Air infrared small target detection method based on target polarization characteristic reconstruction

Abstract

The invention discloses an aerial infrared small target detection method based on target polarization characteristic reconstruction, which belongs to the technical field of infrared target detection, and comprises the following steps: S1, obtaining polarization images of three channels of 0 degree, 60 degrees and 120 degrees; s2, calculating corresponding Stokes vectors according to the polarization images of the three channels; s3, calculating gradient features of a suppression edge according to the Stokes vector, and performing saliency feature extraction; s4, obtaining a gradient feature of a final suppression background edge; s5, obtaining reconstructed polarization characteristics; s6, calculating a local variance stability level; and S7, obtaining a weighting function, and obtaining a final detection result. According to the aerial infrared small target detection method based on target polarization characteristic reconstruction, small target detection can be accurately achieved, the background clutter suppression efficiency is improved, and good detection performance is achieved under the simple sky background and the complex sky background.

Description

Technical Field

[0001] The invention relates to the technical field of infrared target detection, and in particular to a method for detecting small infrared targets in the air based on reconstruction of target polarization characteristics. Background Art

[0002] At present, the detection methods of small targets under clouds can be divided into two categories: detection methods based on single-frame images and detection methods based on frame sequences. The frame sequence-based methods use the state and motion trajectory of the target to achieve target detection. The more advanced multi-frame image detection methods currently include Markov random fields, spatiotemporal filtering, and spatiotemporal contrast filtering. These methods perform well in removing complex background interference, but usually rely on prior information of the target or background, and require a certain correlation between frame sequences.

[0003] The target detection method based on infrared polarization imaging mainly distinguishes the difference between the target and the background through the linear polarization degree and the polarization angle, and highlights the characteristics of the target to be detected. For example, Forssell, Y. Gronau, etc. respectively used the linear polarization degree to establish models to realize field mine detection and car detection; Zhao Yongqiang, etc. realized vehicle target detection under low signal-to-noise ratio and complex background by designing the polarization weighted local contrast method; Ratlif, etc. verified that infrared polarization imaging has stronger detection ability for targets under complex sky background; Hao, etc. combined polarization and Top-Hat transformation to realize the detection of weak targets in the air; Song Minmin, etc. conducted a large number of experiments to verify that infrared polarization imaging has a significant effect in suppressing sea surface clutter.

[0004] Although infrared imaging can effectively detect small targets against a sky background, it still faces many difficulties and challenges when the contrast of the target neighborhood in the scene is low. Polarization information can describe the physical and chemical properties of the target and the background. Even in low-contrast scenes, it can still reflect the difference in polarization information between the target and the background, thereby improving the contrast between the target and the background and providing a new idea for detecting small targets against a complex sky background. At the same time, although infrared polarization imaging technology improves the local contrast of the target neighborhood, it also improves the contrast of the background highlight edges, which brings new difficulties and challenges to target detection. Summary of the invention

[0005] The purpose of the present invention is to provide an aerial infrared small target detection method based on target polarization feature reconstruction, which can accurately realize the detection of small targets, improve the efficiency of suppressing background clutter, and have good detection performance under both simple sky background and complex sky background.

[0006] To achieve the above object, the present invention provides a method for detecting small infrared targets in the air based on target polarization feature reconstruction, comprising the following steps:

[0007] Step S1, obtaining polarization images of three channels of 0°, 60°, and 120°;

[0008] Step S2, calculating the corresponding Stokes vector based on the polarization images of the three channels;

[0009] Step S3, calculating the gradient features of the suppressed edge based on the Stokes vector, and performing significant feature extraction;

[0010] Step S4, obtaining the gradient features for finally suppressing the background edge according to step S3;

[0011] Step S5, obtaining reconstructed polarization characteristics;

[0012] Step S6, calculating the local variance stability level;

[0013] Step S7: Obtain a weight function to obtain a final detection result.

[0014] Preferably, the formula for calculating the gradient feature of the suppressed edge based on the Stokes vector in step S3 is as follows:

[0015]

[0016] Wherein, Gm(I)(x,y) represents the gradient feature of the Stokes vector calculation suppression edge, and I represents the Stokes vector S1 or the Stokes vector S2; Respectively represent (x,y) along α i , β i Directional derivative of direction; where α i , β i Satisfy β i =α i +π / 2.

[0017] Preferably, the significant feature extraction formula in step S3 is as follows:

[0018]

[0019] Among them, SP(I) represents the significant feature; ||·||2 represents the 2-norm; mean ω (I) represents the mean value of ω in the local area of ​​the image; Represents an image smoothed by a Gaussian filter with an area ω as the size; the size of ω is 5×5, and the standard deviation σ of the Gaussian filter is 1 to 3.

