Infrared small target detection method and device based on improved three-layer window local contrast measure

Abstract

The application relates to the field of infrared image processing and discloses an infrared small target detection method and device based on an improved three-layer window local contrast measurement. The method comprises the following steps: (1) using a guided filter to achieve the effects of smoothing and edge preservation of an original thermal infrared image; (2) based on a local contrast measurement theory, a sliding window with a three-layer nested structure is constructed, and a local contrast measurement operator is designed; (3) performing edge filling on the thermal infrared image subjected to the guided filter, traversing the image pixel by pixel with the sliding window, calculating the local contrast measurement value of the image block covered by the current sliding window, and obtaining a local contrast measurement matrix C; and (4) regarding the local contrast measurement matrix C as a local contrast image T, taking the local contrast image T as the infrared small target detection result, and realizing infrared small target detection. The application comprehensively utilizes the guided filter and the local contrast measurement theory, improves the local contrast measurement criterion, better represents the characteristics of small targets, and improves the detection performance of thermal infrared small targets.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of image processing, and in particular to an infrared small target detection method and device based on an improved three-layer window local contrast measure. BACKGROUND

[0002] It is well known that thermal infrared small target detection has a wide range of applications in remote sensing, surveillance, aerospace and many other fields. In the process of remote sensing infrared small target detection, due to long distance imaging and interference of complex environment, infrared small targets usually exhibit small size and lack of texture shape information. Worse still, during the imaging process, the size of moving targets may change within a certain range, and due to the low signal-to-noise ratio of the image, infrared small target detection in the field of remote sensing faces great challenges.

[0003] According to different application scenarios, infrared small target detection is mainly divided into small target detection methods based on multi-frame images and single-frame images.

[0004] The infrared small target detection algorithm based on multi-frame images assumes that the motion trajectory of the infrared small target has continuity, and through the motion state information of the current target, a corresponding algorithm is designed to predict the motion state of all potential targets and determine the reliability of the potential trajectory, thereby completing the infrared small target detection in the thermal infrared image sequence. However, the applicability of the small target detection algorithm based on multi-frame images is limited, and it cannot well adapt to scenes with rapid background changes or discontinuous target motion trajectories, and has a large demand for hardware.

[0005] The infrared small target detection algorithm based on single-frame image is mainly divided into background suppression-based detection method, low-rank sparse representation-based detection method and visual contrast mechanism-based detection method. The background suppression-based detection method mainly depends on the brightness distribution characteristics of the background and small target in the image, and designs a filter to filter the infrared image, so as to achieve the purpose of background signal suppression and small target signal enhancement. However, such a method has poor small target detection effect in complex background. The low-rank sparse representation-based method models the target component and background component in the infrared image as a sparse matrix (or sparse tensor) and a low-rank matrix (or low-rank tensor), and through the design of an optimization algorithm, the sparse matrix (tensor) is obtained as much as possible, thereby realizing thermal infrared small target detection. However, overall, the real-time performance of the low-rank sparse representation-based method is poor, which is not conducive to practical application.

[0006] Inspired by the theory of human visual system, researchers have proposed a number of infrared small target detection methods based on visual contrast mechanism. Among them, the method based on local contrast measure shows good detection performance, which takes into account that the thermal infrared small target is more prominent in vision than its surroundings, and enhances the target by designing an effective local contrast measure operator. Chen et al. first designed a local feature representation algorithm called local contrast measure (LCM). Subsequently, a series of variants of LCM were proposed, such as multi-scale relative local contrast measure (RLCM), three-layer window-based local contrast measure (TLLCM), weighted enhanced local contrast measure (WSLCM), and enhanced robust local contrast measure (SRLCM). Although the local contrast measure algorithm has been widely used in thermal infrared target detection, the existing LCM-based methods still lack satisfactory target detection ability or background suppression ability. SUMMARY

[0007] In view of the deficiencies in the prior art, the purpose of the present application is to provide an infrared small target detection method and device based on improved three-layer window local contrast measure, which makes full use of guided filtering, local contrast measure theory and the performance characteristics of thermal infrared small targets. Considering that local contrast measure can highlight areas with strong differences, the local contrast measure criterion is further improved by introducing a local contrast weighting coefficient, which effectively suppresses the background and enhances the target, improves the detection performance of thermal infrared small targets, and realizes infrared small target detection.

