Elliptical ring top-hat based infrared small target detection method

Abstract

The application discloses an infrared small target detection method based on an elliptical ring Top-Hat, and comprises the following steps: constructing a structure element of four angles of a detection target, wherein the structure element comprises an external structure element, an internal structure element and an elliptical structure element; defining an open operation and a closed operation according to the structure element; obtaining a top-hat transformation operation and a bottom-hat transformation operation through morphological transformation based on the open operation and the closed operation; obtaining a first detection result through fusion based on the top-hat transformation operation and the bottom-hat operation; and segmenting the first detection result based on an adaptive threshold to obtain an infrared small target detection result. The elliptical structure element obtained by constructing the structure element of four angles of the detection target can fully acquire contrast information, and the problem of serious false alarm in the case of a complex background is solved, so that the application can be widely applied to the field of infrared detection technology.

Description

Technical Field

[0001] This invention relates to the field of infrared detection technology, and in particular to an infrared small target detection method based on an elliptical ring Top-Hat. Background Technology

[0002] Infrared target detection primarily employs infrared thermal imagers combined with signal processing technology to achieve automatic target detection, and has been widely applied in various fields. In real-world scenarios, the infrared radiation emitted by an infrared target must travel through the atmosphere before being received by the optical system. During this atmospheric transmission, it suffers attenuation due to atmospheric absorption and scattering, as well as scattering from other particles such as clouds, fog, rain, and snow. Furthermore, due to the development of infrared-resistant materials and interference from complex environments, small targets often lack texture and structural information, and may even be completely obscured by complex and varied backgrounds. Therefore, infrared small target detection remains a challenging problem.

[0003] Existing infrared small target detection methods can generally be divided into two categories: single-frame-based and sequence-based detection methods. Sequence-based detection requires more prior information about the target and background. However, this prior information is difficult to obtain in practical applications. Therefore, many researchers have focused on researching single-frame-based detection methods. Single-frame-based detection methods can be further divided into four categories:

[0004] (1) Background consistency-based models assume that the background is correlated and separate the background from the original image through a designed filter, thereby achieving the purpose of target detection. Two-dimensional minimum mean square error, top-hat transform, top-hat, maximum mean and maximum median filters are the most typical background consistency-based models.

[0005] (2) Methods based on the human visual system utilize the local differences between the target and the background to construct a saliency map that can highlight the target, thereby achieving target detection. This type of method includes Absolute Directional Mean Difference Detector (ADMD), Local Contrast Measure (LCM), Multiscale Patch-based Contrast Measure (MPCM), and Improved Local Contrast Measure (ILCM), etc.

[0006] (3) Methods based on low-rank sparse matrix recovery utilize the sparsity of infrared small targets and the low-rank nature of the background to transform detection into a classification task, thereby achieving the detection of infrared small targets. This type of method includes Infrared Patch-Image (IPI), Partial Sum of the Tensor Nuclear Norm (PSTNN), Non-Convex Rank Approximation Minimization (NRAM), and... Nonconvex tensor fibered rank approximation (NTFRA), etc.

[0007] (4) Deep learning-based models utilize a large amount of training data to learn an abstract representation of infrared small targets, thereby achieving the purpose of target detection. Currently, most methods of this type use simulated datasets or data augmentation methods to train the parameters in the network.

[0008] The existing technology has the following main problems:

[0009] (1) The first issue is the contradiction between detection rate and real-time performance. With a low false alarm rate, filtering-based and contrast-based methods have high real-time performance, but low detection rate. Conversely, low-rank sparse matrix recovery-based methods have high detection rate, but poor real-time performance. However, in engineering applications, both high detection rate and high real-time performance are required.

[0010] (2) The second problem is the poor performance in detecting bipolar targets. Due to the influence of external factors and the characteristics of the target itself, some small targets often exhibit bipolar characteristics, that is, the target gray level can be higher or lower than the background gray level. However, most researchers design infrared small target detectors based on the assumption that the target gray level is higher than the background gray level. Summary of the Invention

[0011] In view of this, embodiments of the present invention provide a real-time detection method for infrared small targets based on elliptical ring Top-Hat.

