A low-resolution SAR image ship detection method based on fuzzy analysis

Abstract

This invention provides a ship detection method for low-resolution SAR images based on fuzzy analysis. By performing land-sea separation technology on the SAR image, it achieves the discrimination and separation of ocean and land regions within the SAR image. Then, the CFAR ship detection method is used to identify and mark ships and false targets in the separated image. Azimuth fuzzy image analysis is used to evaluate the fuzzy position and intensity in the actual image. Finally, based on the fuzzy position and intensity evaluation results, the true ship targets and false targets are analyzed and distinguished, resulting in a low false alarm rate ship detection result with false targets removed. This invention provides a ship detection method for low-resolution spaceborne SAR satellite ship detection that reduces false alarms caused by fuzzy false targets.

Description

Technical Field

[0001] This invention belongs to the field of SAR image analysis and SAR image target detection, specifically relating to a method for ship detection in low-resolution SAR images based on fuzzy analysis. Background Technology

[0002] Synthetic Aperture Radar (SAR) satellites, with their all-weather, multi-polarization, high penetration, and wide coverage characteristics, have become one of the most important information acquisition platforms in the field of remote sensing, and are widely used in agricultural monitoring, topographic mapping, environmental monitoring, disaster relief, and other fields.

[0003] The excellent all-weather, all-time detection capabilities of SAR systems make SAR images a crucial data source for detecting ship targets on the sea surface. Currently, the most extensively researched SAR image-based ship detection method is the Constant False Alarm Rate (CFAR) detection method, which is based on the statistical distribution of clutter. This type of method primarily obtains a specific image intensity threshold by modeling the statistical characteristics of ocean clutter in the target background. The image intensity threshold is then used to distinguish target pixels from background pixels, thereby achieving the identification and detection of ship targets in the image.

[0004] Azimuth ambiguity is caused by the non-ideal azimuth antenna pattern and the finite azimuth sampling frequency. This is because the actual beamwidth of the azimuth antenna is wider than the ideal beamwidth (3dB width of the main lobe of the antenna pattern). At the finite pulse repetition frequency (PRF) sampling, the energy of the antenna pattern sidelobes also enters the SAR echo signal, resulting in azimuth ambiguity after imaging. Azimuth ambiguity leads to the appearance of false targets in the image and reduces the image signal-to-noise ratio.

[0005] Currently, to quickly acquire images of large-scale sea surface scenes, phased array antennas are typically used for electronic scanning (or mechanical antenna pointing) to achieve rapid radar beam scanning, shortening the SAR azimuth synthetic aperture time and obtaining wide-swath, low-azimuth-resolution images. However, the significant reduction in azimuth resolution of SAR images leads to a substantial increase in azimuth blur energy, resulting in severe false targets in the images. These false targets, like the target signals, also appear as sets of bright pixels in the SAR images.

[0006] Conventional CFAR detection methods require sliding a clutter statistical window across the SAR image to classify each pixel. These methods focus on statistical modeling of background clutter and cannot distinguish between real targets and false targets caused by azimuth blur in low-azimuth resolution images, resulting in a large number of false alarms and severely impacting target detection results. The main problems with existing SAR image ship detection methods based on constant false alarm rate (CFAR) detection are:

[0007] Under the weighting of the antenna pattern, the intensity of false targets caused by azimuth ambiguity in SAR images is about 20-30 dB weaker than that of useful targets. For low azimuth resolution SAR systems, the intensity of azimuth ambiguity is high, which corresponds to the good focusing of false targets, and they appear as a set of bright pixels in the image that are consistent with the useful target signal.

[0008] Meanwhile, conventional CFAR ship detection methods primarily model the statistical characteristics of ocean clutter in the target background to obtain specific image intensity thresholds. These thresholds are then used to distinguish target pixels from background pixels, thereby enabling the identification and detection of ship targets in the image. For azimuth-ambiguous false targets in low-resolution SAR images, even if their intensity is relatively weaker than useful targets, it is still much stronger than background ocean clutter. Summary of the Invention

[0009] To address the aforementioned technical problems, this invention provides a ship detection method for low-resolution SAR images based on fuzzy analysis. It combines azimuth fuzzy analysis technology to evaluate false targets in the actual image, integrates land-sea image separation technology with the conventional CFAR ship detection algorithm, and ultimately achieves low false alarm rate target detection for ships in low-resolution SAR images. In this case, after processing by the CFAR ship detection algorithm, azimuth-fuzzy false targets will be detected and identified as ships simultaneously with useful targets, thereby significantly improving the false alarm rate for ship target detection.

