Radar target identification method based on image classification and identification in sea clutter environment

Abstract

The invention discloses a radar target recognition method based on image classification recognition in a sea clutter environment, and the method comprises the steps: collecting each frame of radar observation data in a circular scanning range, carrying out the signal preprocessing of an original echo signal, and carrying out the non-coherent accumulation, and obtaining the amplitude data of each trace point; abnormal value processing is carried out on the amplitude data, and abnormal low value points whose power spectrum density is lower than a preset threshold value are eliminated; performing graying processing on the preprocessed amplitude data, and converting the preprocessed amplitude data into a grayscale image; performing morphological opening operation and morphological closing operation on the grayscale image in sequence so as to suppress image noise and fill up holes in the target; performing edge detection on the image, and enhancing target edge information; and performing connected domain analysis on the image data, and performing connected domain marking and statistics by adopting a scanning line method so as to identify a radar target. According to the invention, effective identification and suppression of the radar target in the sea clutter environment are realized, and the detection and discrimination capability of the radar system on the effective target in the sea clutter meteorological environment is significantly improved.

Description

Technical Field

[0001] This invention relates to the field of radar target detection technology, and in particular to a radar target recognition method based on image classification in a sea clutter environment. Background Technology

[0002] With the widespread application of radar technology in civilian, defense, aerospace, and maritime surveillance fields, the performance requirements for radar systems in target detection, identification, and tracking under different scenarios are increasing. In various application environments such as low-altitude surveillance, intelligent transportation, and marine environments, radar needs to operate stably in a variety of complex environments. However, clutter signals generated under marine conditions have become a significant source of interference affecting the target detection and identification capabilities of radar systems.

[0003] As a major source of interference for radar signals in marine environments, sea clutter exhibits significant environmental correlations and parameter dependencies in its impact patterns. From an environmental perspective, sea clutter intensity is positively correlated with sea state; the higher the Beaufort scale, the more severe the wave breaking, the greater the number of surface scatterers, and the more nonlinearly the clutter power increases. Furthermore, at low grazing angles, the superposition of specular reflection and scattered signals from the sea surface can increase clutter power by 10-20 dB compared to medium-to-high grazing angles, forming a strong interference area. In addition, sea clutter also exhibits significant spatiotemporal correlations. Temporally, it fluctuates periodically with the wave cycle; spatially, it forms continuous clutter blocks along the radar beam direction, with clutter correlation between adjacent range cells reaching over 0.7, significantly different from the discrete distribution characteristics of ground clutter.

[0004] Sea clutter impacts radar system performance throughout the entire target detection, identification, and tracking process. In the detection phase, strong sea clutter masks target echo signals, especially for small-section targets, whose echo amplitude is often lower than the average clutter power. This causes conventional constant false alarm rate (CFAR) detection algorithms to fail. In clutter-nonuniform regions, the CFAR threshold is difficult to adapt to local clutter intensity, and the false alarm rate can rise to over 30%. Simultaneously, weak clutter in calm areas is easily misidentified as targets, increasing the false alarm rate several times compared to clutter-free environments. Furthermore, strong sea clutter interference increases the complexity of radar signal processing. To maintain a good detection rate, algorithm iteration delays ultimately sacrifice the radar's real-time detection capability, indirectly affecting the system's response speed to dynamic targets.

[0005] To address this issue, image recognition and analysis technology has made significant progress in radar clutter suppression in recent years. By extracting the spatiotemporal characteristics of radar echoes and applying image recognition and analysis techniques, targets can be more effectively distinguished from environmental sea clutter, improving the accuracy of clutter identification and suppression. Image recognition and analysis technology can adaptively extract structural features from sea clutter radar signals in space, distinguishing clutter with common characteristics from real targets in the data space, thereby achieving more intelligent and efficient clutter suppression.

