as: ; where ⊕ represents the expansion operator, g1 represents the expansion structure element, , (x, y) represents the image The pixel position coordinates of , (m, n) represent the pixel position coordinates in the structural element g1; S7.2, the image after dilation obtained in step S7.1 F 1 Perform corrosion processing to obtain the image after corrosion processing F 2 , specifically expressed as: ;in, represents the corrosion operator, g 2 represents the corrosion structural element, , ( x , y ) represents an image F 1 Pixel position coordinates, ( m , n ) represents a structural element g 2 The pixel position coordinates in the middle; S7.3, the image after corrosion processing obtained in step S7.2 F 2 Perform flood filling operation to obtain the binary final target area image .

[0015] Preferably, in step S8, the polarization image DoLP t The specific expression is: . BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A flowchart of a target detection method in a dynamic background combined with multiple polarization channels provided by the present invention; Figure 2 A real field experiment scene diagram for camouflaged and obscured target detection provided by an embodiment of the present invention; wherein: Figure 2 (a) is scene 1: a real vehicle target blocked by a small area of ​​trees; Figure 2 (b) is scene 2: a real vehicle target covered by a large area of ​​trees; Figure 2 (c) is scene 3: a scaled-down tank model hidden by grass; Figure 2 (d) is scene 4: a scaled-down tank model with mixed real and fake images; Figure 3 Schematic diagram of Stokes parameter images and polarization degree images corresponding to different scenarios provided by an embodiment of the present invention; Figure 4 Schematic diagram of initial target area images corresponding to different scenarios provided by an embodiment of the present invention; Figure 5 A schematic diagram of adjacent polarization channels selected under a scene and a schematic diagram of corresponding Stokes parameter images and polarization degree images provided in an embodiment of the present invention; Figure 6 Schematic diagram of intermediate calculation result images of target detection corresponding to different scenarios provided by the embodiment of the present invention; Figure 7A schematic diagram for comparing target detection results using different methods provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0017] In order to facilitate ordinary technicians in the field to understand and implement the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0018] In order to solve the practical problems of "pseudo-polarization" clutter interference caused by dynamic backgrounds such as "the slightest movement of the grass", which leads to inaccurate measurement, submersion, and unclear identification of target features, the present invention fully considers the differences in polarization characteristics of light waves reflected by different materials and the differences in data distribution between polarization channels, and proposes a target detection method under dynamic background combined with multiple polarization channels, that is, firstly, a polarization feature parameter image containing background interference is calculated on the basis of obtaining multiple polarization channel images to achieve preliminary extraction of target features; then, each three adjacent polarization images are selected in turn for background extraction, and the feature results are logically operated to obtain a binary mask of the dynamic background; then, background subtraction and morphological filtering are combined to remove background interference and eliminate isolated points to obtain a binary mask of the target, and finally extract the target features; finally, image fusion is performed to "unravel" the cluttered dynamic background to achieve the purpose of eliminating clutter interference and highlighting target features.