[0010] To achieve the above objectives, the present invention adopts the following technical solution:

[0011] A ship detection method based on fuzzy analysis in low-resolution SAR images includes the following steps:

[0012] S1: Increase the intensity distinction between land and ocean regions in SAR images through image grayscale enhancement / attenuation operations, and identify and separate land and ocean regions in SAR images;

[0013] S2: The SAR image after separating the land area in S1 is processed by the constant false alarm rate (CFAR) detection algorithm. The detected ship targets containing real and false targets are marked, the position information of the marked targets in the SAR image is recorded, and the target intensity information of the marked targets is obtained through a statistical averaging window.

[0014] S3: Based on the precise Earth model, orbital parameters, satellite attitude, antenna size and installation angle and position, antenna scanning angle, combined with the SAR system carrier frequency or wavelength, resolution, and system pulse transmission frequency parameters, the position and intensity information of real targets and blurred false targets under different SAR imaging modes and corresponding imaging processes are obtained. The relative position information and relative intensity relationship between real targets and false targets under this imaging mode are further obtained, and the relative intensity comparison threshold is obtained.

[0015] S4: Based on the relative position information and relative intensity relationship between real and false targets in S3, the marked targets in the SAR image in S2 are compared and analyzed and re-marked to obtain the final real target markings in the SAR image, thus achieving ship detection in SAR images with a low false alarm rate.

[0016] Further, S1 includes:

[0017] S11: Using image grayscale enhancement / attenuation operation, by setting the image grayscale threshold, the image pixel weighting coefficient is calculated, and the SAR image is weighted as a whole, so that the high-intensity grayscale areas corresponding to the land area and the low-intensity grayscale areas corresponding to the ocean area in the SAR image are respectively enhanced and attenuated, thereby increasing the intensity distinction between the land area and the ocean area in the SAR image.

[0018] S12: Perform smooth window grayscale histogram calculation on the SAR image obtained in S11 to obtain grayscale statistical information of land and ocean regions in the SAR image;

[0019] S13: Based on the characteristic that the gray intensity of the land area is higher than that of the ocean area, the gray intensity segmentation threshold of land and ocean is evaluated and calculated. Based on the gray intensity segmentation threshold of land and ocean, the SAR image obtained in S11 is segmented to obtain the SAR image of the detected ocean area separated from the land area.

[0020] Further, S2 includes:

[0021] S21: Based on the SAR image resolution and the size of the detected ship target, set the basic parameters of the constant false alarm rate (CFAR) detection algorithm: energy statistical average clutter reference cell number M×N, and number of protection units N. p Target false alarm probability P FA The clutter reference cell number is two-dimensional data, corresponding to the range and azimuth dimensions of the two-dimensional SAR image, respectively.

[0022] S22: Select detection unit i, and estimate the average clutter power intensity P near detection unit i. scat :

[0023]

[0024] And calculate the total number N of reference units corresponding to the current detection unit. ref :

[0025] Total number of reference units N ref = Number of clutter reference elements - Number of protection elements;

[0026] S23: Based on the target false alarm probability P set in step S21 FA Considering the current clutter environment, an arbitrary ship detection algorithm is selected from the Constant False Alarm Rate (CFAR) detection algorithm library, and the threshold factor α of the CFAR detection algorithm is estimated. T Alternatively, you can set the threshold factor α for the constant false alarm rate (CFAR) detection algorithm based on your processing experience. T ;

[0027] S24; Based on the average clutter power intensity P in step S21 scat Set the detection threshold T = α T ·P scat If the intensity of the current detection unit is higher than the detection threshold, the current detection unit is marked as valid (1); otherwise, it is marked as invalid (0).

[0028] S25: Slide through all detection units in the SAR image according to steps S21 to S24 to obtain the binary image of the CFAR detection result;

[0029] S26: Based on the SAR image resolution and the size of the detected ship target, set an effective target size threshold; remove target units smaller than the effective target size threshold from the binary image of the CFAR detection result through erosion operation; restore the image after erosion through dilation operation; mark the target detection bounding box on the restored image to obtain the detection result image, and record the position information of each mark in the SAR image and the maximum intensity information at the mark.