[0006] Although some research has applied image-based methods to radar sea clutter classification and suppression, efficient suppression methods are still under development and refinement in full-scale, severe weather conditions at sea and in practical engineering applications. How to accurately model and classify sea clutter within extremely short radar frame periods to achieve effective identification and suppression has become a key technical issue for improving radar engineering system performance and ensuring target detection accuracy. Therefore, there is an urgent need to develop a method for image-based classification and identification of radar targets in sea clutter environments, providing technical support for efficient radar detection under such conditions. Summary of the Invention

[0007] To address the challenges of target detection in sea clutter environments using existing radar systems, such as the increasing intensity of sea clutter signals with rising Beaufort scale and real-time dynamic changes in meteorological conditions, traditional clutter suppression methods have limited identification and suppression capabilities in strong sea clutter environments. These methods struggle to effectively separate noise, sea clutter, and the true target, and existing single-dimensional suppression techniques are inadequate for simultaneously suppressing multiple types of interference, leading to target features being masked by mixed interference. This invention provides a radar target identification method based on image classification in sea clutter environments. By accumulating the radar echo signal's cycle time and employing an image target recognition method, it achieves effective identification and suppression of radar targets in sea clutter environments. This method can improve the target detection accuracy of radar systems under sea clutter conditions and enhance the adaptability of radar clutter suppression.

[0008] To achieve the above objectives, the technical solution adopted by the present invention is as follows: In a first aspect, the present invention provides a radar target identification method based on image classification in a sea clutter environment, comprising the following steps: S1. Continuous data acquisition and data preprocessing; specifically including: S1.1 Acquire radar observation data for each frame within the ring scan range, perform signal preprocessing on the raw echo signal including pulse compression, moving target display, filtering, noise reduction and amplitude and phase correction, and then perform non-coherent accumulation to obtain the amplitude data of each point trace; S1.2. Perform outlier processing on the amplitude data and remove abnormally low value points where the power spectral density is lower than a preset threshold. S1.3. Convert the preprocessed amplitude data to grayscale to form a grayscale image. S2. Perform morphological opening and morphological closing operations on the grayscale image in sequence to suppress image noise and fill in the holes inside the target. S3. Perform edge detection on the image processed in step S2 to enhance the target edge information; S4. Perform connected component analysis on the image data processed in step S3. Based on the 4-neighborhood connectivity rule, use the scan line method to mark and count connected components in order to identify radar targets.

[0009] In one embodiment of the present invention, step S1.2 specifically includes: setting a power spectral density threshold. The amplitude data below the threshold The values ​​were identified as abnormally low and removed.

[0010] In one embodiment of the present invention, step S1.3 specifically includes: performing grayscale processing on the amplitude of the obtained radar sweep observation data.

[0011] Calculate its maximum and minimum values ​​as the scaling factor, using the following formula:

[0012] in, For the amplitude data of the dot trace, min(D) and max(D) are the minimum amplitude and maximum amplitude in the data frame, respectively. Round() represents the rounding operation.

[0013] In one embodiment of the present invention, in step S2, the morphological opening operation is a dilation-erosion operation, and the morphological closing operation is an erosion-dilation operation; the dilation operation is as follows: a structuring element of a preset shape and size is slid across the image, and when any pixel covered by the structuring element is foreground, the center pixel is retained as foreground; otherwise, the center pixel is dilated to become background; the erosion operation is as follows: a structuring element of a preset shape and size is slid across the image, and when all pixels covered by the structuring element are foreground, the center pixel is retained as foreground; otherwise, the center pixel is eroded to become background; the kernels of the morphological opening and closing operations are both: .

[0014] In one embodiment of the present invention, the shape of the structural element includes a rectangle and a circle, and the size of the structural element is 3.

[0015] In one embodiment of the present invention, in step S3, the edge detection employs a Laplacian filtering algorithm, using a preset 4-neighborhood Laplacian convolution kernel to perform convolution operations on the image to sharpen the edges. The convolution kernel is: .

[0016] In one embodiment of the present invention, in step S4, a 4-neighborhood is selected, containing pixels in the top, bottom, left, and right directions. Under GPU runtime, the scanline method is used to complete the marking through two scans, obtaining the final result. The connected component marking and statistics using the scanline method specifically includes: Step 1: Scan line by line; Scan each row of pixels from left to right and from top to bottom. If the current pixel is foreground, check the labels of the pixels to its left and top. If neither is labeled, assign a new temporary label to the current pixel. If only one side is labeled, the current pixel inherits that label. If both sides are labeled, the current pixel inherits one of the labels and records that the two labels are equivalent. Step 2: Merge equivalent tags to generate the final tag; organize the equivalent tag pairs recorded in Step 1, merge the equivalent tags using a disjoint-set data structure to obtain a unique final tag; traverse the image and replace the temporary tag of each pixel with the merged final tag.