[0008] To achieve the above purpose, the present application provides the following technical solutions:

[0009] The present application discloses an infrared small target detection method based on improved three-layer window local contrast measure, characterized in that it comprises the following steps:

[0010] Step 1): using guided filtering to smooth and edge-preserving process the original thermal infrared image;

[0011] Step 2): based on the theory of local contrast measure, a sliding window with improved three-layer nested structure is constructed, and a local contrast measure operator TrLCM is designed;

[0012] Step 3): edge filling is performed on the thermal infrared image after guided filtering, the sliding window is traversed pixel by pixel through the infrared image, the local contrast measure value of the image block covered by the current sliding window is calculated according to the local contrast measure operator TrLCM, and finally the local contrast measure matrix C is obtained;

[0013] Step 4): the local contrast measure matrix C obtained in step 3) is regarded as a local contrast image T, which is used as the result of infrared small target detection, and the detection of infrared small target is realized.

[0014] The application also discloses an infrared small target detection device based on the improved three-layer window-based local contrast measurement method.

[0015] The denoising and edge preserving module based on guided filtering is used for obtaining a denoised and edge preserved image of the original thermal infrared image, and lays a foundation for the local contrast measurement.

[0016] The local contrast measurement window structure design module is used for designing a sliding window with a three-layer nested window, and capturing a local detection image block in the infrared image.

[0017] The local contrast measurement operator design module is used for designing a local contrast measurement mode, and combining the sliding window with the three-layer nested window to carry out the local contrast measurement.

[0018] The target detection module is used for traversing the infrared image pixel by pixel by the sliding window, carrying out the local contrast measurement calculation, and carrying out the infrared small target detection.

[0019] The target detection result output module is used for outputting the infrared small target detection result image.

[0020] The application has the following beneficial effects:

[0021] 1) The application makes full use of the intrinsic characteristics of small targets, improves the traditional local contrast measurement structure, and designs a sliding window with a three-layer nested structure, which contains ten different parts, i.e., a center block, an isolation circle and eight surrounding neighborhood blocks. Considering that the target edge region or mixed background component has a certain influence on the characteristics of small targets, the isolation circle that can capture the target edge information is added, so that the influence of the target edge region or other background components on the local contrast measurement process can be reduced to a certain extent.

[0022] 2) The application fully considers the influence of the background component on the target, and improves the local contrast measurement criterion. The energy performance characteristics of the background region and the target region are comprehensively considered, the gray difference between the center block and the surrounding neighborhood block is introduced, and an adjustable local contrast weighting coefficient is introduced, so that the target detection ability of the detector is enhanced, and the background suppression ability of the detector is further improved. DETAILED DESCRIPTION

[0023] Figure 1 A local contrast measurement window structure diagram for carrying out the thermal infrared image small target detection of the application;

[0024] Figure 2 A structure diagram of the infrared small target detection device of the application;

[0025] Figure 3 An example frame image of thermal infrared image sequence one for experimental test;

[0026] Figure 4 An example frame image of thermal infrared image sequence two for experimental test;

[0027] Figure 5 Thermal infrared small target detection result of an example frame image of thermal infrared image sequence one;

[0028] Figure 6 Thermal infrared small target detection result of an example frame image of thermal infrared image sequence two;

[0029] Figure 7 Detection result image and corresponding three-dimensional gray visualization image of an example frame image of thermal infrared image sequence one after detection by the proposed method, LCM and TLLCM;

