A ship target classification and recognition method based on ISAR image feature processing

Abstract

The present invention discloses a ship target classification and recognition method based on ISAR image feature processing, and belongs to the technical field of sea detection by unmanned aerial vehicle reconnaissance and surveillance radar. Compared with the traditional method of classifying and identifying ship targets by manually judging ISAR images, which is prone to human misjudgment, low classification efficiency, low recognition accuracy, etc., the present invention adopts a ship target classification and recognition method based on ISAR image feature processing, obtains the classification result of the ship target through corresponding processing, can efficiently and conveniently solve the problem of ship target classification and recognition, improves the recognition accuracy and work efficiency, and greatly reduces human resources.

Description

Technical Field

[0001] The invention relates to the technical field of sea detection by unmanned aerial vehicle (UAV)-borne reconnaissance and surveillance radar, and in particular to a ship target classification and recognition method based on ISAR image feature processing. Background Art

[0002] With the rapid development of electronic communication technology, the reconnaissance and surveillance radar technology suitable for UAV platforms has developed rapidly, and its functions and performance have been continuously enhanced, greatly improving its combat effectiveness and survivability in the future unmanned battlefield. The UAV-borne reconnaissance and surveillance radar system has the characteristics of all-day, all-weather, long-distance, and high resolution, and is one of the indispensable intelligence acquisition equipment for future unmanned combat scenarios. However, in the field of sea detection by UAV-borne reconnaissance and surveillance radar, due to factors such as the wide variety of sea surface targets and the complex and changeable sea surface environment, it is particularly difficult to classify and identify radar detection targets against the sea surface background.

[0003] The above problems need to be solved urgently. To this end, the present invention provides a ship target classification and recognition method based on ISAR image feature processing. Summary of the invention

[0004] The technical problem to be solved by the present invention is how to solve the problem of automatic classification and identification of different ship targets in sea detection scenarios by existing unmanned aerial vehicle reconnaissance and surveillance radar systems, and provides a ship target classification and identification method based on ISAR image feature processing.

[0005] The present invention solves the above technical problems through the following technical solutions, and the present invention comprises the following steps:

[0006] S1: In the sea surface ship target detection mode of the UAV-borne reconnaissance and surveillance radar system, the ISAR imaging processing of the moving ship target is performed using a broadband signal beam to obtain the ship ISAR image;

[0007] S2: By processing the acquired ship ISAR image, the ship target characteristic parameters are extracted and the length and width measurement values ​​of the ship target are calculated;

[0008] S3: Classify and identify the ship target based on the calculated length and width measurements and the preset classification criteria.

[0009] Furthermore, the unmanned aerial vehicle reconnaissance and surveillance radar system includes an antenna unit and a processing unit; the antenna unit is used to control the pointing of the radar beam to achieve two-dimensional electronic scanning in azimuth and elevation, generate and radiate broadband electromagnetic waves and receive electromagnetic wave signals reflected by moving ship targets, and the antenna unit includes X×Y transceiver channels, where X represents the number of transceiver channels in azimuth, and Y represents the number of transceiver channels in distance; the processing unit is used to perform ISAR imaging processing on the received electromagnetic wave signals, and perform feature extraction, ship target length and width detection processing on the generated ISAR images.

[0010] Furthermore, in step S2, the specific processing process is as follows:

[0011] S21: performing ship target slicing processing on the generated ISAR image with a size of 1024×1024 to obtain a ship slice ISAR image with a size of 256×256 for ship feature extraction processing;

[0012] S22: preprocessing the ship slice ISAR image, including noise removal and fringe suppression processing;

[0013] S23: Perform contour extraction on the preprocessed ship slice ISAR image to obtain the ship range and azimuth contour areas, and calculate the length and width measurement values ​​of the ship target through the image resolution and the pixel points occupied by the ship target contour sequence.

[0014] Furthermore, in step S22, the noise removal of the ship slice ISAR image is performed based on a global threshold method, and the specific processing process is as follows:

[0015] S2211: Based on the fact that the target is at the center of the ISAR image and the image size is larger than 4 times the target size, the target area is eliminated, and the noise scattering area σ0 of the background area of ​​the ISAR image is counted to obtain its mean μ and standard deviation σ;

[0016] S2212: Calculate the threshold value T = μ + k × σ, where k is an empirical constant;

[0017] S2213: Backscatter value x for each pixel in the ship slice ISAR image i,j After threshold processing, the ISAR image with noise removed is obtained.