[0019] In the embodiment of the present invention, four real field experimental scenes for camouflage obscuration target detection are arranged, namely: Scene 1: Real vehicle target under small area obscuration of trees (see Figure 2 (a) in the figure); Scene 2: A real vehicle target covered by a large area of ​​trees (see Figure 2 (b) in the figure); Scene 3: A scaled tank model hidden by the bush (see Figure 2 (c) in the figure); Scene 4: A scaled tank model with mixed real and fake images (see Figure 2(d) in the figure); wherein, dynamic background interference factors include swaying branches, swinging leaves, and swaying grass. The imaging camera used is a developed dynamic rotation tuned polarization imaging camera, which can realize the acquisition of 180 polarization channel information with a minimum interval of 1° in the range of 0°~180°. The camera used to acquire images is a CCD black-and-white industrial camera with an imaging resolution of 1280×960 and a spectral response range of 380~1050nm; the experimental weather conditions are sunny and cloudy, and the temperature is 16.5℃. During the field polarization information acquisition experiment, the target remains stationary, and environmental factors such as wind speed will cause dynamic changes in the background such as branches, leaves, and grass, which will lead to errors in the corresponding spatial positions between the collected sequence images, thereby affecting the reconstructed image quality and target detection accuracy. So far, the present invention proposes a method for target detection under dynamic background combining multiple polarization channels. The flow chart is as follows. Figure 1 As shown, the following steps are included: Step 1: Collect a number of polarization information images of the target scene at different polarization angles, and calculate polarization characteristic parameter images of the polarization information images according to the polarization imaging principle, wherein the polarization characteristic parameter images include Stokes parameter images and polarization degree images; Furthermore, the specific implementation process of step one is: S11: A polarization information image is collected using a developed dynamic rotation tuning polarization imaging camera. The polarization imaging camera is a time-sharing imaging system, in which a polarizer that can be rotated at any angle is integrated in its optical path. By rotating the polarizer to 36 angle positions such as 0°, 5°, ..., 170°, and 175°, 36 polarization information images at different angles are obtained; S12: Calculate the Stokes parameter image based on the polarization information image collected in step S11, ignore the circular polarization component, and calculate the polarization intensity value of the collected polarization information image. With polarization angle θ i The relationship is: ; in, S 0 , S 1 , S 2 All represent Stokes parameter images; By collecting 36 polarization information images at different angles, the polarization intensity value can be With polarization angle θ i The relationship is used to construct a linear equation system to solve the Stokes parameter image; Specifically, when 36 polarization information images with adjacent directions of 0°, 5°, ..., 170°, and 175° and the same interval angle are collected, the corresponding Stokes parameter image can be obtained as follows: ; in, ; S13: Based on the Stokes parameter image obtained in step S12, the polarization degree image can be obtained: .

[0020] According to the above steps, if Figure 3 As shown, the Stokes parameter images of the embodiment of the present invention in different scenes are shown. S 0 , Stokes parameter image S 1 , Stokes parameter image S 2 and polarization image DoLP ;from Figure 3 It can be seen from the figure that due to the differences in the polarization characteristics of light waves reflected by different materials, the polarization image DoLP In the image, man-made targets are clearly distinguished from the natural background; however, due to the dynamic changes in the background such as branches, leaves, and grass, the polarization image DoLP The target features in the scene are seriously interfered by the background clutter, and even the target is completely submerged (Scenes 2, 3, and 4), making it difficult to achieve accurate detection.

[0021] Step 2: Using a threshold segmentation algorithm to perform threshold segmentation of the background area and the target area on the polarization image obtained in step 1, so as to obtain a binary initial target area image; Furthermore, the specific implementation process of step 2 is: The polarization images of the four different scenes calculated in step S13 are DoLP The OTSU maximum inter-class variance adaptive threshold segmentation algorithm is used to segment the background area and the target area to obtain a binary initial target area image. , which can be expressed as: ; in, T t Represents the threshold, and the area with a logical value of 1 is the initial target area.

[0022] The threshold segmentation algorithm can be a manually selected threshold segmentation method, or an OTSU maximum inter-class variance, a threshold segmentation based on maximum entropy, and an iterative threshold segmentation method that does not require human intervention and can automatically select a suitable threshold according to image features.

[0023] In the embodiment of the present invention, the initial target area images in different scenes are as follows: Figure 4 As shown. Figure 4 It can be seen that the binary initial target area image contains not only the real target, but also the dynamic background clutter caused by "slight movement", which will greatly interfere with target detection.