[0030] Figure 8 Detection result image and corresponding three-dimensional gray visualization image of an example frame image of thermal infrared image sequence one after detection by IPI and PSTNN;

[0031] Figure 9 Detection result image and corresponding three-dimensional gray visualization image of an example frame image of thermal infrared image sequence two after detection by the proposed method, LCM and TLLCM;

[0032] Figure 10 Detection result image and corresponding three-dimensional gray visualization image of an example frame image of thermal infrared image sequence two after detection by IPI and PSTNN. DETAILED DESCRIPTION

[0033] In order to make the objects, technical solutions and advantages of the present application clearer, the present application will be described in detail below with specific examples. The specific examples are described below to simplify the present application. However, it should be recognized that the present application is not limited to the described examples, and various modifications of the present application are possible without departing from the basic principles, and these equivalent forms also fall within the scope defined by the appended claims of the present application.

[0034] The technical solutions in the embodiments of the present application will be described clearly and completely below with the accompanying drawings in the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0035] The basic steps of the infrared small target detection method based on improved three-layer window local contrast measure of the present application mainly include:

[0036] Step 1: Use guided filtering to achieve smoothing and edge preservation of the original thermal infrared image;

[0037] In this embodiment, the guided filtering parameters used are as follows: the local window radius r = 2 and the regularization parameter eps = 0.001 are set to obtain a smooth and edge-preserving thermal infrared image f with a size of u×v.

[0038] Step 2: Based on the theory of local contrast measurement, construct an improved three-layer nested sliding window and design a local contrast measurement operator;

[0039] Specifically, a novel three-layer nested sliding window is designed, which contains ten different parts: a central block, an isolation ring, and eight surrounding neighboring blocks. The central block is represented by M0, and the eight surrounding neighboring blocks are represented by M1, M2, ..., M8.

[0040] The average gray value m0 of the center block M0 is

[0041]

[0042] Where N0 represents the side length scale of the central block M0, (i,j)∈M0 means that the pixel (i,j) belongs to the central block M0, and I0(i,j) represents the gray value of the pixel located at position (i,j) on the central block M0.

[0043] The t-th surrounding neighboring block M t grayscale mean m t for

[0044]

[0045] Where, N t Represents the t-th surrounding neighbor block M t The side length scale, (i,j)∈M t This indicates that pixel (i,j) belongs to the t-th surrounding neighborhood block M. t I t (i,j) represents the t-th surrounding neighbor block M. t The gray value of the pixel located at position (i,j), where t takes values ​​of 1, 2, ..., 8.

[0046] By designing a local contrast weighting coefficient γ, the background suppression and target detection capabilities of the local contrast operator can be further improved.

[0047]

[0048] Where m0 represents the average gray value of the center block M0, max(m t), t = 1, 2,..., 8 represents the maximum value of the average gray mean of eight surrounding neighborhood blocks, n and β are adjustment coefficients, which are set to 3 and 1.005 respectively.

[0049] Finally, the improved three-layer window-based local contrast measure operator TrLCM is obtained as

[0050]

[0051] Step 3: Perform edge padding on the guided-filtered thermal infrared image, traverse the infrared image pixel by pixel with a sliding window, calculate the local contrast measure value of the image block covered by the current sliding window according to the local contrast measure operator TrLCM, and obtain the local contrast measure matrix C;

[0052] Specifically, in order to enable the pixel points located at the edge of the image to be subjected to local contrast measure, the guided-filtered thermal infrared image f is subjected to boundary padding, that is, the value "0" is filled on the boundary of the image matrix, and the padding scale s is

[0053]

[0054] where [·] represents the floor function, N0represents the side length scale of the center block M0, which is set to 3, N circle represents the width of the isolation ring, which is set to 1, N surrounding represents the side length scale of the surrounding neighborhood block, and has N surrounding =N1=N2=...=N8=5.

[0055] Thus, the boundary-padded image matrix F is obtained, which has a size of (u+7)×(v+7), and the local contrast measure matrix C is established, which has a size of u×v.