[0017] In one embodiment of the present invention, the radar target identification method further includes: Based on the size of the area and the centroid of the target, combined with the target obtained by conventional signal processing, the target set and the pixel target size occupied by the image are obtained; Based on the servo angle and distance dimensions occupied by different frame dimensions, the actual size of the target is calculated, and the distribution map of the actual area of ​​the target and the sea clutter is used to determine whether the group region is sea clutter or the target. The power intensity of the actual target is determined by the size of the area occupied and the depth of the grayscale image, thus obtaining the overall sea state level of the target.

[0018] In a second aspect, the present invention provides a computer-readable storage medium storing computer instructions which are executed by a processor using the method described above.

[0019] Thirdly, the present invention provides a computer program product, the computer program product storing computer instructions, the computer instructions being executed by a processor using the method described above.

[0020] The beneficial effects achieved by this invention are as follows: This invention provides a radar target identification method based on image classification in sea clutter environments, which significantly improves the radar system's ability to detect and distinguish effective targets in sea clutter meteorological environments. Its main advantages and applications are as follows: 1. By combining image recognition algorithms to identify targets, the accuracy of target identification is improved. This invention performs joint clustering of power amplitude in radar echoes within a complete circle range with physical feature data from adjacent frames. It does not rely on single-dimensional data features but combines them with surrounding attribute features, which greatly improves the ability to distinguish real targets in sea clutter environments and effectively reduces the false alarm rate.

[0021] 2. Adaptive clutter identification in varying environments and sea states: This invention can obtain the current sea state in real time based on the returned signal strength, and obtain sea state maps under different Beaufort wind levels for adaptive threshold adjustment, thereby improving the algorithm's generalization ability and environmental adaptability.

[0022] 3. High compatibility and engineering practicality: This invention has low requirements for radar hardware and data format, and can be easily integrated into existing engineering code. It can be applied to GPU programming and improve the detection of various targets in sea clutter scenarios without affecting normal detection. Comprehensive decision-making improves the reliability of target recognition.

[0023] 4. Intelligent decision-making and signal link optimization: This invention can integrate adaptive sea state discrimination, image recognition and other algorithms to achieve full adaptation from signal processing to multi-target screening. At the same time, it can also reduce the false alarm rate of the system, improve the continuous monitoring and remote unattended operation capabilities, and meet the high-level requirements of modern intelligent sensing systems. Attached Figure Description

[0024] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the drawings without creative effort.

[0025] Figure 1 This is a flowchart of the radar target recognition method of the present invention.

[0026] Figure 2 This is a statistical distribution diagram of sea clutter in this invention.

[0027] Figure 3 This is a diagram showing the result of grayscale processing according to the present invention.

[0028] Figure 4 This is a diagram showing the result of the opening operation of this invention.

[0029] Figure 5 This is a diagram showing the result of the closing operation in this invention.

[0030] Figure 6 This is a diagram showing the connected regions and target detection results of the present invention. Detailed Implementation

[0031] The embodiments of the technical solution of the present invention will now be described in detail with reference to the accompanying drawings. These embodiments are merely illustrative of the technical solution of the present invention and are therefore intended to limit the scope of protection of the present invention.

[0032] Unless otherwise defined, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains; the terminology used in this invention is for the purpose of describing particular embodiments only and is not intended to limit the invention; the terms “comprising” and “having”, and any variations thereof, in the specification, claims and foregoing description of the drawings are intended to cover non-exclusive inclusion.

[0033] In the description of the embodiments of this invention, technical terms such as "first" and "second" are used only to distinguish different objects and should not be construed as indicating or implying relative importance or implicitly specifying the number, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this invention, "multiple" means two or more, unless otherwise explicitly defined.

[0034] In this invention, the reference to "embodiment" means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least some embodiments of the invention. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described in this invention can be combined with other embodiments.

[0035] In the description of the embodiments of this invention, the term "and / or" is merely a description of the relationship between associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, in this invention, the character " / " generally indicates that the preceding and following associated objects have an "or" relationship.