[0030] Further, S3 includes:

[0031] S31: Based on the accurate Earth model, orbital parameters, satellite attitude, antenna size, antenna installation angle and position, and antenna beam scanning angle, obtain the current antenna beam illumination status of the SAR system;

[0032] S32: Taking into account the antenna beam illumination state of the SAR system in S31 and the current imaging mode of the SAR system, the antenna pattern of the SAR system is introduced to obtain the gain relationship of the antenna beam pattern corresponding to the beam scanning in the current imaging mode.

[0033] S33: In the current imaging mode, based on the relative slant range position relationship between the SAR payload and the target, obtain the azimuth time slant range history and azimuth Doppler spectrum of the corresponding target.

[0034] S34: By using the azimuth time slant range history in S33 and combining it with the pulse transmission frequency, the position information of the real target and the false target corresponding to the blurred signal in different imaging modes is obtained. By combining the azimuth Doppler spectrum in S33 with the antenna pattern weighting in S32, the intensity information of the real target and the false target corresponding to the blurred signal in the current imaging mode is obtained. The real target and the false target corresponding to the blurred signal in the current imaging mode are denoted as the real target and the blurred false target, respectively.

[0035] S35: By using the position and intensity information of the real target and the blurred false target corresponding to different imaging modes in S34, obtain the azimuthal relative position and relative intensity information of the blurred false target relative to the corresponding real target of any order.

[0036] Further, S4 includes:

[0037] S41: Decompose the position and intensity information of the marked targets in S2, sort the targets at the same distance position, and perform azimuth position difference operation to obtain the azimuth position difference of the marked targets at the same distance.

[0038] S42: For the marked targets detected at the same distance gate, if the difference in their azimuth position and the relationship between their blur strength and the position of the real target and the blurred false target in the corresponding imaging mode in S3 are consistent with the position of the real target and the blurred false target, then the marked target at the distance gate position is determined to be either a real target or a blurred false target.

[0039] S43: Re-mark all marked targets in the detection result map in S26 to obtain the ship detection results corresponding to the real targets.

[0040] Beneficial effects:

[0041] This invention uses a fuzzy image analysis method to evaluate the fuzzy position and intensity in a real image. Based on the evaluation results of the fuzzy position and intensity, it analyzes, judges, and distinguishes between real and false targets on ships, thereby obtaining a ship detection result with a low false alarm rate after removing fuzzy false targets. Attached Figure Description

[0042] Figure 1 This is a flowchart of the ship detection method based on fuzzy analysis for low-resolution SAR images according to the present invention.

[0043] Figure 2 This is the original low-resolution SAR image with orientation;

[0044] Figure 3a, Figure 3b , Figure 3c This is a magnified view of the original low-resolution SAR image; in which... Figure 3a Corresponding to the left part, Figure 3b The corresponding part, Figure 3c Corresponding to the right part;

[0045] Figure 4 This is a low-resolution SAR image with azimuth after being labeled using the constant false alarm rate (CFAR) method.

[0046] Figure 5a , Figure 5b , Figure 5c This refers to a low-resolution SAR image with azimuth after being labeled using the constant false alarm rate (CFAR) method; among which... Figure 5a Corresponding to the left part, Figure 5b The corresponding part, Figure 5c Corresponding to the right part;

[0047] Figure 6 This is a low-resolution SAR image of the azimuth after relabeling following fuzzy analysis.

[0048] Figure 7a , Figure 7b , Figure 7c This is a low-resolution SAR image of the azimuth orientation relabeled after fuzzy analysis; among which, Figure 7a Corresponding to the left part, Figure 7b The corresponding part, Figure 7c Corresponding to the right part. Detailed Implementation

[0049] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0050] like Figure 1 As shown, the ship detection method based on fuzzy analysis in low-resolution SAR images according to the present invention includes the following steps:

[0051] S1: Increase the intensity distinction between land and ocean regions in SAR images through image grayscale enhancement / attenuation operations, and perform identification and separation of land and ocean regions in SAR images, including:

[0052] S11: Using image grayscale enhancement / attenuation operation, by setting the image grayscale threshold, the image pixel weighting coefficient is calculated, and the SAR image is weighted as a whole, so that the high-intensity grayscale areas corresponding to the land area and the low-intensity grayscale areas corresponding to the ocean area in the SAR image are respectively enhanced and attenuated, and a SAR image with higher intensity distinction between land and ocean areas is obtained.