[0012] On one hand, embodiments of the present invention provide an infrared small target detection method based on elliptical ring Top-Hat, including:

[0013] Construct structural information of the target from multiple angles, including external structural elements, internal structural elements, and elliptical structural elements;

[0014] Based on the structural information, define opening and closing operations;

[0015] Based on the opening and closing operations, the top-hat transformation operation and the bottom-hat transformation operation are obtained through morphological transformation.

[0016] Based on the top-hat transformation operation and the bottom-hat transformation operation, a first detection result is obtained by fusion;

[0017] The first detection result is segmented based on an adaptive threshold to obtain the infrared small target detection result.

[0018] Optionally, the construction of structural information of the target from multiple angles includes:

[0019] Construct external structural elements of the target from multiple angles;

[0020] Construct internal structural elements of the target from multiple angles;

[0021] Based on the external structural element and the internal structural element, the elliptical structural element is obtained by subtraction.

[0022] Optionally, in the step of defining opening and closing operations based on the structural information, the expressions for opening and closing operations are:

[0023]

[0024]

[0025] in, For opening operation, For closing operations, B is an external structural element. b It is an intermediate structural element whose size lies between the outer and inner structural elements. It is an elliptical structural element. For erosion calculation, This is for the expansion operation.

[0026] Optionally, in the step of obtaining the top-hat transformation operation and the bottom-hat transformation operation through morphological transformation based on the opening operation and the closing operation, the expressions for the top-hat transformation operation and the bottom-hat transformation operation are as follows:

[0027]

[0028]

[0029] in, This is the result of the top hat transformation operation. This is the result of the bottom hat transformation operation. For infrared images, min() takes the minimum value and max() takes the maximum value.

[0030] Optionally, obtaining the first detection result by fusing the top hat transformation operation and the bottom hat transformation operation includes:

[0031] By performing top-hat transformation and bottom-hat transformation operations, the results of the top-hat transformation and bottom-hat transformation operations in the four directions are obtained.

[0032] The initial detection target is obtained by adding the top cap transformation results from the four directions to the bottom cap transformation results;

[0033] The initial detection targets in the four directions are multiplied together to obtain the first detection result.

[0034] Optionally, in the step of segmenting the first detection result based on an adaptive threshold to obtain the infrared small target detection result, the formula for calculating the adaptive threshold is:

[0035]

[0036] Where T is the adaptive threshold. The mean of the first test results. The standard deviation of the first test result. It is a constant.

[0037] Optionally, the step of segmenting the first detection result based on an adaptive threshold to obtain the infrared small target detection result includes:

[0038] The adaptive threshold is calculated using the adaptive threshold calculation formula.

[0039] The first detection result is compared with the adaptive threshold. When the first detection result is greater than the adaptive threshold, the gray level of the detected target is set to the first threshold to obtain the infrared small target detection result.

[0040] When the first detection result is less than or equal to the adaptive threshold, the gray value of the detected target is set as the second threshold and used as the background pixel.

[0041] On the other hand, embodiments of the present invention provide an infrared small target detection device based on an elliptical ring Top-Hat, comprising:

[0042] The first module is used to construct structural information of the target from multiple angles, including external structural elements, internal structural elements, and elliptical structural elements.

[0043] The second module is used to define opening and closing operations based on the structural information.

[0044] The third module is used to obtain the top-hat transformation operation and the bottom-hat transformation operation through morphological transformation based on the opening operation and the closing operation;

[0045] The fourth module is used to fuse the top cap transformation calculation result and the bottom cap transformation calculation result to obtain the first detection result;

[0046] The fifth module is used to compare the first detection result with the adaptive threshold to obtain the infrared small target detection result.

[0047] On the other hand, embodiments of the present invention provide an electronic device, including a processor and a memory; the memory is used to store a program; the processor executes the program to implement the infrared small target detection method based on elliptical ring Top-Hat as described above.

[0048] On the other hand, embodiments of the present invention provide a computer-readable storage medium storing a program that is executed by a processor to implement the aforementioned infrared small target detection method based on elliptical ring Top-Hat.