[0036] To effectively suppress the impact of sea clutter on radar target detection and improve the stability and accuracy of radar system target detection under complex weather conditions, this invention provides a radar target recognition method based on image classification in sea clutter environments, such as... Figure 1 As shown, the method includes the following steps: S1. Continuous data acquisition and data preprocessing; specifically including: S1.1 Acquire radar observation data for each frame within the ring scan range, perform signal preprocessing on the raw echo signal including pulse compression, moving target indication (MTI), filtering, noise reduction and amplitude and phase correction, and then perform non-coherent accumulation to obtain the amplitude data of each point trace. S1.2. Perform outlier processing on the amplitude data and remove abnormally low value points where the power spectral density is lower than a preset threshold. S1.3. Convert the preprocessed amplitude data to grayscale to form a grayscale image. S2. Perform morphological opening and morphological closing operations on the grayscale image in sequence to suppress image noise and fill in the holes inside the target. S3. Perform edge detection on the image processed in step S2 to enhance the target edge information; S4. Perform connected component analysis on the image data processed in step S3. Based on the 4-neighborhood connectivity rule, use the scan line method to mark and count connected components in order to identify radar targets.

[0037] In some embodiments, step S1.1 specifically includes: for each frame of radar observation data within the ring scan range, performing signal processing and non-coherent accumulation to obtain the amplitude at each point:

[0038]

[0039] Where N is the number of target points in the ring scan, and M is the total number of sampling points per frame. The magnitude at each point.

[0040] The signal processing operation includes receiving the echo signal reflected back from the target object by the radar antenna, and then performing signal preprocessing operations such as pulse compression, moving target indication (MTI), filtering, denoising, amplitude and phase correction on the received raw echo signal to enhance the weak target signal, suppress background clutter and improve the accuracy of subsequent feature extraction.

[0041] Further, step S1.2 specifically includes: setting a power spectral density threshold. The amplitude data below the threshold The values ​​were identified as abnormally low and removed.

[0042] For each frame, due to zero-padding interpolation, the total energy is distributed across more frequency points, resulting in a decrease in the power spectral density value at individual frequency points, thus causing outliers.

[0043] Among them, the power spectral density threshold It can be adjusted according to the actual situation. For each power density.

[0044] Further, step S1.3 specifically includes: calculating the maximum and minimum values ​​of all amplitude data, mapping each amplitude data to a grayscale range of 0-255 through linear scaling, and rounding to obtain the grayscale value.

[0045] The amplitude of the acquired radar sweep data is converted to grayscale:

[0046] Calculate its maximum and minimum values ​​as the scaling factor, using the following formula:

[0047] in, For the point trace amplitude data, min(D) and max(D) are the minimum amplitude and maximum amplitude in the frame data, respectively. Round() means rounding to the nearest integer, that is, if the calculated new grayscale value contains decimals, it is rounded to the nearest integer.

[0048] In some embodiments, in step S2, the morphological opening operation is a dilation followed by erosion operation, and the morphological closing operation is an erosion followed by dilation operation; the dilation operation is as follows: a structuring element of a preset shape and size is slid across the image, and when any pixel covered by the structuring element is foreground, the center pixel is retained as foreground; otherwise, the center pixel is dilated to become background; the erosion operation is as follows: a structuring element of a preset shape and size is slid across the image, and when all pixels covered by the structuring element are foreground, the center pixel is retained as foreground; otherwise, the center pixel is eroded to become background.

[0049] Optionally, the shape of the structural element can be a rectangle, a circle, etc., and the size of the structural element is 3.

[0050] In this operation, the opening operation, which performs dilation before erosion, helps to close small holes inside the foreground object and small black spots on the object, effectively filling internal voids. The closing operation, which performs erosion before dilation, helps to remove noise and can eliminate extremely small internal targets. In this embodiment, the opening operation is performed first, followed by the closing operation.

[0051] In some embodiments, in step S3, the edge detection employs a Laplacian filtering algorithm, using a preset 4-neighborhood Laplacian convolution kernel to perform convolution operations on the image to sharpen the edges.