[0053] S12: Perform smooth window grayscale histogram calculation on the SAR image obtained in S11 to obtain grayscale statistical information of land and ocean regions in the SAR image;

[0054] S13: Based on the characteristic that the gray intensity of the land area is higher than that of the ocean area, the gray intensity segmentation threshold of land and ocean is evaluated and calculated. Based on the gray intensity segmentation threshold, the SAR image obtained in S11 is segmented to obtain the SAR image of the detected ocean area that has been separated from the land area.

[0055] S2: Using the Constant False Alarm Rate (CFAR) detection algorithm, the SAR image after separating the land area in S1 is processed. Detected ship targets containing both real and false targets are marked, and the location information of the marked targets in the SAR image is recorded. Target intensity information of the marked targets is obtained through a statistical averaging window, including:

[0056] S21: Based on the SAR image resolution and the size of the detected ship target, set the basic parameters of the constant false alarm rate (CFAR) detection algorithm: energy statistical average clutter reference cell number M×N, and protection cell number N. p Target false alarm probability P FA The clutter reference element number is two-dimensional data, corresponding to the range and azimuth dimensions of the two-dimensional SAR image, respectively.

[0057] S22: Select detection unit i, and estimate the average clutter power intensity P near detection unit i. scat :

[0058]

[0059] Where i represents the label of the SAR image detection unit;

[0060] And calculate the total number N of reference units corresponding to the current detection unit. ref :

[0061] Total number of reference units N ref = Number of clutter reference elements - Number of protection elements;

[0062] Here, the total number of reference units N refIt is necessary to differentiate and count the detection units based on their positions in the image, using the total number of actual effective reference units near the detection unit as the standard. For example, the effective reference units corresponding to the detection units on the left edge of the image only include the upper, lower, and right clutter reference units.

[0063] S23: Based on the target false alarm probability P set in step S21 FA Considering the current clutter environment, a specific CFAR algorithm is selected from the detection algorithm library (the main common algorithms are CA-CFAR, SO-CFAR, and GO-CFAR) to estimate the constant false alarm rate (CFAR) detection algorithm threshold factor α. T (Taking the classic CA-CFAR detection algorithm as an example, detection threshold factor) Alternatively, you can set the algorithm threshold factor α based on your processing experience. T ;

[0064] S24; Based on the average clutter power intensity P in step S21 scat Set the detection threshold T = α T ·P scat If the intensity of the current detection unit is higher than the detection threshold, the current detection unit is marked as valid (1); otherwise, it is marked as invalid (0).

[0065] S25: Slide through all detection units in the image according to steps S21 to S24 to obtain the binary image of CFAR detection results;

[0066] S26: Based on the SAR image resolution and the size of the detected ship target, set an effective target size threshold; remove target units smaller than the threshold in the binary image of the CFAR detection result through erosion operation; restore the image after erosion through dilation operation; mark the target detection bounding box in the restored image to obtain the detection result image, and record the position information of each mark in the SAR image and the maximum intensity information at the mark.

[0067] S3: Based on a precise Earth model, orbital parameters, satellite attitude, antenna size and installation angle and position, and factors affecting the antenna scanning angle, combined with SAR system carrier frequency or wavelength, resolution, and system pulse transmission frequency parameters, different SAR imaging modes and corresponding imaging processes are used to obtain the position and intensity information of real targets and blurred false targets under different imaging modes. Furthermore, the relative position information and relative intensity relationship between real targets and false targets under this imaging mode are obtained, and the relative intensity comparison threshold is derived, including:

[0068] S31: Based on the accurate Earth model, orbital parameters, satellite attitude, antenna size, antenna installation angle and position, and antenna beam scanning angle, obtain the current antenna beam illumination status of the SAR system;

[0069] S32: Taking into account the antenna beam state of the SAR system in S31 and the current imaging mode (strip mode, scanning mode, spotting mode, sliding spotting mode, etc.) of the SAR system, the antenna pattern of the SAR system is introduced to obtain the gain relationship of the antenna beam pattern corresponding to the beam scanning in the current imaging mode.