[0024] Step 3: According to the polarization information image obtained in step 1, three adjacent polarization information images are selected in turn by utilizing the correlation between polarization channels, and the corresponding polarization characteristic parameter images are calculated according to the polarization imaging principle. The polarization characteristic parameter images specifically include a Stokes parameter image and a polarization degree image. Further, see Figure 5 (Using scenario 1 as an example, other scenarios are the same). The specific implementation process of step 3 is as follows: S31: Based on the polarization information images of the multiple polarization channels obtained in step 1, three adjacent images are selected in sequence: , as well as ,in, j ≤ i ≤180°-2* j , j =5° is the minimum interval angle of the intensity image set of multiple polarization channels; S32: Solve the Stokes parameter images corresponding to the three adjacent polarization information images: ; S33: Based on the Stokes parameter image obtained in step S32, the polarization degree image is obtained by solving DoLP k , specifically expressed as: ,in, k =1,2 …, 35,36.

[0025] Step 4: Using a threshold segmentation algorithm to perform threshold segmentation of the background area and the target area on the polarization image obtained in step 3, to obtain a binary initial background area image; See also Figure 6 , the threshold segmentation algorithm is used to implement the threshold segmentation of the background area and the target area on the polarization image obtained in step 3, and a binary initial background image is obtained, in which the area with a logical value of 1 is the initial background area.

[0026] Furthermore, the specific implementation process of step 4 is as follows: Step S33 calculates the polarization degree image DoLP kThe OTSU maximum inter-class variance adaptive threshold segmentation algorithm is used to segment the background area and the target area to obtain a binary initial background image. , expressed as: ; in, represents the threshold value, The area is the initial background area.

[0027] The threshold segmentation algorithm can be a manually selected threshold segmentation method, or an OTSU maximum inter-class variance, a threshold segmentation based on maximum entropy, and an iterative threshold segmentation method that does not require human intervention and can automatically select a suitable threshold according to image features.

[0028] Step 5: Perform an OR logic operation on the initial background area image obtained in step 4 to obtain a binary final background area image; Furthermore, the specific implementation process of step five is: The initial background area image calculated in step 4 Perform a logical operation of "OR" to obtain the final binary background image. ,Right now: ; in, represents the union of multiple initial background area images, The area is the final background area.

[0029] In the embodiment of the present invention, the final background area images in different scenes are as follows: Figure 6 As shown. Figure 6 It can be seen that the final background area image after binarization contains all the clutter interference caused by the dynamic background position error during the acquisition of the polarization image, while the real target has been fully suppressed, and the target area and the background area are clearly divided, which lays the foundation for the technology proposed in the present invention to remove dynamic background interference based on multiple polarization channels.

[0030] Step 6: Perform background difference operation on the initial target area image obtained in step 2 and the final background area image obtained in step 5 to obtain a binary refined target area image; Furthermore, the specific implementation process of step six is: The initial target area image obtained in step S21 and the final background area image obtained in step 5 Perform background difference operation to obtain a binary refined target image ,Right now: ; in, The area is the refined target area.

[0031] In the embodiment of the present invention, the refined target area in different scenes refers to Figure 6 As shown. Figure 6 It can be seen that most of the dynamic background clutter is eliminated, and the target position and edge contour are clear; although some target features are lost (scenario 1), most of the important information is retained in the edge feature points, and complete target detection can be achieved by combining with subsequent steps; in addition, although some background clutter remains in the image, compared with the target, they are all outliers and isolated pixels.

[0032] Step 7: The binary target image obtained in step 6 is processed by morphological filtering algorithms such as dilation and erosion to remove background pixels and fill isolated internal pixels in the target area, thereby detecting the complete target and obtaining the final target area image; Furthermore, the specific implementation process of step seven is: S71: Performing dilation processing in morphological operation on the binary refined target area image obtained in step 6 to obtain an image after dilation processing F 1 ,Right now: ; Among them, ⊕ represents the expansion operator, g 1 represents the dilated structural element, , ( x , y ) represents an image The pixel position coordinates in , (m, n) represent the pixel position coordinates in the structural element g1; After the expansion process in step S71, the boundary of the target area can be expanded so that the same broken area can be bridged; S72: dilate the image after step S71 F 1 Perform corrosion processing in morphological operation to obtain the image after corrosion processing F 2 ,Right now: ; in, represents the corrosion operator, g 2 represents the corrosion structural element, , ( x , y ) represents the pixel position coordinates in image F1, (m, n) represents the pixel position coordinates in structure element g2; The remaining noise points can be eliminated by the corrosion process in step S72; S73: Finally, the image after the corrosion process obtained in step S72 F 2 Through flood filling operation, isolated internal pixels (0 pixels surrounded by 1) are filled to obtain the binary final target area image. .