[0056] By moving the sliding window of the three-layer nested structure pixel by pixel on the boundary-padded image matrix F, the image block covered thereby is divided into ten parts, namely one center block M0, one isolation ring and eight surrounding neighborhood blocks M t , t = 1, 2,..., 8, and the local contrast measure value of the image block covered by the current sliding window is calculated using the local contrast measure operator TrLCM, which is taken as the local contrast measure value c F (x, y) of the pixel (x, y) in the boundary-padded image matrix F, and satisfies the constraint condition 8≤x≤u+7, 8≤y≤v+7; the local contrast measure value c F (x, y) is stored in the local contrast measure matrix C, and satisfies the corresponding relationship C(x-7, y-7) = c F (x, y).

[0057] Step 4: Treat the local contrast measurement matrix C obtained in step 3) as the local contrast image T, and use it as the infrared small target detection result to realize the detection of infrared small targets.

[0058] Corresponding to the aforementioned embodiment of an infrared small target detection method based on an improved three-layer window local contrast metric, the present invention also provides an embodiment of an infrared small target detection device based on an improved three-layer window local contrast metric.

[0059] Figure 2 This is a block diagram illustrating an infrared small target detection device based on an improved three-layer window local contrast metric according to an exemplary embodiment, such as... Figure 2 As shown, the device includes:

[0060] A denoising and edge-preserving module based on guided filtering is used to acquire a denoised and edge-preserving image of the original thermal infrared image, laying the foundation for local contrast measurement. In a preferred embodiment, this module uses a guided filter to obtain a smoothed and edge-preserving thermal infrared image. The size of image f is u×v;

[0061] The local contrast measurement window structure design module is used to design a sliding window with three nested windows to capture local detection image blocks in infrared images. The sliding window with three nested windows contains ten different parts: a central block, an isolation circle, and eight surrounding neighborhood blocks. The central block is represented by M0, and the eight surrounding neighborhood blocks are represented by M1, M2, ..., M8. The isolation circle is designed to capture target edge information, target edge regions, or background components. By adding the isolation circle, the influence of target edge regions or other background components during the local contrast measurement process is reduced.

[0062] A local contrast measurement operator design module is used to design a local contrast measurement method and to carry out local contrast measurement in conjunction with a sliding window with three nested windows. In a preferred embodiment, the module designs a local contrast measurement operator TrLCM and performs edge filling on the thermal infrared image after guided filtering. The sliding window traverses the infrared image pixel by pixel and calculates the local contrast measurement value of the image block covered by the current sliding window according to the local contrast measurement operator TrLCM, and finally obtains the local contrast measurement matrix C.

[0063] The target detection module is used to traverse the infrared image pixel by pixel through the sliding window, perform local contrast measurement calculation, and perform infrared small target detection. This module regards the local contrast measurement matrix C obtained by the local contrast measurement operator design module as the local contrast image T, and uses it as the infrared small target detection result to perform infrared small target detection.

[0064] The target detection result output module is configured to output an infrared small target detection result image.

[0065] As to the apparatus in the above-mentioned embodiments, the specific manners in which the respective modules perform operations have been described in detail in the embodiments related to the method, and thus will not be described in detail here.

[0066] As to the apparatus embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to the parts of the method embodiments. The apparatus embodiments described above are merely illustrative, and the respective modules in the apparatus are a logical functional division, and in actual implementation, there can be another division manner, for example, a plurality of modules can be combined or integrated into another unit. In addition, the connection between the displayed or discussed modules can be a communication connection through some interfaces, which can be electrical or other forms. Part or all of the modules can be selected according to actual needs to achieve the purpose of the scheme of the present application. Those skilled in the art can understand and implement without paying creative labor. The following takes the disclosed real thermal infrared image sequence as an example to illustrate the specific implementation manner, so as to embody the technical effects of the present application, and the specific steps in the embodiments will not be described in detail.

