Noise reduction method for oil spill detection image based on X-band marine radar

Abstract

The invention discloses a noise reduction method for an oil spill detection image based on an X-band marine radar, and the method comprises the implementation steps: S1, collecting an X-band marine radar image, and carrying out the coordinate transformation of the X-band marine radar image; s2, same-frequency interference noise suppression processing is carried out on the image after coordinate conversion, a first noise suppression image is obtained, and the same-frequency interference noise suppression processing comprises an Otsu threshold segmentation method, transverse and longitudinal proportion feature recognition and neighborhood mean value replacement; and S3, salt and pepper noise suppression processing is carried out on the first noise suppression image, a noise reduction image is finally output, and the salt and pepper noise suppression processing comprises Bradley adaptive threshold segmentation, size screening identification and three-dimensional fitting repair. According to the method, a method of combining same-frequency interference noise and salt and pepper noise reduction is adopted, and particularly, the normal value of the noise area is fitted by using a three-dimensional fitting technology, so that the noise area in the image is restored, and the recognizability of the oil spill characteristic in the radar image is remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a noise reduction method for oil spill detection images based on X-band marine radar. Background Art

[0002] Marine oil spills pose a threat to global ecosystems, maritime activities, and economies, and the resulting environmental disasters require an efficient and timely response. X-band marine radar, with its advantages of strong environmental adaptability and real-time monitoring, holds great promise for identifying marine oil spill areas. However, in practical applications, co-channel interference noise and salt-and-pepper noise in marine radar images degrade image quality, hindering the identification of oil spill areas. Although denoising techniques for X-band marine radar images have a substantial research foundation, they still face challenges in marine oil spill detection applications, such as insufficient noise suppression, weak target feature retention, and poor adaptability and robustness. Therefore, exploring a denoising method for X-band marine radar oil spill detection images is crucial. Summary of the Invention

[0003] The present invention provides a denoising method for oil spill detection images based on X-band marine radar, which is used to overcome the problem that during the denoising process of X-band marine radar images, co-frequency interference noise and salt-and-pepper noise are insufficiently suppressed, thereby destroying weak features of oil spills.

[0004] To achieve the above object, the present invention proposes a method for denoising an oil spill detection image based on an X-band marine radar, the method comprising:

[0005] S1, collecting X-band marine radar images, converting the Cartesian coordinate system of the marine radar images into a polar coordinate system with the azimuth angle as the horizontal axis and the ship distance as the vertical axis, to obtain a coordinate conversion image;

[0006] S2, performing co-frequency interference noise suppression processing on the coordinate conversion image, including:

[0007] S21, performing binary segmentation on the coordinate transformation image using the Otsu threshold segmentation method, and extracting abnormal areas with grayscale values ​​higher than the background area;

[0008] S22, traversing the connected domains of the abnormal area, calculating the horizontal and vertical ratios of each connected domain, and marking the connected domains with a horizontal and vertical ratio less than a preset threshold as co-frequency interference noise areas;

[0009] S23. Calculate the total area ratio of all co-channel interference noise regions. If the total ratio exceeds a preset threshold, perform neighborhood mean replacement on each co-channel interference noise region using mean filtering, and execute S24. If the total ratio does not exceed the preset threshold, directly output the first noise suppression image.

[0010] S24, using the image after the neighborhood mean replacement process as a new coordinate transformation image, and performing the processes of S21 to S23 again;

[0011] S3, performing salt and pepper noise suppression processing on the first noise suppressed image, including:

[0012] S31, processing the first noise-suppressed image using Bradley adaptive threshold segmentation, dividing the grayscale image of the first noise-suppressed image into a plurality of sub-windows, calculating a dynamic threshold based on the pixel mean of the sub-windows and a preset sensitivity, and generating a binary mask;

[0013] S32, traversing the connected domains of the binary mask, calculating the area of ​​each connected domain, marking the connected domains whose area is smaller than the area threshold as salt and pepper noise areas, and obtaining a salt and pepper noise marked image;

[0014] S33. Normalize the coordinates of the salt and pepper noise marked image to a preset range to obtain a normalized image; perform three-dimensional fitting processing on the normalized image, including vertical polynomial fitting, horizontal trigonometric function fitting, and environmental Gaussian distribution fitting, and perform weighted fusion calculation on the three-dimensional fitting processing results to generate a restoration value, and finally output a denoised image.