[0018] Furthermore, in step S2213, the threshold-crossing processing formula is as follows:

[0019]

[0020] Among them, x i,jrepresents the backscatter value of the corresponding pixel in the ISAR image, p i,j It represents the backscatter value of each pixel in the ISAR image after noise removal.

[0021] Furthermore, in step S22, when the window function is determined, the specific process of stripe suppression is as follows:

[0022] S2221: Calculate the peak sidelobe ratio α according to the window function added during ISAR imaging;

[0023] S2222: Calculate the amplitude of the strong scattering point in the distance unit i;

[0024] S2223: Determine and calculate a threshold G for fringe suppression in distance unit i according to the size of the side lobe, perform fringe suppression in the distance unit according to the threshold, and set points smaller than the threshold to zero;

[0025] S2224: Repeat the above steps for all distance units;

[0026] S2225: Repeat the processing in the above steps S2221-S2224 for all orientation units.

[0027] Furthermore, in step S2223, the calculation formula of the threshold G for stripe suppression is as follows:

[0028] G=α×z

[0029] Among them, α is the peak sidelobe ratio, which is calculated by the window function; z is the amplitude value of the strong scattering point.

[0030] Furthermore, in step S23, the specific steps of obtaining the ship target profile sequence are as follows:

[0031] S2311: Design a circle with a radius equal to the length of a set number of units as a morphological unit, perform two dilation operations on the ISAR image, and connect the targets distributed as scattered points into continuous targets;

[0032] S2312: Then perform two more corrosion operations to restore the target to its original outline;

[0033] S2313: Screen the ISAR image content, count the areas of each continuous part in the image, and only keep the part with the largest area as the target;

[0034] S2314: Use Canny operator to extract the outline of the ship target;

[0035] S2315: Eliminate the curve parts that are not connected to the ship target contour, calculate the length of each curve, and only retain the longest part as the ship target contour;

[0036] S2316: Take the pixel point in the upper left corner of the contour sequence as the starting point, and search the eight neighborhoods of the pixel point in a clockwise direction from the starting point to see if there are other pixels. If there are, keep the horizontal and vertical coordinates of the new pixel point, and continue to search with the new pixel point as the center pixel point until the coordinates of all the pixels in the contour sequence are saved to obtain the ship target contour sequence, including N a The pixels in the distance direction and N b Pixels in the direction.

[0037] Furthermore, in step S23, the specific process of calculating the measured values ​​of the target length and width of the ship is as follows:

[0038] S2321: Get range resolution ρ a =c / 2B×cosθ, in the same mode, the azimuth resolution is considered equal to the range resolution, where B is the bandwidth of the wideband signal transmitted by the ISAR mode radar, θ is the angle between the radar line of sight and the target moving direction, and c is the speed of light;

[0039] S2322: Calculate the ship target length L according to the ship target profile sequence a 、Width L b The measured values ​​and calculation formulas are as follows:

[0040] L a =N a ×ρ a ;

[0041] L b =N b ×ρ b .

[0042] Furthermore, in step S3, the calculated length and width measurement values ​​of the ship target are compared with the set large, medium and small length and width thresholds of the ship target to achieve ship target classification and identification.

[0043] Compared with the prior art, the present invention has the following advantages: compared with the traditional ship target classification and identification method of manually judging ISAR images, which is easy to cause human misjudgment, low classification efficiency, low recognition accuracy and other problems, the ship target classification and identification method based on ISAR image feature processing adopts the ship target classification and identification method based on ISAR image feature processing, and obtains the classification result of the ship target through corresponding processing, which can efficiently and conveniently solve the problem of ship target classification and identification, improve the recognition accuracy and work efficiency, and greatly reduce human resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 12 is a schematic diagram of the structure of a UAV-borne reconnaissance and surveillance radar system according to an embodiment of the present invention;

[0045] Figure 2 This is the ISAR image preprocessing process in an embodiment of the present invention;

[0046] Figure 3 1 is a schematic diagram of a process for extracting ship contour features based on ISAR image segmentation in an embodiment of the present invention;

[0047] Figure 4 It is a schematic diagram of the process of ship target classification and identification in an embodiment of the present invention. DETAILED DESCRIPTION

[0048] The following is a detailed description of an embodiment of the present invention. This embodiment is implemented on the premise of the technical solution of the present invention, and a detailed implementation method and a specific operation process are given, but the protection scope of the present invention is not limited to the following embodiment.