[0067] Embodiment

[0068] The drawings illustrated by the embodiments of the present application can make the purpose, technical scheme and advantage of the present application more clear and understandable. It should be noted that the specific embodiments described here are merely used to explain the present application, and are not used to limit the present application. Any equivalent replacement, improvement, etc. made within the method idea and principle provided by the present application should be included in the protection scope of the present application.

[0069] Next, the disclosed infrared image sequence is taken as the research object to develop the infrared small target detection algorithm verification. In order to evaluate the algorithm effectiveness from the qualitative point of view, the intuitive result image of small target detection and the corresponding three-dimensional gray visualization image are used to evaluate the thermal infrared small target evaluation result; in order to evaluate the small target detection result from the quantitative point of view, the 3D-ROC evaluation index system is used, which includes AUC TD , AUC BS , AUC SNPR , AUC TDBS and AUC ODP , etc., which comprehensively evaluate the small target detection result from the target detection ability, background suppression ability and detector effectiveness, and the specific definitions are as follows:

[0070] AUC (D,F) represents the AUC value of the (P D , P F ) curve, which represents the effectiveness of the detector,

[0071] 0 < AUC (D,F) ≤ 1

[0072] AUC (D,τ) represents the AUC value of the (P D , τ) curve, characterizing the target detection capability of the detector,

[0073] 0 < AUC (D , τ) ≤ 1

[0074] AUC (F,τ) represents the AUC value of the (P D , τ) curve, characterizing the background rejection capability of the detector,

[0075] 0 < AUC (F , τ) ≤ 1

[0076] AUC TD may represent the joint detection capability of the detector,

[0077] 0 < AUC TD = AUC (D,F) + AUC (D,τ) ) ≤ 2

[0078] AUC BS may represent the joint background rejection capability of the detector,

[0079] -1 < AUC BS = AUC (D,F) - AUC (F,τ) ) ≤ 1

[0080] AUC TDBS may represent the combined target detection and background rejection capability of the detector,

[0081] -1 < AUC TDBS = AUC (D,τ) - AUC (F,τ) ) ≤ 1

[0082] AUC SNPR may represent the signal-to-noise ratio of the detector,

[0083]

[0084] AUC ODP may represent the total detection probability of the detector,

[0085] -1 < AUC ODP = AUC (D,F) + AUC (D , τ) - AUC (F , τ) ≤ 2

[0086] Generally, the 3D-ROC evaluation index system is divided as follows:

[0087] (1) Target detection capability (TD): AUC (D,τ) and AUC TD ;

[0088] (2) Background suppression capability (BS): AUC (F,τ) , AUC BS and AUC SNPR ;

[0089] (3) Detector effectiveness: AUC (D,F) , AUC TDBS and AUC ODP .

[0090] For two published real thermal infrared image sequences containing 100 frames, sequence one is characterized by slow-moving aircraft in the background, complex background, and contains high-light buildings, suspected targets and linear edge structures, and the image size is 256x256; sequence two is characterized by low local contrast between target and background, severe noise in the background and containing suspected targets, and the sensor contains inherent noise, and the background image size is 320x240. An example frame of the infrared image sequence is shown in Figure 3 and Figure 4 , and the corresponding target detection results are shown in Figure 5 and Figure 6 . From the detection results, the background is greatly suppressed and the target is well highlighted. Table 1 is the quantitative index of the small target detection results of thermal infrared image sequence one using LCM, TLLCM, IPI, PSTNN and the proposed method, and Table 2 is the quantitative index of the small target detection results of thermal infrared image sequence two using LCM, TLLCM, IPI, PSTNN and the proposed method. The bold and underlined values represent the optimal AUC results and the suboptimal AUC results, respectively.