[0070] S33: In the current imaging mode, based on the relative slant range position relationship between the SAR payload and the target, obtain the azimuth time slant range history and azimuth Doppler spectrum of the corresponding target.

[0071] S34: By using the azimuth time slant range history in S33 and combining it with the pulse transmission frequency (PRF), the position information of the real target and the blurred false target in different imaging modes can be obtained. By combining the azimuth Doppler spectrum in S33 with the antenna pattern weighting in S32, the intensity information of the real target and the blurred false target in the current imaging mode can be obtained.

[0072] S35: By using the position and intensity information of the real target and the blurred false target corresponding to different imaging modes in S34, obtain the azimuthal relative position and relative intensity information of the blurred false target relative to the corresponding real target of any order.

[0073] S4: Based on the relative position information and relative intensity relationship between real and false targets in S3, the marked targets in the SAR image in S2 are compared and analyzed, and then re-marked to obtain the final real target markings in the SAR image, achieving low false alarm rate SAR image ship detection, including:

[0074] S41: Decompose the position and intensity information of the marked targets in S2, sort the targets at the same distance position, and perform azimuth position difference operation to obtain the azimuth position difference of the marked targets at the same distance.

[0075] S42: For a target being detected at the same distance gate, if the difference in its azimuth position and the relationship between the blur strength and the position of the real target and the blurred false target in the corresponding imaging mode in S3, then the marked target at the distance gate position is determined to be either a real target or a blurred false target (taking the basic frontal side view strip mode as an example, the blurred false targets appear in pairs on both sides of the real target in the azimuth direction, and the blur strength is about 18 to 30 dB weaker than the real target).

[0076] S43: Re-mark all marked targets in the detection result map in S26 to obtain the ship detection results corresponding to the real targets.

[0077] Specifically, taking the processing of ocean ship images with a C-band resolution of 2m (range) × 100m (azimuth) as an example (which has already undergone the S1 processing described above, removing land information and retaining only the ocean scene), the basic system parameters are shown in Table 1.

[0078] Table 1. Simulation Parameters for the Example

[0079]

[0080]

[0081] Figure 2 The simulated image is a sea surface image containing several ships obtained after processing the original image through S1. Only the first-order azimuth ambiguity signal is considered in the simulated image.

[0082] In S2, the constant false alarm rate algorithm is used to identify and mark targets in the image and obtain target location and intensity information;

[0083] In S3, the relative position and relative intensity information of the target and the fuzzy false target are evaluated through parameter calculation;

[0084] In S4, the target in S2 is re-marked by comparing parameter information to obtain the true target position, thus realizing low false alarm SAR ship detection images.

[0085] Further, S2 includes:

[0086] S21: Set the number of energy statistical average clutter reference units, the number of protection units, and the false alarm probability of the target according to the image resolution requirements (2m in the range direction and 100m in the azimuth direction);

[0087] S22: Perform pixel-by-pixel unit clutter power average intensity assessment on each pixel in the image;

[0088] S23: Use the CA-CFAR detection method and set the CA-CFAR detection threshold factor weights;

[0089] S24: Combine the detection threshold factor value in S23 with the average intensity of clutter power to set the detection threshold;

[0090] S25: Traverse any pixel unit in the image to obtain binary data of the detection result;

[0091] S26: Set the target detection threshold, label the binary data detected in S25, and obtain the SAR image with the bounding box label, such as... Figure 4 , Figure 5a , Figure 5b , Figure 5c As shown in the figure, the solid lines in the diagram represent the actual target markers.

[0092] The constant false alarm ship detection algorithm based on background clutter statistics in S2 can be replaced by other similar marker ship detection algorithms, provided that these detection algorithms can provide the position and intensity information of the marked detected targets (all targets, including real targets and blurred false targets) in the scene after detection.