[0015] Furthermore, the expression of the longitudinal polynomial fitting is:

[0016]

[0017] Where i and j are the normalized horizontal and vertical coordinates respectively; V(i,j) represents the fitted grayscale value at the vertical position (i,j); m represents the order of the polynomial; q n represents the nth longitudinal fitting coefficient.

[0018] Furthermore, the expression for the transverse trigonometric function fitting is:

[0019] T(i,j)=p1+p2×sin(2πj)+p3×cos(2πj) (2)

[0020] Where T(i, j) represents the fitting grayscale value at position (i, j) in the longitudinal direction; p1, p2, and p3 represent the transverse fitting coefficients, respectively.

[0021] Furthermore, the expression for fitting the environmental Gaussian distribution is:

[0022]

[0023] Where E(i, j) represents the grayscale value at image (i, j); μ represents the mean of the surrounding area; and σ represents the standard deviation of the fluctuation.

[0024] Furthermore, the expression for generating the repair value by weighted fusion calculation is:

[0025] P(i,j)=W v ×V(i,j)+W t ×T(i,j)+Env(i,j) (4)

[0026] Where W v and W t They represent the weights corresponding to the longitudinal fitting value and the transverse fitting value respectively; Env(i,j) is a function value belonging to f(E(i,j))=, representing the environment fitting value at image (i,j).

[0027] The present invention has the following beneficial effects:

[0028] 1. The present invention uses X-band navigation radar equipment, which has a wider application base compared with satellite images, laser fluorescence and other methods.

[0029] 2. This invention utilizes a two-stage cascade process: first suppressing co-channel interference noise, then repairing salt-and-pepper noise. This eliminates large-scale, structured co-channel interference, preventing it from interfering with salt-and-pepper noise point location and size screening. It then refines the salt-and-pepper noise. This achieves efficient and comprehensive noise suppression in X-band navigation radar oil spill detection images, maximally preserving critical oil spill target feature information. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0031] Figure 1 Schematic diagram of the processing flow of the present invention. DETAILED DESCRIPTION

[0032] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0033] This embodiment provides a method for reducing noise of oil spill detection images based on X-band marine radar. Figure 1As shown, the method includes:

[0034] S1, collecting X-band marine radar images, converting the Cartesian coordinate system of the marine radar images into a polar coordinate system with the azimuth angle as the horizontal axis and the ship distance as the vertical axis, to obtain a coordinate conversion image;

[0035] S2, performing co-frequency interference noise suppression processing on the coordinate conversion image, including:

[0036] S21, perform Otsu threshold segmentation on the coordinate transformation image, and calculate the grayscale histogram of the coordinate transformation image at different grayscale resolutions (such as 2 10 ,2 11 ,2 12 ,2 13 etc.) to perform hierarchical traversal of thresholds, calculate the maximum inter-class variance under each level, and extract abnormal areas with grayscale values ​​higher than the background area;

[0037] S22, traversing the connected domains of the abnormal area, calculating the horizontal and vertical ratios of each connected domain, and marking the connected domains with a horizontal and vertical ratio less than a preset threshold value of 1:N as co-frequency interference noise areas; wherein the value range of N is 2 to 512;

[0038] S23. Calculate the total area ratio of all co-channel interference noise regions. If the total ratio exceeds a preset threshold value P%, perform neighborhood mean replacement on each co-channel interference noise region using mean filtering, and execute S24. If the total ratio does not exceed the preset threshold, directly output the first noise suppression image. The value range of P is 0 to 20.

[0039] S24, using the image after the neighborhood mean replacement process as a new coordinate transformation image, and performing the processes of S21 to S23 again;

[0040] S3, performing salt and pepper noise suppression processing on the first noise suppressed image, including:

[0041] S31, processing the first noise-suppressed image using Bradley adaptive threshold segmentation, dividing the grayscale image of the first noise-suppressed image into J×J windows, where J ranges from 10 to 20, calculating the pixel mean of each sub-window, setting the product of the pixel mean and a sensitivity K as a dynamic threshold, where K ranges from 0.6 to 0.8, and generating a binary mask;

[0042] S32, traversing the connected domains of the binary mask, calculating the area of ​​each connected domain, marking the connected domains whose area is less than an area threshold L as salt and pepper noise areas, where L ranges from 400 to 1000, and obtaining a salt and pepper noise labeled image;

[0043] S33. Normalize the coordinates of the salt and pepper noise marked image to the range of [0, 1] to obtain a normalized image; perform three-dimensional fitting processing on the normalized image, including vertical polynomial fitting, horizontal trigonometric function fitting, and environmental Gaussian distribution fitting, and perform weighted fusion calculation on the three-dimensional fitting processing results to generate a restoration value, and finally output a denoised image.