[0049] This invention also discloses a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device can read the computer instructions from the computer-readable storage medium and execute the computer instructions, causing the computer device to perform the aforementioned method.

[0050] The embodiments of the present invention include at least the following beneficial results: The present invention constructs structuring elements at four angles of the target to obtain elliptical structuring elements that can fully acquire contrast information, thus solving the problem of severe false alarms in complex backgrounds; The present invention obtains the first detection result by fusing the top-hat transformation operation result with the bottom-hat transformation operation result, thus solving the problem of insufficient dark target datasets; The present invention obtains the top-hat transformation operation and the bottom-hat transformation operation through morphological transformation, enabling real-time detection of bipolar infrared small targets. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 A flowchart of an infrared small target detection method based on an elliptical ring Top-Hat provided in an embodiment of the present invention;

[0053] Figure 2 This is a graph showing the change in image size of the same target as the distance changes, provided in an embodiment of the present invention.

[0054] Figure 3 This is a schematic diagram of the minimum circumscribed ellipse of the target circumscribed rectangle provided in an embodiment of the present invention;

[0055] Figure 4 A diagram illustrating the construction process of structural elements provided in an embodiment of the present invention;

[0056] Figure 5 The following are the result diagrams of three different fusion methods provided in the embodiments of the present invention;

[0057] Figure 6 The detection results of bipolar small targets using three fusion methods provided in the embodiments of the present invention are shown in the figure.

[0058] Figure 7 ROC curves of several bipolar small target detection methods provided in this embodiment of the invention on six datasets;

[0059] Figure 8 This is a schematic diagram of an infrared small target detection device based on an elliptical ring Top-Hat provided in an embodiment of the present invention. Detailed Implementation

[0060] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0061] On the one hand, embodiments of the present invention provide an infrared small target detection method based on elliptical ring Top-Hat, such as Figure 1 The method shown includes, but is not limited to, steps S100-S500:

[0062] S100: Construct structural information of the target from multiple angles, including external structural elements, internal structural elements, and elliptical structural elements.

[0063] Optionally, an infrared image is acquired through a lens. This image contains the background and small infrared targets. The lens focal length directly determines the size of the image formed by the target, that is, how many pixels it occupies on the focal plane. The number of pixels occupied by the image of each target on the focal plane can be calculated from the target size, the distance between the target and the detector, and the spatial resolution. The calculation formula is as follows:

[0064]

[0065]

[0066] in, Indicates spatial resolution. Indicates the pixel size of the detector. Indicates the focal length of the detector; Indicates the size of the image formed by the target. Indicates the size of the target. This indicates the distance between the target and the detector.

[0067] The International Society for Optical Engineering defines small infrared targets as those with a size smaller than [missing information]. When the detector is fixed, the focal length and pixel size remain unchanged. According to the formula for calculating the number of pixels occupied by the target on the focal plane, the reasons for the change in the size of small targets are as follows: (1) The size of the target itself: when the distance between the target and the camera is fixed, the size of the target image on the same camera is determined by the size of the target itself; (2) The distance between the target and the camera: when the target is fixed, the size of the target image on the camera is determined by the distance between the target and the camera. Through the analysis of the above two types of infrared small target data, this embodiment of the invention found that: different infrared small targets can be bounded by a rectangular frame. As the infrared target decreases, the ratio of the length to the width of the bounding rectangle becomes smaller, until it is 1 (at which point the bounding rectangle is a square). The process of the change in the length to width ratio can be quantitatively expressed by the roundness. Figure 2 This describes the change in the size of the same target as the distance increases. Figure 2 It can be seen that as the size of the infrared target decreases, the circularity increases. The formula for calculating circularity is as follows:

[0068]

[0069] in, Indicates roundness; Represents the area of ​​the outline region; Indicates the perimeter of the outline region.