[0093] Further, S3 includes:

[0094] S31: Based on the accurate Earth model, orbital parameters, satellite attitude, antenna size, antenna installation angle and position, and antenna beam scanning angle, obtain the current antenna beam illumination status of the SAR system;

[0095] S32: Taking into account the antenna beam state of the SAR system in S31 and the strip imaging mode adopted by the SAR system in this embodiment, the antenna pattern of the SAR system is introduced to obtain the gain relationship of the antenna beam pattern corresponding to the beam scanning in the current imaging mode.

[0096] S33: In the current imaging mode, based on the relative slant range position relationship between the SAR payload and the target, obtain the azimuth time slant range history and azimuth Doppler spectrum of the corresponding target.

[0097] S34: By using the azimuth time slant range history in S33 and combining it with the pulse transmission frequency (PRF), the position information of the real target and the blurred false target in the current imaging mode is obtained. By using the azimuth Doppler spectrum in S33 and combining it with the antenna pattern weighting in S32, the intensity information of the real target and the blurred false target in the current imaging mode is obtained.

[0098] S35: Using the position and intensity information of the real target and the blurred false target in the current imaging mode in S34, obtain the azimuth relative position and relative intensity information of the blurred false target relative to the corresponding real target in ±1 order.

[0099] Further, S4 includes:

[0100] S41: Decompose the position and intensity information of the marked targets in S2, sort the targets at the same distance position, and perform azimuth position difference operation to obtain the azimuth position difference of the marked targets at the same distance.

[0101] S42: For a target being detected at the same distance gate, if the difference in its azimuth position and the relationship between the blur strength and the position of the real target and the blurred false target in the corresponding imaging mode in S3, then the marked target at the distance gate position is determined to be either a real target or a blurred false target (taking the basic frontal side view strip mode as an example, the blurred false targets appear in pairs on both sides of the real target in the azimuth direction, and the blur strength is about 18 to 30 dB weaker than the real target).

[0102] S43: Re-mark all marked targets in the detection result image of S2 to obtain the ship detection results corresponding to the real targets, such as... Figure 6 , Figure 7a , Figure 7b , Figure 7c As shown in the figure, solid lines represent real target markers, and dashed lines represent blurred false target markers.

[0103] Figure 3a , Figure 3b , Figure 3c , Figure 5a , Figure 5b , Figure 5c , Figure 7a , Figure 7b , Figure 7c These correspond to the original SAR image containing first-order azimuth ambiguity, the SAR image after preliminary target labeling, and the SAR image relabeled based on ambiguity analysis. It can be seen that the method proposed in this invention can distinguish false targets generated by ambiguity in the image, thereby significantly reducing the false alarm rate of ship detection in SAR images.

Claims

1. A low resolution SAR image ship detection method based on fuzzy analysis, characterized in that, It comprises the following steps: S1: Increase the intensity distinction of land areas and ocean areas in the SAR image by image gray scale enhancement / decay operation, and identify and separate the land areas and ocean areas in the SAR image; S2: Process the SAR image after separating the land areas in S1 by constant false alarm CFAR detection algorithm, mark the detected ship targets containing real targets and false targets, record the position information of the marked targets in the SAR image, and obtain the target intensity information of the marked targets by statistical average window; S3: According to the accurate earth model, orbit parameters, satellite attitude, antenna size and installation angle and position, antenna scanning angle, combined with the SAR system carrier frequency or wavelength, resolution, system pulse transmission frequency index parameters, through different SAR imaging modes and corresponding imaging processes, the position and intensity information of the real targets and blurred false targets under different imaging modes are obtained, the relative position information and relative intensity relationship of the real targets and false targets under this kind of imaging mode are further obtained, and the relative intensity contrast threshold is obtained; S4: Based on the relative position information and relative intensity relationship of the real targets and false targets in S3, the marked targets in the SAR image in S2 are compared and analyzed and re-marked to obtain the final real target mark of the SAR image, and the low false alarm rate SAR image ship detection is realized.

2. The low resolution SAR image ship detection method based on fuzzy analysis according to claim 1, characterized in that, The S1 comprises: S11: Use image gray scale enhancement / decay operation, set image gray scale threshold, calculate image pixel weighting coefficient, and perform overall weighting on the SAR image to make the high-intensity gray scale region corresponding to the land area and the low-intensity gray scale region corresponding to the ocean area in the SAR image respectively obtain corresponding enhancement and decay weighting, and increase the intensity distinction of the land area and the ocean area in the SAR image; S12: Perform smooth window gray scale histogram calculation on the SAR image obtained in S11 to obtain the gray scale statistical information of the land area and the ocean area in the SAR image; S13: According to the characteristic that the gray scale intensity of the land area is higher than that of the ocean area, evaluate and calculate the land-ocean gray scale segmentation threshold, and perform threshold segmentation on the SAR image obtained in S11 based on the land-ocean gray scale segmentation threshold to obtain the SAR image of the ocean area from which the land area has been detected and separated.