[0020] Preferably, the salient features obtained in step S3 are used to calculate the gradient weights of the Stokes vectors S1 and S2 using formula (1) to obtain the final gradient feature image GT that suppresses the background edge:

[0021]

[0022] Preferably, by suppressing the gradient feature image GT of the background edge, the suppression PF of the background polarization feature is achieved, which is defined as:

[0023] PF = GT·S1 + GT·S2 (4);

[0024] The calculation formula of the reconstructed polarization characteristic PF is as follows:

[0025]

[0026] Among them, RDoLP represents the reconstructed polarization characteristics; S0 represents the first component in the Stokes vector, which represents the intensity image of the scene.

[0027] Preferably, the local variance stability level is calculated by the local stability of variance VSL, which is defined as:

[0028]

[0029] in, Indicates the local variance stability level; i indicates the pixel number in the nested window, i = 0 to 8, i = 0 indicates the inner window of the nested structure, and i = 1-8 indicates the outer window of the nested structure; represents the set of all pixels in the pixel numbered i in the nested window at the jth scale; Represents the set of the first N largest pixels in the pixel.

[0030] Preferably, for any window, the weight W of the center image is j (x,y) is represented as:

[0031]

[0032] in, Indicates the variance stability level of the inner window; Represents the standard deviation of VSL between pixels in the outer window; ζ is a positive number, and its value is 0.5; TED j It represents the minimum value of the variance stability level difference between the central pixel and the surrounding pixels, which is defined as:

[0033]

[0034] Then the final target saliency image SupD in the final detection result is expressed as:

[0035] SupD=max{W j ·RDoLP}(9).

[0036] Therefore, the present invention adopts the above-mentioned aerial infrared small target detection method based on target polarization feature reconstruction, and analyzes the difference in polarization characteristics between the target and the background by analyzing a large amount of sky background infrared polarization data. Specifically, the polarization feature images S1 and S2 of the target area are both isotropic, while the gradient features of the background edge area are anisotropic. Therefore, this characteristic is used to constrain the polarization Stokes parameters, further reconstruct the polarization characteristics of the scene, effectively suppress the background information, and highlight the target.

[0037] At the same time, this method realizes the design of weight function by defining the variance stability level, and realizes small target detection under complex sky background. Compared with other methods, the signal-to-clutter ratio gain (SCRG) of the detection result is improved by about 40%; in different types of scenes (large continuous clouds, complex clouds), the experimental results show that this method has better robustness.

[0038] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a flow chart of an embodiment of a method for detecting small infrared targets in the air based on reconstruction of target polarization characteristics of the present invention;

[0040] Figure 2 1 is an original image (left) and a three-dimensional intensity distribution diagram (right) of scene 1 in an embodiment of the present invention;

[0041] Figure 3 1 is the detection result of scene 1 in an embodiment of the present invention (left) and its three-dimensional intensity distribution diagram (right);

[0042] Figure 4 1 is an original image (left) and a three-dimensional intensity distribution diagram (right) of scene 2 in an embodiment of the present invention;

[0043] Figure 5 2 is the detection result of scene 2 (left) and its three-dimensional intensity distribution diagram (right) in an embodiment of the present invention.

[0044] Figure 6 is a schematic diagram of the nested structure in steps S6 and S7 of an embodiment of the present invention, Figure 6 (a) in the figure shows the reconstructed polarization characteristics; Figure 6 (b) in the figure represents the residual edge window; Figure 6 (c) in the figure represents the target area window. DETAILED DESCRIPTION

[0045] The technical solution of the present invention is further described below through the accompanying drawings and embodiments.

[0046] Unless otherwise defined, technical or scientific terms used in the present invention shall have the common meanings understood by one having ordinary skills in the field to which the present invention belongs.

[0047] Embodiment 1

[0048] like Figure 1 As shown, the present invention provides a method for detecting small infrared targets in the air based on reconstruction of target polarization characteristics, comprising the following steps:

[0049] Step S1, obtaining polarization images of three channels of 0°, 60°, and 120°;

[0050] Step S2, calculating the corresponding Stokes vector based on the polarization images of the three channels;

[0051] Step S3, calculating the gradient features of the suppressed edge based on the Stokes vector, and performing significant feature extraction;

[0052] Step S4, obtaining the gradient features for finally suppressing the background edge according to step S3;

[0053] Step S5, obtaining reconstructed polarization characteristics;

[0054] Step S6, calculating the local variance stability level;

[0055] Step S7: Obtain a weight function to obtain a final detection result.