[0052] For radar target detection, Laplacian filtering is an image processing technique used to enhance image edges, making the image clearer. It is based on second-order differentiation, achieving a sharpening effect by calculating the difference in gray levels of pixels within the image's neighborhood. Laplacian filters can be implemented using different convolution kernels; common kernels include 4-neighbor and 8-neighbor templates. In this embodiment, a 4-neighbor kernel is selected, and the chosen kernel is:

[0053] By applying the convolution kernel to each pixel of the image and adding the calculation result to the original pixel value, a sharpened image is obtained. During the processing, negative values ​​are handled and the grayscale range of the resulting image is limited to avoid grayscale overflow. During the convolution process, padding with zeros is used at the edge positions.

[0054] In some embodiments, step S4, calculating connected components, is a commonly used technique in image analysis to identify sets of adjacent pixels with the same pixel value in an image. The core principle is to assign a unique label to each independent connected region for subsequent calculations, etc.

[0055] In this embodiment, a 4-neighborhood is selected, encompassing pixels in the top, bottom, left, and right directions. Common connected component calculations employ a filling method to iteratively or recursively expand the foreground of all neighboring areas, marking them as a single connected component. However, in real-world engineering applications, the real-time processing performance requirements for each frame are higher. In a GPU-based environment, a scan-line method is used, completing the marking process in two scans to obtain the final result.

[0056] The scan-line method for connected component labeling and statistics specifically includes: Step 1: Scan line by line; Scan each row of pixels from left to right and from top to bottom. If the current pixel is foreground, check the labels of the pixels to its left and top. If neither is labeled, assign a new temporary label to the current pixel. If only one side is labeled, the current pixel inherits that label. If both sides are labeled, the current pixel inherits one of the labels and records that the two labels are equivalent. Step 2: Merge equivalent tags to generate the final tag; organize the equivalent tag pairs recorded in Step 1, merge the equivalent tags using a disjoint-set data structure to obtain a unique final tag; traverse the image and replace the temporary tag of each pixel with the merged final tag.

[0057] Using this method in conjunction with GPU programming, a large image generated after completing a full circle of data in a sea clutter environment has a linear time complexity of O (width×height), low memory usage, and higher execution efficiency.

[0058] The final count includes the number of neighborhoods, the area occupied by each neighborhood, and the centroid of each neighborhood.

[0059] Specifically, the present invention also provides a specific embodiment: for a sea clutter radar signal with a range gate number of Nr of 13350 and a pulse number of Nf of 12, the following processing is performed.

[0060] (a) Basic signal processing and data preprocessing Signal processing operations include receiving the echo signal reflected back from the target object by the radar antenna, and then expanding the original data dimensions to Nr = 16384 and Nf = 16 to prepare for subsequent Fast Fourier Transform (FFT). This includes pulse compression, Moving Target Indication (MTI), filtering, denoising, amplitude and phase correction, etc.

[0061]

[0062] Perform a Fourier transform on the expanded target echo signal to convert the time-domain signal... ( Transform to the frequency domain, denoted as

[0063] Calculate the power spectral density of the signal by taking the square of the modulus of the frequency domain signal and adding a local minimum value Esp=1e-6 to avoid calculation errors.

[0064]

[0065] Finally, non-coherent accumulation is performed to obtain the final data.

[0066]

[0067] 2. Outlier Handling For each frame, due to zero-padding interpolation, the total energy is distributed across more frequency points, resulting in a decrease in the power spectral density value at individual frequency points, thus causing outliers.

[0068] Where the threshold The threshold in this embodiment will be adjusted according to the actual situation. Using the mean of 7000-8000 points as the average noise floor, the data was plotted under different distributions to obtain... Figure 2 Determine whether the distribution curve of sea clutter is satisfied.

[0069] 3. Grayscale processing The processed data is stored in a cache, and the total number of ring scan frames is 450. The amplitude of the obtained ring scan data is then converted to grayscale, as follows: Figure 3 As shown:

[0070] Calculate its maximum and minimum values ​​as the scaling factor. The calculation formula is as follows:

[0071] If the calculated new grayscale value contains decimals, it should be rounded to the nearest integer to obtain the total frame data size.