[0070] Optionally, due to atmospheric attenuation, during long-range infrared detection, the target gradually becomes elliptical as the object distance increases. When the distance is sufficiently far, the original infrared target will become a point infrared target. Furthermore, based on the relationship between the target and its circularity, any small infrared target can be bounded by a rectangular frame. Once the bounding rectangle is determined, the minimum bounding ellipse of this rectangle is also determined, and the ellipticity of the minimum bounding ellipse is the same as the aspect ratio of the rectangle. Figure 3 This demonstrates the minimum bounding ellipse of the target bounding rectangle. Based on the general equation of an ellipse and the formula for the area of ​​an ellipse, solving for the minimum bounding ellipse of the rectangle is transformed into a convex optimization problem:

[0071]

[0072]

[0073] because and We can obtain:

[0074]

[0075]

[0076] Therefore, the value of pi can be obtained as:

[0077]

[0078] For the reasons mentioned above, an elliptical structural element was constructed. When the target is as small as a single pixel, the circularity is... Therefore, the circularity of the elliptical structural element is set to 0.8. Figure 4 The process of constructing structural elements is demonstrated. First, external structural elements are constructed at four angles of the target. Then, internal structural elements are constructed at four angles of the target. Finally, elliptical structural elements are obtained by subtraction. In one embodiment of the present invention, the four angles are... .

[0079] S200: Define opening and closing operations based on the structural information.

[0080] Optionally, based on the structural information, the expressions for opening and closing operations in this step are defined as follows:

[0081]

[0082]

[0083] in, For opening operation, For closing operations, B is an external structural element. b It is an intermediate structural element whose size lies between the outer and inner structural elements. It is an elliptical structural element. It is an internal structural element and external structural elements The edge area between, For erosion calculation, This is for the expansion operation.

[0084] Existing techniques using a ring-shaped top-hat transform construct a concentric structuring element. However, this concentric structuring element acquires contrast information in all directions at the same scale, thus failing to adequately suppress background information and leading to a high false alarm rate in complex backgrounds. Furthermore, considering the uncertainty of the target orientation, a structuring element with four directions is proposed.

[0085] S300: Based on the opening and closing operations, the top cap transformation operation and the bottom cap transformation operation are obtained through morphological transformation.

[0086] Optionally, in the step of obtaining the top-hat transformation operation and the bottom-hat transformation operation through morphological transformation based on the opening operation and the closing operation, the expressions for the top-hat transformation operation and the bottom-hat transformation operation are as follows:

[0087]

[0088]

[0089] in, The result of the top-hat transform operation is used to detect bright targets in infrared images; The result of the bottom cap transformation is used to detect dark targets in infrared images; For infrared images, `min()` takes the minimum value, and `max()` takes the maximum value. Depending on the specific conditions, the target's temperature can be higher or lower than the background temperature, and the corresponding infrared radiation intensity may be higher or lower than the background, exhibiting a "bipolar" imaging characteristic. Due to the bipolar characteristic of the target, it will present two opposite shapes on the image plane: one with a high center and low edges, and the other with a low center and high edges. Therefore, infrared small target detection algorithms must detect not only bright targets but also dark targets to meet practical application requirements.

[0090] S400: Based on the top cap transformation operation and the bottom cap transformation operation, the first detection result is obtained by fusion.

[0091] Optionally, in order to detect bipolar targets, this embodiment of the invention fuses the top-hat transformation and bottom-hat transformation in four angular directions. Common fusion methods include addition, multiplication, and extremum fusion. Under a bright background, extremum fusion may cause target loss, so this embodiment of the invention fuses using addition and multiplication, which can form three fusion methods. (1) Summing the top-hat transformation and bottom-hat transformation in the four directions; (2) Multiplying the top-hat transformation and bottom-hat transformation in the four directions respectively; (3) First adding the top-hat transformation and bottom-hat transformation in the corresponding directions, and then multiplying the results in the four directions. The fusion process of these three fusion methods is as follows:

[0092] (1) Summation of top hat transformation and bottom hat transformation in 4 directions:

[0093]

[0094] (2) Multiply the top-hat transformation and bottom-hat transformation in the four directions respectively:

[0095]

[0096] (3) First, add the top hat transformation and bottom hat transformation in the corresponding directions, and then multiply the results in the four directions:

[0097]

[0098] in, It is the result of the top-hat transformation operation. It is the result of the bottom-hat transformation operation. This is the first test result. This is the initial test result.