[0091] Table 1 Quantitative index of detection results of thermal infrared image sequence one using LCM, TLLCM, IPI, PSTNN and the proposed method

[0092]

[0093]

[0094] Table 2 Quantitative index of detection results of thermal infrared image sequence two using LCM, TLLCM, IPI, PSTNN and the proposed method

[0095] Method AUC (D,F) ]] AUC (D,τ) ]]> AUC (F,τ) ]]> AUC TD ]]> AUC BS ]]> AUC SNPR ]]> AUC TDBS ]]> AUC ODP ]] The present invention 1.000 0.920 ]]> ​ 0.005 1.920 ]]> ​ 0.995 1.828e2 0.915 1.915 LCM 0.999 ]]> ​ 0.994 0.526 1.992 0.473 1.889 0.468 1.466 TLLCM 1.000 0.828 0.005 1.828 0.995 1.634e2 ]]> ​ 0.823 ]]> ​ 1.823 ]]> ​ IPI 0.909 0.896 0.386 ]]> ​ 1.805 0.524 ]]> ​ 2.321 0.510 1.419 PSTNN 1.000 0.493 0.005 1.493 0.995 9.851e1 0.488 1.488

[0096] Figure 7 and Figure 8 For the thermal infrared image sequence, the comparison chart of small target detection results and the corresponding three-dimensional gray visualization image of LCM, TLLCM, IPI, PSTNN and the method proposed in the application can be seen that the background suppression ability of LCM is very poor, and the target size is distorted, the background suppression ability of IPI algorithm is weak, and there is high false alarm, although TLLCM and PSTNN can enhance the target well, but cannot completely suppress the background, it can be seen that the method proposed in the application not only has good target detection performance, but also has excellent background suppression ability. Figure 9 and Figure 10 For the thermal infrared image sequence, the comparison chart of small target detection results and the corresponding three-dimensional gray visualization image of LCM, TLLCM, IPI, PSTNN and the method proposed in the application can be seen that the background suppression ability of LCM is very poor, and the target size is distorted, the background suppression ability of IPI algorithm is weak, and there is high false alarm, although TLLCM and PSTNN can enhance the target well, but cannot completely suppress the background, it can be seen that the method proposed in the application not only has good target detection performance, but also has excellent background suppression ability.

[0097] The drawings of the embodiments of the application can make the purpose, technical scheme and advantage of the application more clear and clear. It should be explained that the specific embodiments described here are only used to explain the application, and are not used to limit the application. Any equivalent replacement, improvement, etc. within the method idea and principle provided by the application should be included in the protection scope of the application.

Claims

1. An infrared small target detection method based on an improved three-layer window local contrast metric, characterized in that, Includes the following steps: Step 1): Use guided filtering to smooth and preserve the edges of the original thermal infrared image; Step 2): Based on the theory of local contrast measurement, construct an improved three-layer nested sliding window and design the local contrast measurement operator TrLCM; Specifically: Design an improved three-layer nested sliding window structure containing ten distinct parts: a central block, an isolation ring, and eight surrounding neighborhood blocks. The central block is denoted by M0, and the eight surrounding neighborhood blocks are denoted by M1, M2, ..., M8. The isolation ring is designed to capture target edge information, target edge regions, or background components. By adding the isolation ring, the influence of target edge regions or other background components during local contrast measurement is reduced. The average gray value m0 of the center block M0 is Where N0 represents the side length scale of the central block M0, (i,j)∈M0 means that the pixel (i,j) belongs to the central block M0, and I0(i,j) represents the gray value of the pixel located at position (i,j) on the central block M0. The t-th surrounding neighboring block M t grayscale mean m t for Where, N t Represents the t-th surrounding neighbor block M t The side length scale, (i,j)∈M t This indicates that pixel (i,j) belongs to the t-th surrounding neighborhood block M. t I t (i,j) represents the t-th surrounding neighbor block M. t The gray value of the pixel located at position (i,j), where t takes values ​​of 1, 2, ..., 8; By designing a local contrast weighting coefficient γ, the background suppression and target detection capabilities of the local contrast measurement operator can be further improved. Where m0 represents the average gray value of the center block M0, max(m t ),t=1,2,…,8 represents the maximum value of the average gray value of the eight surrounding neighboring blocks, and n and β are customizable adjustment coefficients; Finally, the improved three-layer window local contrast measurement operator TrLCM is: Wherein, γ is the weighting coefficient for local contrast in the design; Step 3): Fill the edges of the thermal infrared image after the guided filtering. Traverse the infrared image pixel by pixel with the sliding window. Calculate the local contrast measurement value of the image block covered by the current sliding window according to the local contrast measurement operator TrLCM. Finally, obtain the local contrast measurement matrix C. Step 4): Treat the local contrast measurement matrix C obtained in Step 3) as the local contrast image T, and use it as the infrared small target detection result to realize the detection of infrared small targets.