[0072]

[0073] (ii) Opening and closing operations In this embodiment, the opening operation is performed first, followed by the closing operation, such as... Figure 4 and Figure 5 As shown. In this embodiment, both opening and closing operations use a three-dimensional kernel. The selected opening and closing kernels are:

[0074] (III) Edge Detection For radar target detection, Laplacian filtering is an image processing technique used to enhance image edges, making the image clearer. In this embodiment, a 4-neighborhood filter is selected, and the chosen convolution kernel is:

[0075] This convolution kernel is applied to each pixel of the image, and the result is added to the original pixel value.

[0076] (iv) Statistical calculation of connected components In this embodiment, a 4-neighborhood is selected, encompassing pixels in the top, bottom, left, and right directions. The scan-line method is employed, completing the marking process in two scans to obtain the final result, resulting in the final neighborhood set and target set, as shown below. Figure 6 As shown.

[0077] Based on the size of the area and the centroid of the target, combined with the target obtained through conventional signal processing, the target set and the pixel size of the target in the image can be obtained.

[0078] Based on the servo angle and distance dimensions occupied by different frame dimensions, the actual size of the target is calculated. Based on the distribution map of the actual area of ​​the target and the sea clutter, it can be determined whether the group area is sea clutter or the target.

[0079] The power intensity of the actual target can be determined by the size of the area occupied and the depth of the grayscale image, thus obtaining the overall sea state level of the target.

[0080] In summary, the radar target identification method based on image classification in a sea clutter environment provided by this invention significantly improves the radar system's ability to detect and distinguish effective targets in sea clutter meteorological environments. Its main advantages and applications are as follows: 1. By combining image recognition algorithms to identify targets, the accuracy of target identification is improved. This invention performs joint clustering of power amplitude in radar echoes within a complete circle range with physical feature data from adjacent frames. It does not rely on single-dimensional data features but combines them with surrounding attribute features, which greatly improves the ability to distinguish real targets in sea clutter environments and effectively reduces the false alarm rate.

[0081] 2. Adaptive clutter identification in varying environments and sea states: This invention can obtain the current sea state in real time based on the returned signal strength, and obtain sea state maps under different Beaufort wind levels for adaptive threshold adjustment, thereby improving the algorithm's generalization ability and environmental adaptability.

[0082] 3. High compatibility and engineering practicality: This invention has low requirements for radar hardware and data format, and can be easily integrated into existing engineering code. It can be applied to GPU programming and improve the detection of various targets in sea clutter scenarios without affecting normal detection. Comprehensive decision-making improves the reliability of target recognition.

[0083] 4. Intelligent decision-making and signal link optimization: This invention can integrate adaptive sea state discrimination, image recognition and other algorithms to achieve full adaptation from signal processing to multi-target screening. At the same time, it can also reduce the false alarm rate of the system, improve the continuous monitoring and remote unattended operation capabilities, and meet the high-level requirements of modern intelligent sensing systems.

[0084] In some embodiments, the present invention provides a computer-readable storage medium storing computer instructions that are executed by a processor as described in any of the above embodiments, for a radar target identification method based on image classification in a sea clutter environment.

[0085] Computer-readable storage media may take the form of any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may, for example, include, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples (not an exhaustive list) of readable storage media may include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0086] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0087] Embodiments of the present invention may also be computer program products, which include computer program instructions that, when executed by a processor, cause the processor to perform the steps in a radar target recognition method based on image classification in a sea clutter environment according to various embodiments of the present invention, as described in the "Exemplary Methods" section above.

[0088] The steps of the method of the present invention are not limited to the specific order described above, unless otherwise specifically stated. Furthermore, in some embodiments, the invention may also be implemented as a program recorded on a recording medium, the program comprising machine-readable instructions for implementing the method according to the invention. Therefore, the invention also covers recording media storing programs for performing the method according to the invention.

[0089] Although the invention has been described with reference to preferred embodiments, various modifications can be made and components can be replaced with equivalents without departing from the scope of the invention. In particular, the technical features mentioned in the various embodiments can be combined in any manner as long as there is no structural conflict. The invention is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A radar target recognition method based on image classification in a sea clutter environment, characterized in that, Includes the following steps: S1. Continuous data acquisition and data preprocessing; specifically including: S1.1 Acquire radar observation data for each frame within the ring scan range, perform signal preprocessing on the raw echo signal including pulse compression, moving target display, filtering, noise reduction and amplitude and phase correction, and then perform non-coherent accumulation to obtain the amplitude data of each point trace; S1.