[0099] like Figure 6 The images show the results of fusion using three different fusion methods. The first column displays the original image. In this embodiment, the image is normalized, and the target is magnified and displayed in the upper left or lower right corner. The second column shows the detection results using the first fusion method, showing many false alarms. Because the first fusion method adds the results of the top-hat and bottom-hat transformations in four directions, false alarms are also accumulated. The third column shows the detection results using the second fusion method, showing no target detected. Since bright and dark targets have opposite imaging characteristics, multiplying the results of the top-hat and bottom-hat transformations in four directions results in an image with all grayscale values ​​of zero. The fourth column shows the detection results using the third fusion method, showing that bipolar targets can be effectively detected. Because the third fusion method first adds the results of the top-hat and bottom-hat transformations in the corresponding directions, it ensures the completeness of the detection results. Then, it multiplies the results in four directions, further enhancing the detection of small targets. Therefore, this invention chooses the third fusion method to fuse the results of the top-hat and bottom-hat transformations in four directions.

[0100] Optionally, in this embodiment of the invention, the top hat transformation operation and the bottom hat transformation operation are first performed to obtain the top hat transformation operation results and the bottom hat transformation operation results in four directions; then the top hat transformation operation results and the bottom hat transformation operation results in four directions are added together to obtain the initial detection target; finally, the initial detection targets in four directions are multiplied together to obtain the first detection result.

[0101] S500: The first detection result is segmented based on an adaptive threshold to obtain the infrared small target detection result.

[0102] Optionally, after the fusion process, the contrast of the region containing the small infrared target in the image will be at its maximum, appearing as a peak in the image. Step S500 includes, but is not limited to, steps S510-S530:

[0103] S510: The adaptive threshold is calculated using the adaptive threshold calculation formula.

[0104] Optionally, the adaptive threshold calculation formula in this invention is:

[0105]

[0106] Where T is the adaptive threshold. The mean of the first test results. The standard deviation of the first test result. It is a constant.

[0107] S520: Compare the first detection result with the adaptive threshold. When the first detection result is greater than the adaptive threshold, set the grayscale of the detected target to the first threshold to obtain the infrared small target detection result.

[0108] Optionally, in one embodiment of the present invention, the first threshold is 255. The first detection result is compared with an adaptive threshold. When the first detection result is greater than the adaptive threshold, the grayscale of the detected target is set to 255 to obtain the infrared small target detection result.

[0109] S530: When the first detection result is less than or equal to the adaptive threshold, the gray value of the detected target is set to the second threshold as a background pixel.

[0110] Optionally, in one embodiment of the present invention, the second threshold is 0. The first detection result is compared with the adaptive threshold. When the first detection result is less than or equal to the adaptive threshold, the grayscale value of the detected target is set to 0, which is used as a background pixel.

[0111] In practical applications, the gray value of a target may be higher or lower than the background due to the influence of internal and external factors. Most existing infrared small target detection methods and datasets are for targets with gray values ​​higher than the background.

[0112] Therefore, in this embodiment of the invention, dark targets are artificially embedded into an infrared background image and a background image with a single bright target, constructing two datasets with dark targets. To ensure that the targets are smoothly embedded into the images, taking advantage of the fact that the gray-level distribution of the targets conforms to a Gaussian distribution, one embodiment of the invention employs a random adaptive fusion strategy to embed the simulated targets into the background image with high quality to complete the data synthesis. The specific process is as follows:

[0113] Step 1: Randomly generate the target's location within the image size range. Based on the definition of small targets in SPIE (the size of a small target is less than 0.12% of the image size), a suitable target size is constructed. ;

[0114] Step 2: Calculate the mean of the horizontal and vertical coordinates based on the target's location and size. and The variance of randomly selected x and y axes , Generate Gaussian matrix ;

[0115]

[0116] Step 3: with Centered on the original image, remove background image blocks of the same size as the simulation target from the original image. And calculate the mean of the image patch. Finally, a random selection was made. , and These represent the adjustment parameters of the fixed and randomly varying parts of the Gaussian scaling factor, respectively, used to construct dark targets of different intensities.