2. The infrared small target detection method based on improved three-layer window local contrast measurement according to claim 1, characterized in that, Step 1) specifically involves: using a guided filter to obtain a smoothed and edge-preserving thermal infrared image. The size of image f is u×v.

3. The infrared small target detection method based on improved three-layer window local contrast measurement according to claim 1, characterized in that, Step 3) specifically refers to: To enable local contrast measurement of pixels located at image edges, the thermal infrared image f after guided filtering is filled with boundary values, specifically "0"s, on the boundaries of the image matrix. The filling scale s is... Where [·] denotes the floor function, N0 represents the side length scale of the center block M0, and N circle N represents the width of the isolation ring. surrounding Let N represent the side length scale of the surrounding neighboring blocks, and let N be the side length scale of the surrounding neighboring blocks. surrounding =N1=N2=…=N8; This yields the image matrix F after boundary filling, with a size of (u+2s)×(v+2s), and a local contrast metric matrix C with a size of u×v is also established. By moving a three-layer nested sliding window pixel-by-pixel across the image matrix F after boundary padding, the covered image patch is divided into ten parts: a central patch M0, an isolation circle, and eight surrounding neighboring patches M0. t The algorithm calculates the local contrast metric value of the image patch covered by the current sliding window using the local contrast metric operator TrLCM, and uses this local contrast metric value as the local contrast metric value c of the pixel (x,y) at position F in the edge-filled image matrix F. F (x, y), and satisfying the constraints s+1≤x≤s+u, s+1≤y≤s+v; the local contrast metric c F (x, y) are stored in the local contrast metric matrix C, and the correspondence C(xs, ys) = c F (x,y).

4. The infrared small target detection method based on improved three-layer window local contrast measurement according to claim 1, characterized in that, Step 4) specifically involves treating the local contrast measurement matrix C obtained in step 3) as a local contrast image T, and using it as the infrared small target detection result to achieve the detection of infrared small targets.

5. An infrared small target detection device based on an improved three-layer window local contrast metric, implementing the method of claim 1, characterized in that, include: The denoising and edge-preserving module based on guided filtering is used to acquire the denoised and edge-preserving image of the original thermal infrared image, laying the foundation for local contrast measurement. The local contrast measurement window structure design module is used to design a sliding window with three nested windows to capture local detection image blocks in infrared images. The local contrast measurement operator design module is used to design local contrast measurement criteria and, in conjunction with a sliding window with three nested windows, to carry out local contrast measurement. The target detection module is used to traverse the infrared image pixel by pixel through the sliding window, perform local contrast measurement calculations, and detect small infrared targets. The target detection result output module is used to output infrared small target detection result images.

6. The apparatus according to claim 5, characterized in that, The local contrast measurement window structure design module describes a sliding window with three nested windows, comprising ten distinct parts: a central block, an isolation ring, and eight surrounding neighborhood blocks. The central block is denoted by M0, and the eight surrounding neighborhood blocks are denoted by M1, M2, ..., M8. The isolation ring aims to capture target edge information, target edge regions, or background components. By adding the isolation ring, the influence of target edge regions or other background components during local contrast measurement is reduced.

Images