[0033] In the embodiment of the present invention, the final target area images in different scenes refer to Figure 6 As shown, from Figure 6 It can be seen that isolated background pixels have been removed well and relatively complete real target features have been retained.

[0034] Step 8: Use the final target area image obtained in step 7 as the target mask and fuse it with the polarization image obtained in step 1 to obtain a polarization image that eliminates dynamic background interference. DoLP t .

[0035] Furthermore, the specific implementation process of step eight is: The final target area image obtained by the binarization in step S73 is a mask, and the polarization degree obtained in step S13 is DoLP By fusing the images, we can get a polarization image that eliminates dynamic background interference. DoLP t ,Right now: ; In the embodiment of the present invention, the final detection result images in different scenes are as follows: Figure 7 As shown, from Figure 7 It can be seen that compared with the traditional DoLP The method proposed in the present invention can solve the problems of large background interference and low detection accuracy in target detection under dynamic background, and can realize accurate and effective detection of targets under dynamic background.

[0036] In addition, in order to verify the effectiveness and advancement of the target detection method under dynamic background combined with multi-polarization channels proposed in the present invention, the traditional inter-frame difference method is used as a comparison. Figure 1 The scenes are detected separately.

[0037] like Figure 7 As shown in Figure 2, with the help of the differences between images of different polarization channels, the inter-frame difference method can detect targets in scenes such as occlusion and mixing, and compared with images DoLPIt can fully suppress the interference of dynamic background clutter, but it is sensitive to the dynamically changing background and the target with changing light intensity, and it is difficult to completely eliminate samples of no interest and thus achieve effective segmentation of the target and background. The method proposed in the present invention has good target detection performance under dynamic background interference. While effectively separating the background from the target, it can effectively ensure the detection of target details and edge information, and achieve the purpose of "unraveling the mystery" from the cluttered dynamic background, eliminating clutter interference and highlighting the target features. The method proposed in the present invention solves the practical problems of "pseudo-polarization" clutter interference caused by dynamic backgrounds such as mountain jungles that "sway in the slightest breeze", which leads to inaccurate measurement, submersion, and unclear identification of target features.

[0038] Although the disclosure is disclosed as above, the protection scope of the disclosure is not limited thereto. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the disclosure, and these changes and modifications will fall within the protection scope of the present invention.

Claims

1. A method for target detection in a dynamic background using multiple polarization channels, characterized in that: The method comprises the following steps: S1, collecting a plurality of polarization information images of a target scene at different polarization angles, and calculating a polarization degree image of the polarization information image according to the polarization imaging principle; S2, using a threshold segmentation algorithm to perform threshold segmentation of the background area and the target area on the polarization image obtained in step S1 to obtain a binary initial target area image; S3, according to the polarization information images obtained in step S1, select three adjacent polarization information images in sequence, and calculate the polarization degree image corresponding to each of the three adjacent polarization information images; S4, using a threshold segmentation algorithm to perform threshold segmentation of the background area and the target area on the polarization image obtained in step S3 to obtain a binary initial background area image; S5, performing an OR logic operation on the initial background area image obtained in step S4 to obtain a binary final background area image; S6, performing background difference operation on the initial target area image obtained in step S2 and the final background area image obtained in step S5 to obtain a binary refined target area image; S7, filtering the refined target area image obtained in step S6 to obtain a final target area image; S8: Using the final target area image obtained in step S7 as a target mask, and fusing it with the polarization degree image obtained in step S1 to obtain a polarization degree image with dynamic background interference eliminated. DoLP t .