[0117]

[0118] Step 4: Based on the selected and Generate Gaussian matrix scaling factor ;

[0119]

[0120] Step 5: Construct the dark target .

[0121]

[0122] Examples of three different fusion methods are shown below. Figure 5 As shown, Figure 5 (a) is the original grayscale image. Figure 5 (b) is a directly embedded image. Figure 5 (c) is the result image of the random adaptive fusion embedding. The random adaptive fusion eliminates the embedding boundary traces, making the generated image closer to the real scene.

[0123] like Figure 6 As shown Figure 5The images show the results of detecting bipolar targets using three fusion methods. The first column displays the original image, which has been normalized, and the target is magnified and displayed in the upper left or lower right corner. The second column shows the detection results using the first fusion method, showing many false alarms. This is because the first fusion method adds the results of the top-hat and bottom-hat transforms in all four directions, resulting in an accumulation of false alarms. The third column shows the detection results using the second fusion method, showing no target detected. Since bright and dark targets have opposite imaging characteristics, multiplying the results of the top-hat and bottom-hat transforms in all four directions yields an image with all grayscale values ​​of zero. The fourth column shows the detection results using the third fusion method, showing that both bipolar targets are effectively detected. The third fusion method first adds the results of the top-hat and bottom-hat transforms in the corresponding directions, ensuring the completeness of the detection results, and then multiplies the results in all four directions, further enhancing the detection of small targets.

[0124] One embodiment of this invention further employs Signal-to-Noise Ratio Gain (SNRG) and Background Suppression Factor (BSF) to perform a more objective quantitative analysis of the proposed method. To verify the performance of the proposed method, this embodiment compares it with Absolute Directional Mean Difference (ADMD), Ring Top-Hat (RTH), Local Contrast Measure (LCM), Multiscale Patch-based Contrast Measure (MPCM), Multiscale Gray Difference Weighted Image Entropy (MGDWIE), Infrared Patch-Image (IPI), Partial Sum of the Tensor Nuclear Norm (PSTNN), Non-Convex Rank Approximation Minimization (NRAM), and... The detection performance of six methods using the nonconvex tensor fiber rank approximation (NTFRA) paradigm is presented. Tables 1 and 2 list the average signal-to-noise ratio gain (SNRG) and background suppression factor (BSF) obtained using different methods on the six datasets, with the best results marked in bold and the second-best results marked in underline. For each evaluation, the highest value represents the best result. The results in the tables show that the proposed elliptic ring top-hat-based infrared small target detection method achieves the highest SNRG and BSF across all datasets, indicating that the proposed method can effectively suppress the background while enhancing the target in various scenarios.

[0125] Table 1. Average SNRG values ​​obtained by different methods

[0126]

[0127] Table 2. Average BSF values ​​obtained by different methods

[0128]

[0129] like Figure 7 The ROC (Receiver Operating Characteristic) curves of six different methods on six datasets are shown. The ROC curve of the proposed elliptic ring top-hat infrared small target detection method is located in the upper left corner of all datasets, indicating that the proposed method achieves the best detection performance, especially on datasets 4, 5, and 6. This demonstrates that the proposed method achieves a high detection probability with a low false alarm rate.

[0130] The processing efficiency of different methods on six datasets was compared, and the results are shown in Table 3. RTH is the fastest among all tested methods, but it cannot effectively suppress background noise and can only process bright targets. The method of this embodiment is only slower than RTH on datasets 1, 2, 4, 5, and 6. On the third dataset, ADMD is slightly better than the method of this embodiment; however, ADMD is sensitive to noise and can only process bright targets. The efficiency of the optimization-based method is lower than that of the contrast-based method. Furthermore, the efficiency of the optimization-based method varies significantly for different images. In summary, the method proposed in this invention can detect small infrared targets of different complexities and polarities in real time.