[0049] like Figure 1 As shown, it is a schematic diagram of the composition of the unmanned aerial vehicle reconnaissance and surveillance radar system in the present invention. The unmanned aerial vehicle reconnaissance and surveillance radar system is installed on an unmanned aerial vehicle (carrier platform). The system includes an antenna unit and a processing unit, wherein the antenna unit is used to control the direction of the radar beam to realize azimuth and elevation two-dimensional electronic scanning, generate and radiate broadband electromagnetic waves and receive electromagnetic wave signals reflected by ship targets; the processing unit is used to perform ISAR (high-resolution inverse synthetic aperture) imaging processing on the received electromagnetic wave signals, and perform feature extraction and ship target length and width detection processing on the generated ISAR images.

[0050] In this embodiment, the antenna unit includes X×Y transmitting and receiving channels in azimuth and elevation.

[0051] like Figure 2 As shown in FIG. 1 , the ISAR image preprocessing process of the ship target obtained by the present invention mainly includes noise removal and fringe suppression, which can improve the quality of the ISAR image.

[0052] As more specific, in this embodiment, the noise removal of the ship slice ISAR image is performed based on the global threshold method, and the specific processing process is as follows:

[0053] 1): Based on the fact that the target is in the center of the ISAR image and the image size is larger than 4 times the target size, the target area is eliminated, and the noise scattering area σ0 of the background area of ​​the ISAR image is counted to obtain its mean μ and standard deviation σ;

[0054] 2): Calculate the threshold value T = μ + k × σ, where k is an empirical constant;

[0055] 3): The scattering value x for each pixel in the ship slice ISAR image i,j After threshold processing, the ISAR image with noise removed is obtained.

[0056] As a more specific example, the threshold processing formula is as follows:

[0057]

[0058] Among them, x i,j Represents the backscattering value of the corresponding pixel, p i,j The backscatter value of each pixel in the ISAR image after noise removal.

[0059] It should be noted that, in general, in ISAR images, most of the fringe interference comes from the influence of the side lobes of the target strong scattering points. The windowing method is used to suppress the side lobes. When calculating the size G of the side lobe, the influence of the window function needs to be considered, and the side lobe can be obtained by the following formula:

[0060] G=α×z

[0061] Where: α is the peak-to-sidelobe ratio, and z is the amplitude value of the strong scattering point.

[0062] In this embodiment, a strong scattering point refers to a reflection point in an area such as the bow or hull of a ship where the size of the ship target is larger than the wavelength λ of the radar electromagnetic wave signal. The radar electromagnetic wave signal amplitude of the reflection point in such a target area is strong, which is reflected in the large backscattering value corresponding to the pixel point in the ISAR image.

[0063] When the window function is determined during the fringe suppression process, the peak sidelobe ratio can be calculated. By measuring the amplitude of the strong scattering point, the size of the sidelobe can be obtained, that is, the threshold of fringe suppression. The specific process of fringe suppression is as follows:

[0064] 1): Calculate the peak sidelobe ratio α according to the window function added during ISAR imaging;

[0065] 2): Calculate the amplitude value of the strong scattering point in distance unit i;

[0066] 3): Determine the threshold G for calculating fringe suppression in distance unit i according to the size of the side lobe, perform fringe suppression in the distance unit according to the threshold, and set the points smaller than the threshold to zero;

[0067] 4): Repeat the above steps for all distance units;

[0068] 5): Repeat the above process for all orientation units.

[0069] like Figure 3As shown in the figure, the ship contour extraction process based on ISAR image segmentation of the present invention is carried out. The pre-processed ISAR image is processed by contour feature extraction to obtain the ship target distance and azimuth shape area, and the ship target length and width measurement values ​​are calculated by image resolution and pixel points occupied by the ship target contour sequence.

[0070] It should be noted that the edge in the image is manifested as discontinuities in color, texture, and brightness. Finding these discontinuities is finding the edge of the target. The targets in ISAR images are distributed as discrete, isolated scattering points, and there is speckle noise. Directly smoothing and extracting the edge of the target will result in multiple closed ellipses and multiple non-smooth curves, and the contour of the target cannot be fully extracted.