3. The low resolution SAR image ship detection method based on fuzzy analysis according to claim 2, characterized in that, The S2 comprises: S21: According to the SAR image resolution and the size of the detected ship target, set the basic parameters of the constant false alarm CFAR detection algorithm: energy statistical average clutter reference cell number MxN, protection unit number N p , target false alarm probability P FA , wherein the clutter reference cell number is two-dimensional data, corresponding to the range and azimuth dimensions of the two-dimensional SAR image respectively; S22: select the detection unit i, estimate the average intensity P of the clutter power near the detection unit i scat : And the total number of reference units corresponding to the current detection unit N is calculated ref : Reference cell total number N ref = clutter reference cell number - guard cell number S23: According to the target false alarm probability P set in step S21 FA , in the constant false alarm CFAR detection algorithm library, select any ship detection algorithm according to the current clutter environment, estimate the constant false alarm detection algorithm threshold factor a T , or set the constant false alarm detection algorithm threshold factor a T according to the processing experience. S24; according to the step S21 clutter power average intensity P scat , set detection threshold T = a T ·P scat ; if the current detection unit intensity is higher than the detection threshold, mark the current detection unit as valid 1, otherwise mark it as invalid 0; S25: Slide through all detection units in the SAR image according to steps S21 to S24 to obtain a CFAR detection result binary image; S26: According to the SAR image resolution and the size of the detected ship target, set an effective detection target specification threshold; remove the target units smaller than the effective detection target specification threshold in the CFAR detection result binary image by erosion operation; perform image restoration on the image after erosion by dilation operation; mark the detection result image and record the position information of each mark in the SAR image and the maximum intensity information at the mark.

4. The low resolution SAR image ship detection method based on fuzzy analysis according to claim 3, characterized in that, The S3 comprises: S31: obtaining a current SAR system antenna beam irradiation state according to a precise earth model, orbit parameters, satellite attitude, antenna size, antenna installation angle and position, and antenna beam scanning angle; S32: comprehensively considering the SAR system antenna beam irradiation state in S31 and a current imaging mode adopted by the SAR system, introducing an antenna directional diagram of the SAR system, and obtaining a gain relationship of a beam scanning corresponding antenna directional diagram in the current imaging mode; S33: in the current imaging mode, obtaining a position-time-range history and an azimuth Doppler spectrum corresponding to a target according to a relative slant range position relationship between the SAR load and the target; S34: obtaining position information of real targets and blurred signal corresponding false targets in different imaging modes through the position-time-range history in S33 combined with a pulse transmission frequency, and obtaining intensity information of the real targets and the blurred signal corresponding false targets in the current imaging mode through the azimuth Doppler spectrum in S33 combined with the antenna directional diagram weighting in S32, the real targets and the blurred signal corresponding false targets in the current imaging mode being respectively denoted as real targets and blurred false targets; S35: obtaining azimuth relative position information and relative intensity information of an arbitrary order of blurred false targets relative to corresponding real targets through the position information and the intensity information of the real targets and the blurred false targets in different imaging modes in S34.

5. The low resolution SAR image ship detection method based on fuzzy analysis according to claim 4, characterized in that, The S4 includes: S41: decomposing the position information and the intensity information of the marked targets in S2, sorting the targets at the same range position, and performing an azimuth position difference operation to obtain an azimuth position difference of the marked targets located at the same range gate; S42: for the marked targets detected at the same range gate, if the azimuth position difference and the blur intensity-weakness relationship thereof are consistent with the position of the real targets and the blurred false targets in the corresponding imaging mode in S3, it is determined that the marked targets at the distance gate position correspond to the real targets or the blurred false targets; S43: re-marking all the marked targets in the detection result map in S26 to obtain a ship detection result corresponding to the real targets.

Images