[0131] Table 3 Comparison of detection times for different methods (unit: seconds)

[0132]

[0133] The following example illustrates the application of the infrared small target detection method based on the elliptical ring Top-Hat of this invention:

[0134] Construct structural information of the target from multiple angles, including external structural elements, internal structural elements, and elliptical structural elements; define opening and closing operations based on the structural information; obtain top-hat and bottom-hat transformation operations through morphological transformation based on the opening and closing operations; obtain a first detection result by fusing the top-hat and bottom-hat transformation operations; segment the first detection result based on an adaptive threshold to obtain an infrared small target detection result.

[0135] The infrared small target detection method based on elliptical ring Top-Hat of the present invention has the following beneficial effects:

[0136] 1. This invention constructs structural elements at four angles of the target to obtain elliptical structural elements that can fully acquire contrast information, thus solving the problem of serious false alarms in complex background situations.

[0137] 2. This invention obtains the first detection result by fusing the top-hat transformation operation result and the bottom-hat transformation operation result, thus solving the problem of insufficient dark target dataset;

[0138] 3. This invention obtains top-hat transformation and bottom-hat transformation operations through morphological transformation, which can detect bipolar infrared small targets in real time.

[0139] On the other hand, such as Figure 8 As shown, this embodiment of the invention provides an infrared small target detection device based on an elliptical ring Top-Hat, comprising:

[0140] The first module 801 is used to construct structural information of the target from multiple angles, wherein the structural information includes external structural elements, internal structural elements and elliptical structural elements.

[0141] The second module 802 is used to define opening and closing operations based on the structural information;

[0142] The third module 803 is used to obtain the top cap transformation operation and the bottom cap transformation operation through morphological transformation based on the opening operation and the closing operation;

[0143] The fourth module 804 is used to fuse the top cap transformation calculation result and the bottom cap transformation calculation result to obtain the first detection result;

[0144] The fifth module 805 is used to compare the first detection result with the adaptive threshold to obtain the infrared small target detection result.

[0145] On the other hand, embodiments of the present invention provide an electronic device, including a processor and a memory; the memory is used to store a program; the processor executes the program to implement the infrared small target detection method based on elliptical ring Top-Hat as described above.

[0146] On the other hand, embodiments of the present invention provide a computer-readable storage medium storing a program that is executed by a processor to implement the aforementioned infrared small target detection method based on elliptical ring Top-Hat.

[0147] This invention also discloses a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device can read the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, causing the computer device to perform... Figure 1 The method shown.

[0148] In some alternative embodiments, the functions / operations mentioned in the block diagrams may not occur in the order shown in the operation diagrams. For example, depending on the functions / operations involved, two consecutively shown blocks may actually be executed substantially simultaneously, or the blocks may sometimes be executed in reverse order. Furthermore, the embodiments presented and described in the flowcharts of this invention are provided by way of example to provide a more comprehensive understanding of the technology. The disclosed methods are not limited to the operations and logic flows presented herein. Alternative embodiments are contemplated in which the order of various operations is altered and sub-operations described as part of a larger operation are executed independently.

[0149] Furthermore, although the invention has been described in the context of functional modules, it should be understood that, unless otherwise stated, one or more of the described functions and / or features may be integrated into a single physical device and / or software module, or one or more functions and / or features may be implemented in a separate physical device or software module. It is also understood that a detailed discussion of the actual implementation of each module is unnecessary for understanding the invention. Rather, given the properties, functions, and internal relationships of the various functional modules in the apparatus disclosed herein, the actual implementation of the module will be understood within the scope of conventional skill of an engineer. Therefore, those skilled in the art can implement the invention as set forth in the claims using ordinary techniques without excessive experimentation. It is also understood that the specific concepts disclosed are merely illustrative and not intended to limit the scope of the invention, which is determined by the full scope of the appended claims and their equivalents.

[0150] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0151] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0152] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0153] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0154] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0155] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

[0156] The above is a detailed description of the preferred embodiments of the present invention, but the present invention is not limited to the embodiments described. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention, and these equivalent modifications or substitutions are all included within the scope defined by the claims of this application.