2. Perform outlier processing on the amplitude data and remove abnormally low value points where the power spectral density is lower than a preset threshold. S1.

3. Convert the preprocessed amplitude data to grayscale to form a grayscale image. S2. Perform morphological opening and morphological closing operations on the grayscale image in sequence to suppress image noise and fill in the holes inside the target. S3. Perform edge detection on the image processed in step S2 to enhance the target edge information; S4. Perform connected component analysis on the image data processed in step S3. Based on the 4-neighborhood connectivity rule, use the scan line method to mark and count connected components in order to identify radar targets.

2. The radar target recognition method based on image classification in a sea clutter environment according to claim 1, characterized in that, Step S1.2 specifically includes: setting a power spectral density threshold. The amplitude data below the threshold The values ​​were identified as abnormally low and removed.

3. The radar target recognition method based on image classification in a sea clutter environment according to claim 1, characterized in that, Step S1.3 specifically includes: converting the amplitude of the obtained radar sweep observation data into grayscale. Calculate its maximum and minimum values ​​as the scaling factor. The calculation formula is as follows: in, For the amplitude data of the dot trace, min(D) and max(D) are the minimum amplitude and maximum amplitude in the data frame, respectively. Round() represents the rounding operation.

4. The radar target recognition method based on image classification in a sea clutter environment according to claim 1, characterized in that, In step S2, the morphological opening operation is a dilation-erosion operation, and the morphological closing operation is an erosion-dilation operation. The dilation operation involves sliding a structuring element of a preset shape and size across the image. When any pixel covered by the structuring element is foreground, the center pixel is retained as foreground; otherwise, the center pixel is dilated to become background. The erosion operation involves sliding a structuring element of a preset shape and size across the image. When all pixels covered by the structuring element are foreground, the center pixel is retained as foreground; otherwise, the center pixel is eroded to become background. The kernels for both the morphological opening and closing operations are: 。 5. The radar target recognition method based on image classification in a sea clutter environment according to claim 4, characterized in that, The shape of the structural element includes rectangle and circle, and the size of the structural element is 3.

6. The radar target recognition method based on image classification in a sea clutter environment according to claim 1, characterized in that, In step S3, the edge detection employs a Laplacian filtering algorithm, using a preset 4-neighborhood Laplacian convolution kernel to perform convolution operations on the image to sharpen the edges. The convolution kernel is: 。 7. The radar target recognition method based on image classification in a sea clutter environment according to claim 1, characterized in that, In step S4, a 4-neighborhood is selected, which includes pixels in the top, bottom, left, and right directions. In the GPU running environment, the scan line method is used to complete the marking through two scans to obtain the final result. The method of using the scan line method for connected component labeling and statistics specifically includes: Step 1: Scan line by line; Scan each row of pixels from left to right and from top to bottom. If the current pixel is foreground, check the labels of the pixels to its left and top. If neither is labeled, assign a new temporary label to the current pixel. If only one side is labeled, the current pixel inherits that label. If both sides are labeled, the current pixel inherits one of the labels and records that the two labels are equivalent. Step 2: Merge equivalent tags to generate the final tag; organize the equivalent tag pairs recorded in Step 1, merge the equivalent tags using a disjoint-set data structure to obtain a unique final tag; traverse the image and replace the temporary tag of each pixel with the merged final tag.

8. The radar target recognition method based on image classification in a sea clutter environment according to claim 7, characterized in that, The radar target identification method also includes: Based on the size of the area and the centroid of the target, combined with the target obtained by conventional signal processing, the target set and the pixel target size occupied by the image are obtained; Based on the servo angle and distance dimensions occupied by different frame dimensions, the actual size of the target is calculated, and the distribution map of the actual area of ​​the target and the sea clutter is used to determine whether the group region is sea clutter or the target. The power intensity of the actual target is determined by the size of the area occupied and the depth of the grayscale image, thus obtaining the overall sea state level of the target.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that are executed by a processor according to any one of claims 1 to 8.

10. A computer program product, characterized in that, The computer program product stores computer instructions, which are executed by a processor using the method as described in any one of claims 1 to 8.

Images