[0056] Preferably, the formula for calculating the gradient feature of the suppressed edge based on the Stokes vector in step S3 is as follows:

[0057]

[0058] Wherein, Gm(I)(x,y) represents the gradient feature of the Stokes vector calculation suppression edge, and I represents the Stokes vector S1 or the Stokes vector S2; Respectively represent (x,y) along α i , β i Directional derivative of direction; where α i , β i Satisfy β i =α i +π / 2.

[0059] Directional derivatives can characterize the rate of change of image pixels along a certain direction. Therefore, in the background area with large continuous distribution, the modulus of the directional derivatives in each direction is a very small positive value. At the edge of the cloud background, the modulus of the directional derivative is large only in the direction of gradient change and the opposite direction. In the target area, the modulus of the directional derivative is greater than 0 in all directions and is evenly distributed. Therefore, the gradient obtained by formula (1) is a high gradient in the target area. At the background edge, since each set of orthogonal directions must have a very small directional derivative in one direction, it is a low gradient.

[0060] The significant feature extraction formula in step S3 is as follows:

[0061]

[0062] Among them, SP(I) represents the significant feature; ||·||2 represents the 2-norm; mean ω (I) represents the mean value of ω in the local area of ​​the image; Represents an image smoothed by a Gaussian filter with an area ω as the size; the size of ω is 5×5, and the standard deviation σ of the Gaussian filter is 1 to 3.

[0063] In step S4, the salient features obtained in step S3 are used to calculate the gradient weights of Stokes vectors S1 and S2 using formula (1) to obtain the final gradient feature image GT that suppresses the background edge:

[0064]

[0065] In step S5, the background polarization feature suppression PF is achieved by suppressing the gradient feature image GT of the background edge, which is defined as:

[0066] PF = GT·S1 + GT·S2 (4);

[0067] The calculation formula of the reconstructed polarization characteristic PF is as follows:

[0068]

[0069] Among them, RDoLP represents the reconstructed polarization characteristics; S0 represents the first component in the Stokes vector, which represents the intensity image of the scene.

[0070] Compared with the original linear polarization image, the reconstructed polarization features suppress a large number of non-target features and only retain the polarization features of the target and a small amount of residual background edge features.

[0071] In step S6, since the reconstructed polarization feature RDolp suppresses most background edges, there are still some weak edges and sparse noise that may affect the accuracy of target detection. Therefore, a two-layer nested window optimization weighting function is introduced to further suppress the weak edge background while retaining the real target. In order to accurately estimate the weight, the local variance stability level is calculated by the local stability of the variance VSL, which is defined as:

[0072]

[0073] in, Indicates the local variance stability level; i indicates the pixel number in the nested window, i = 0 to 8, i = 0 indicates the inner window of the nested structure, and i = 1-8 indicates the outer window of the nested structure; represents the set of all pixels in the pixel numbered i in the nested window at the jth scale; Represents a pixel (i.e. Figure 6 The set of the top N largest pixels in the image (a total of nine pixels numbered 0-8).

[0074] Considering that the size of small objects in actual scenes cannot be predicted in advance, the algorithm sets 5 pixels of different scales to cover most of the small object sizes, namely 5×5, 7×7, 9×9, 11×11, and 13×13. The two-layer nested sliding window consists of 3×3 pixels, which are divided into central pixels (numbered 0) and surrounding pixels (numbered 1-8), as shown in Figure 1. Figure 6 As shown, each pixel contains N×N pixels, and N should be slightly larger than the size of the target.

[0075] In step S7, for any window, the weight W of the center image is j (x,y) is represented as:

[0076]

[0077] in, Indicates the variance stability level of the inner window; Represents the standard deviation of VSL between pixels in the outer window; in order to prevent the denominator from being zero, ζ is set to a very small positive number, whose value is 0.5; TED j It represents the minimum value of the variance stability level difference between the central pixel and the surrounding pixels, which is defined as:

[0078]

[0079] The weight obtained by formula (8) has the following characteristics: First, when the center of the inner window is the true target, The outer window pixels are the background area, and the pixel distribution is relatively stable, which satisfies Smaller, TED j is large, so the weight of the central pixel is large; when the center of the inner window is located at the continuous background, is approximately equal to 0, and TED j It must be a finite positive value. Therefore, no matter whether all pixels in the outer window are in the continuous background, the weight in the continuous background is close to 0. Finally, when the center of the inner window is at the edge of the background, due to the continuity of the edge, some pixels in the surrounding pixels may have VSL that satisfies TED j is a positive value close to 0. is also a finite value, that is, the background edge weight value is extremely small. In summary, the weight function defined based on the polarization local variance characteristic can effectively suppress the residual background of the image and highlight the real target.