[0071] In this embodiment, according to the characteristics of ISAR images, the image is first dilated and eroded and then the Canny operator is used to extract the edge. The specific steps of obtaining the ship target contour sequence are as follows:

[0072] 1): A circle with a radius of 6 unit lengths (one unit length corresponds to the range resolution of the ISAR image) is designed as the morphological unit, and the ISAR image is expanded twice to connect the targets distributed as scattered points into continuous targets;

[0073] 2): Then perform two more corrosion operations to restore the target to its original outline;

[0074] 3): Filter the ISAR image content, count the areas of each continuous part in the image, and only retain the part with the largest area as the target;

[0075] 4): Use Canny operator to extract the outline of the ship target;

[0076] 5): Eliminate the curve parts that are not connected to the ship target contour, calculate the length of each curve, and only retain the longest part as the ship target contour;

[0077] 6): Take the pixel point in the upper left corner of the contour sequence as the starting point, and search the eight neighborhoods of the pixel point in a clockwise direction from the starting point to see if there are other pixels. If there are, keep the horizontal and vertical coordinates of the new pixel point, and continue to search with the new pixel point as the center pixel point until the coordinates of all the pixels in the contour sequence are saved to obtain the ship target contour sequence, including N a The pixels in the distance direction and N b Pixels in the direction.

[0078] In this embodiment, the specific process of calculating the measured values ​​of the ship target length and width is as follows:

[0079] 1): For ISAR mode radar, the bandwidth of the transmitted broadband signal is B, and the range resolution is ρ a =c / 2B×cosθ, generally the same mode can be approximately considered as the azimuth resolution ρ b ≈ρ a ;

[0080] 2): Calculate the ship length (distance) L according to the ship target profile sequence a , Width (direction) L b Measurements:

[0081] L a =N a ×ρ a

[0082] L b =N b ×ρ b

[0083] Among them, c is the speed of light, θ is the angle between the radar line of sight and the target movement direction, and the ship movement direction and the bow direction are approximately considered to be consistent.

[0084] like Figure 4 As shown in the figure, it is the process of ship target classification and recognition of the present invention, for the calculated ISAR ship target length and width, the ship target large, medium and small classification and recognition results are given according to the preset classification criteria.

[0085] The calculation and processing processes described in the present invention are all implemented by software design, and the measured values ​​of the length and width of the ship target are automatically obtained through software processing, and large, medium and small classification results are given according to classification criteria.

[0086] In this embodiment, the calculated length and width measurement values ​​of the ship target are compared with the set large, medium and small length and width thresholds of the ship target to achieve ship target classification and identification.

[0087] In summary, the ship target classification and recognition method based on ISAR image feature processing in the above embodiment is compared with the traditional method of classifying and recognizing ship targets by manually judging ISAR images, which is prone to human misjudgment, low classification efficiency, low recognition accuracy and other problems. The ship target classification and recognition method based on ISAR image feature processing is adopted, and the classification result of the ship target is obtained through corresponding processing, which can efficiently and conveniently solve the problem of ship target classification and recognition, improve the recognition accuracy and work efficiency, and greatly reduce human resources.