2. The method for detecting targets in a dynamic background by combining multiple polarization channels according to claim 1, characterized in that: The specific process of step S1 includes the following steps: S1.

1. Collect several polarization information images using a polarization camera. The polarization camera is a time-sharing imaging system. A polarizer that can be rotated at any angle is integrated in its optical path. By rotating the polarizer to multiple angles, more than three polarization information images with equally spaced angles are obtained. S1.

2. Calculate the polarization intensity value of each polarization information image based on the polarization imaging principle , specifically expressed as: ； in, θ i represents the polarization angle; S 0. S 1 , S 2 denotes Stokes parameter images; S1.3, according to the polarization intensity value of the polarization information image obtained in step S1.2, a linear equation system is constructed to solve the Stokes parameter image, which is specifically expressed as: ; Wherein, N represents N adjacent polarization information images with the same interval angle; S 0. S 1 , S 2. S 3 represent Stokes parameter images, S 3 is approximately equal to 0; S1.

4. Determine the polarization degree image based on the Stokes parameter image obtained in step S1.3 DoLP , specifically expressed as: 。 3. The method for target detection in a dynamic background using multiple polarization channels according to claim 2, characterized in that: In step S2, the expression of the binary initial target area image is: ； Wherein, Tt represents the threshold value; Represents the binary initial target area image; area as the initial target area.

4. The method for detecting targets in a dynamic background by combining multiple polarization channels according to claim 3, characterized in that: The specific process of step S3 includes the following steps: S3.1, according to the polarization information image obtained in step S1, select three adjacent polarization information images in sequence , as well as ,in, , j Indicates the minimum interval angle of polarizer rotation; S3.

2. Solve the Stokes parameter image corresponding to each of the three adjacent polarization information images, which is specifically expressed as: ； S3.

3. Based on the Stokes parameter image obtained in S3.2, solve the corresponding polarization image DoLP k , specifically expressed as: ； in, .

5. The method for detecting targets in a dynamic background by combining multiple polarization channels according to claim 4, characterized in that: In step S4, the expression of the binary initial background area image is: ； in, represents the threshold value; k represents the polarization image DoLP k The corresponding serial number; Represents the binary initial background area image; The area is the initial background area.

6. The method for detecting targets in a dynamic background by combining multiple polarization channels according to claim 5, characterized in that: In step S5, the expression of the final background area image is: ； in, Represented as the final background area image; Indicates the union of multiple initial background area images; The area is the final background area.

7. The method for detecting targets in a dynamic background by combining multiple polarization channels according to claim 6, characterized in that: In step S6, the binary refined target area image is specifically expressed as: ； in, Represents the refined target image of binarization; The area is the refined target area.

8. The method for detecting targets in a dynamic background by combining multiple polarization channels according to claim 7, characterized in that: The specific process of step S7 includes the following steps: S7.1, dilate the refined target area image obtained in step S6 to obtain an image after dilation F 1, specifically expressed as: ； Among them, ⊕ represents the expansion operator, g1 represents the expansion structure element, , (x, y) represents the image The pixel position coordinates of , (m, n) represent the pixel position coordinates in the structural element g1; S7.2, the image after dilation obtained in step S7.1 F 1 Perform corrosion processing to obtain the image after corrosion processing F 2, specifically expressed as: ； in, represents the corrosion operator, g 2 indicates corrosion structural elements, , ( x , y ) represents an image F 1 pixel position coordinate, ( m , n ) represents a structural element g 2. Pixel position coordinates; S7.3, the image after corrosion processing obtained in step S7.2 F 2 Perform flood filling operation to obtain the binary final target area image .

9. The method for detecting targets in a dynamic background by combining multiple polarization channels according to claim 8, characterized in that: In step S8, the polarization image DoLP t The specific expression is: 。

Images