[0080] The final target saliency image SupD in the final detection result is expressed as:

[0081] SupD=max{W j ·RDoLP}(9).

[0082] Figure 2 , Figure 3 Scene 1 in the video shows a simple cloud background. Figure 4 , Figure 5 Scene 2 in Figure 1 shows the target detection effect against a complex cloud background. Figure 2 , Figure 3 The three-dimensional intensity distribution in Figure 4 , Figure 5 From the three-dimensional intensity distribution diagram, we can see that Figure 3 , Figure 5 The target information in the image is more obvious, and the noisy noise can be effectively suppressed.

[0083] The present invention can not only effectively suppress noise and extract target information under simple backgrounds, but also effectively highlight targets under complex backgrounds and low-contrast scenes, greatly reducing background residues.

[0084] Therefore, the present invention adopts the above-mentioned aerial infrared small target detection method based on target polarization feature reconstruction to accurately realize the detection of small targets, improve the efficiency of suppressing background clutter, and have good detection performance under both simple sky background and complex sky background.

[0085] It is worth noting that the contents not elaborated in detail in the present invention are all prior art and are well known to those skilled in the art.

[0086] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solution of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solution to deviate from the spirit and scope of the technical solution of the present invention.

Claims

1. A method for detecting small infrared targets in the air based on reconstruction of target polarization characteristics, characterized in that: The following steps are involved: Step S1, obtaining polarization images of three channels of 0°, 60°, and 120°; Step S2, calculating the corresponding Stokes vector according to the polarization images of the three channels; Step S3, calculating the gradient features of the suppressed edge based on the Stokes vector, and performing significant feature extraction; Step S4, obtaining the gradient features for finally suppressing the background edge according to step S3; Step S5, obtaining reconstructed polarization characteristics; Step S6, calculating the local variance stability level; Step S7: Obtain a weight function to obtain a final detection result.

2. The method for detecting small infrared targets in the air based on target polarization feature reconstruction according to claim 1 is characterized in that: The formula for calculating the gradient feature of the suppressed edge based on the Stokes vector in step S3 is as follows: Where Gm(I)(x,y) represents the gradient feature of the Stokes vector calculation suppression edge I represents the Stokes vector S1 or Stokes vector Respectively represent (x,y) along α i , β i Directional derivative of direction; where α i , β i Satisfy β i =α i +π / 2.

3. The method for detecting small infrared targets in the air based on target polarization feature reconstruction according to claim 1, characterized in that: The significant feature extraction formula in step S3 is as follows: Among them, SP(I) represents the significant feature; ||·||2 represents the 2-norm; mean ω (I) represents the mean value of ω in the local area of ​​the image; Represents an image smoothed by a Gaussian filter with an area ω as the size; the size of ω is 5×5, and the standard deviation σ of the Gaussian filter is 1 to 3.

4. The method for detecting small infrared targets in the air based on target polarization feature reconstruction according to claim 3 is characterized in that: The salient features obtained in step S3 are used to calculate the gradient weights of Stokes vectors S1 and S2 using formula (1) to obtain the final gradient feature image GT that suppresses the background edge:

5. The method for detecting small infrared targets in the air based on target polarization feature reconstruction according to claim 4 is characterized in that: By suppressing the gradient feature image GT of the background edge, the suppression PF of the background polarization feature is achieved, which is defined as: PF = GT·S1 + GT·S2 (4); The calculation formula of the reconstructed polarization characteristic PF is as follows: Among them, RDoLP represents the reconstructed polarization characteristics; S0 represents the first component in the Stokes vector, which represents the intensity image of the scene.

6. The method for detecting small infrared targets in the air based on target polarization feature reconstruction according to claim 5, characterized in that: The local variance stability level is calculated by the local stability of variance VSL, which is defined as: in, Indicates the local variance stability level; i indicates the pixel number in the nested window, i = 0 to 8, i = 0 indicates the inner window of the nested structure, and i = 1-8 indicates the outer window of the nested structure; represents the set of all pixels in the pixel numbered i in the nested window at the jth scale; Represents the set of the first N largest pixels in the pixel.

7. The method for detecting small infrared targets in the air based on target polarization feature reconstruction according to claim 6, characterized in that: For any window, the weight W of the center image j (x,y) is represented as: in, Indicates the variance stability level of the inner window; Represents the standard deviation of VSL between pixels in the outer window; ζ is a positive number, and its value is 0.5; TED j It represents the minimum value of the variance stability level difference between the central pixel and the surrounding pixels, which is defined as: Then the final target saliency image SupD in the final detection result is expressed as: SupD=max{W j ·RDoLP} (9)。

Images