Claims

1. An infrared small target detection method based on elliptical ring Top-Hat, characterized in that, The method comprises the following steps: constructing structure information of four angles of a detection target, wherein the structure information comprises an external structure element, an internal structure element and an elliptical structure element; defining an opening operation and a closing operation according to the structure information; obtaining a top-hat transformation operation and a bottom-hat transformation operation through morphological transformation based on the opening operation and the closing operation; obtaining a first detection result through fusion based on the top-hat transformation operation and the bottom-hat transformation operation; segmenting the first detection result based on an adaptive threshold to obtain an infrared small target detection result; the step of constructing structure information of four angles of a detection target comprises the following steps: constructing an external structure element on four angles of a detection target; constructing an internal structure element on four angles of a detection target; obtaining an elliptical structure element through difference based on the external structure element and the internal structure element; in the step of obtaining a top-hat transformation operation and a bottom-hat transformation operation through morphological transformation based on the opening operation and the closing operation, the expressions of the top-hat transformation operation and the bottom-hat transformation operation are as follows: wherein, is a top-hat transform operation result, is a bottom-hat transform operation result, is a pixel value of any point of the infrared image; represents an index value of a pixel point of the infrared image; min() is a minimum value, and max() is a maximum value; is an opening operation, is a closing operation, is an external structure element; the step of obtaining a first detection result through fusion based on the top-hat transformation operation and the bottom-hat transformation operation comprises the following steps: obtaining top-hat transformation operation results and bottom-hat transformation operation results in four angle directions through the top-hat transformation operation and the bottom-hat transformation operation; adding the top-hat transformation operation results and the bottom-hat transformation operation results in the four angle directions to obtain initial detection targets; multiplying the initial detection targets in the four angle directions to obtain the first detection result.

2. The elliptical ring Top-Hat based infrared small target detection method according to claim 1, characterized in that, in the step of defining an opening operation and a closing operation according to the structure information, the expressions of the opening operation and the closing operation are as follows: wherein, represents an infrared image; represents an index value of an infrared image pixel; is an opening operation, is a closing operation, is an external structuring element, B b is an intermediate structuring element of size between the external structuring element and the internal structuring element, is an elliptical structuring element, is an erosion operation, is a dilation operation.

3. The elliptical ring Top-Hat based infrared small target detection method according to claim 1, wherein, in the step of segmenting the first detection result based on an adaptive threshold to obtain an infrared small target detection result, the calculation formula of the adaptive threshold is as follows: wherein T is an adaptive threshold, is a mean value of the first detection results, is a standard deviation of the first detection results, is a constant.

4. The elliptical ring Top-Hat based infrared small target detection method according to claim 1, wherein, the step of segmenting the first detection result based on an adaptive threshold to obtain an infrared small target detection result comprises the following steps: calculating the adaptive threshold through the adaptive threshold calculation formula; comparing the first detection result with the adaptive threshold, when the first detection result is greater than the adaptive threshold, setting the gray value of the detection target as a first threshold to obtain the infrared small target detection result; when the first detection result is less than or equal to the adaptive threshold, setting the gray value of the detection target as a second threshold as a background pixel.

5. A device for implementing the elliptical ring Top-Hat based infrared small target detection method according to any one of claims 1-4, characterized in that, The method comprises the following steps: a first module is configured to construct structure information of four angles of a detection target, wherein the structure information comprises an external structure element, an internal structure element and an elliptical structure element; a second module is configured to define an opening operation and a closing operation according to the structure information; a third module is configured to obtain a top-hat transformation operation and a bottom-hat transformation operation through morphological transformation based on the opening operation and the closing operation; a fourth module is configured to fuse the top-hat transformation operation results and the bottom-hat transformation operation results to obtain a first detection result; a fifth module is configured to compare the first detection result with an adaptive threshold to obtain an infrared small target detection result.

6. An electronic device, comprising: The method comprises a processor and a memory. The memory is configured to store a program. The processor is configured to execute the program to implement the method according to any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, The storage medium stores a program, and the program is executed by the processor to implement the method in any one of claims 1 to 4.

Images