[0088] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A ship target classification and recognition method based on ISAR image feature processing, characterized in that: The following steps are involved: S1: In the sea surface ship target detection mode of the UAV-borne reconnaissance and surveillance radar system, the ISAR imaging processing of the moving ship target is performed using a broadband signal beam to obtain the ship ISAR image; S2: By processing the acquired ship ISAR image, the ship target characteristic parameters are extracted and the length and width measurement values ​​of the ship target are calculated; In step S2, the specific processing process is as follows: S21: performing ship target slicing processing on the generated ISAR image with a size of 1024×1024 to obtain a ship slice ISAR image with a size of 256×256 for ship feature extraction processing; S22: preprocessing the ship slice ISAR image, including noise removal and fringe suppression processing; S23: extracting the contour of the preprocessed ship slice ISAR image, obtaining the ship's range and azimuth contour areas, and calculating the length and width measurement values ​​of the ship target through the image resolution and the pixel points occupied by the ship target contour sequence; S3: Classify and identify the ship target according to the preset classification criteria based on the calculated length and width measurement values ​​of the ship target; In step S23, the specific steps of obtaining the ship target profile sequence are as follows: S2311: Design a circle with a radius equal to the length of a set number of units as a morphological unit, perform two dilation operations on the ISAR image, and connect the targets distributed as scattered points into continuous targets; S2312: Then perform two more corrosion operations to restore the target to its original outline; S2313: Screen the ISAR image content, count the areas of each continuous part in the image, and only keep the part with the largest area as the target; S2314: Use Canny operator to extract the outline of the ship target; S2315: Eliminate the curve parts that are not connected to the ship target contour, calculate the length of each curve, and only retain the longest part as the ship target contour; S2316: Take the pixel point in the upper left corner of the contour sequence as the starting point, and search the eight neighborhoods of the pixel point in a clockwise direction from the starting point to see if there are other pixels. If there are, keep the horizontal and vertical coordinates of the new pixel point, and continue to search with the new pixel point as the center pixel point until the coordinates of all the pixels in the contour sequence are saved to obtain the ship target contour sequence, including N a The pixels in the distance direction and N b Pixels in the azimuth direction; In step S23, the specific process of calculating the measured values ​​of the ship target length and width is as follows: S2321: Get range resolution ρ a =c / 2B×cosθ, in the same mode, the azimuth resolution is considered equal to the range resolution, where B is the bandwidth of the wideband signal transmitted by the ISAR mode radar, θ is the angle between the radar line of sight and the target moving direction, and c is the speed of light; S2322: Calculate the ship target length L according to the ship target profile sequence a 、Width L b The measured values ​​and calculation formulas are as follows: L a =N a ×ρ a ; L b =N b ×ρ b .

2. The ship target classification and recognition method based on ISAR image feature processing according to claim 1 is characterized in that: The unmanned aerial vehicle reconnaissance and surveillance radar system includes an antenna unit and a processing unit; the antenna unit is used to control the direction of the radar beam to realize two-dimensional electronic scanning in azimuth and elevation, generate and radiate broadband electromagnetic waves and receive electromagnetic wave signals reflected by moving ship targets, and the antenna unit includes X×Y transceiver channels, where X represents the number of transceiver channels in azimuth, and Y represents the number of transceiver channels in distance; the processing unit is used to perform ISAR imaging processing on the received electromagnetic wave signals, and perform feature extraction, ship target length and width detection processing on the generated ISAR images.

3. The ship target classification and recognition method based on ISAR image feature processing according to claim 2 is characterized in that: In step S22, the noise removal of the ship slice ISAR image is performed based on the global threshold method, and the specific processing process is as follows: S2211: Based on the fact that the target is at the center of the ISAR image and the image size is larger than 4 times the target size, the target area is eliminated, and the noise scattering area σ0 of the background area of ​​the ISAR image is counted to obtain its mean μ and standard deviation σ; S2212: Calculate the threshold value T = μ + k × σ, where k is an empirical constant; S2213: Backscatter value x for each pixel in the ship slice ISAR image i,j After threshold processing, the ISAR image with noise removed is obtained.

4. The ship target classification and recognition method based on ISAR image feature processing according to claim 3 is characterized in that: In step S2213, the threshold-crossing processing formula is as follows: Among them, x i,j represents the backscatter value of the corresponding pixel in the ISAR image, p i,j It represents the backscatter value of each pixel in the ISAR image after noise removal.

5. The ship target classification and recognition method based on ISAR image feature processing according to claim 3 is characterized in that: In step S22, when the window function is determined, the specific process of stripe suppression is as follows: S2221: Calculate the peak sidelobe ratio α according to the window function added during ISAR imaging; S2222: Calculate the amplitude of the strong scattering point in the distance unit i; S2223: Determine and calculate a threshold G for fringe suppression in distance unit i according to the size of the side lobe, perform fringe suppression in the distance unit according to the threshold, and set points smaller than the threshold to zero; S2224: Repeat the above steps for all distance units; S2225: Repeat the processing in the above steps S2221-S2224 for all orientation units.

6. The ship target classification and recognition method based on ISAR image feature processing according to claim 5 is characterized in that: In step S2223, the calculation formula of the threshold G for stripe suppression is as follows: G=α×z Among them, α is the peak sidelobe ratio, which is calculated by the window function; z is the amplitude value of the strong scattering point.

7. The ship target classification and recognition method based on ISAR image feature processing according to claim 1 is characterized in that: In step S3, the calculated length and width measurement values ​​of the ship target are compared with the set large, medium and small length and width thresholds of the ship target to achieve ship target classification and identification.

Images