[0044] Specifically, the expression of the longitudinal polynomial fitting is:

[0045]

[0046] Where i and j are the normalized horizontal and vertical coordinates respectively; V(i,j) represents the fitted grayscale value at the vertical position (i,j); m represents the order of the polynomial; q n represents the nth longitudinal fitting coefficient.

[0047] Specifically, the expression for the transverse trigonometric function fitting is:

[0048] T(i,j)=p1+p2×sin(2πj)+p3×cos(2πj) (2)

[0049] Where T(i, j) represents the fitting grayscale value at position (i, j) in the longitudinal direction; p1, p2, and p3 represent the transverse fitting coefficients, respectively.

[0050] Specifically, the expression for the environmental Gaussian distribution fitting is:

[0051]

[0052] Where E(i, j) represents the grayscale value at image (i, j); μ represents the mean of the surrounding area; and σ represents the standard deviation of the fluctuation.

[0053] Specifically, the expression for generating the repair value by weighted fusion calculation is:

[0054] P(i,j)=W v ×V(i,j)+W t ×T(i,j)+Env(i,j) (4)

[0055] Where W v and W t They represent the weights corresponding to the longitudinal fitting value and the transverse fitting value respectively; Env(i,j) is a function value belonging to f(E(i,j))=, representing the environment fitting value at image (i,j).

[0056] The present invention has the following beneficial effects:

[0057] 1. The present invention uses X-band navigation radar equipment, which has a wider application base compared with satellite images, laser fluorescence and other methods.

[0058] 2. This invention utilizes a two-stage cascade process: first suppressing co-channel interference noise, then repairing salt-and-pepper noise. This eliminates large-scale, structured co-channel interference, preventing it from interfering with salt-and-pepper noise point location and size screening. It then refines the salt-and-pepper noise. This achieves efficient and comprehensive noise suppression in X-band navigation radar oil spill detection images, maximally preserving critical oil spill target feature information.

[0059] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for denoising oil spill detection images based on X-band marine radar, characterized in that: include: S1, collecting X-band marine radar images, converting the Cartesian coordinate system of the marine radar images into a polar coordinate system with the azimuth angle as the horizontal axis and the ship distance as the vertical axis, to obtain a coordinate conversion image; S2, performing co-frequency interference noise suppression processing on the coordinate conversion image, including: S21, performing binary segmentation on the coordinate transformation image using the Otsu threshold segmentation method, and extracting abnormal areas with grayscale values ​​higher than the background area; S22, traversing the connected domains of the abnormal area, calculating the horizontal and vertical ratios of each connected domain, and marking the connected domains with a horizontal and vertical ratio less than a preset threshold as co-frequency interference noise areas; S23. Calculate the total area ratio of all co-channel interference noise regions. If the total ratio exceeds a preset threshold, perform neighborhood mean replacement on each co-channel interference noise region using mean filtering, and execute S24. If the total ratio does not exceed the preset threshold, directly output the first noise suppression image. S24, using the image after the neighborhood mean replacement process as a new coordinate transformation image, and performing the processes of S21 to S23 again; S3, performing salt and pepper noise suppression processing on the first noise suppressed image, including: S31, processing the first noise-suppressed image using Bradley adaptive threshold segmentation, dividing the grayscale image of the first noise-suppressed image into a plurality of sub-windows, calculating a dynamic threshold based on the pixel mean of the sub-windows and a preset sensitivity, and generating a binary mask; S32, traversing the connected domains of the binary mask, calculating the area of ​​each connected domain, marking the connected domains whose area is smaller than the area threshold as salt and pepper noise areas, and obtaining a salt and pepper noise marked image; S33. Normalize the coordinates of the salt and pepper noise marked image to a preset range to obtain a normalized image; perform three-dimensional fitting processing on the normalized image, including vertical polynomial fitting, horizontal trigonometric function fitting, and environmental Gaussian distribution fitting, and perform weighted fusion calculation on the three-dimensional fitting processing results to generate a restoration value, and finally output a denoised image.

2. The method for denoising an oil spill detection image based on an X-band marine radar according to claim 1, characterized in that: The expression of the longitudinal polynomial fitting is: Where i and j are the normalized horizontal and vertical coordinates respectively; V(i,j) represents the fitted grayscale value at the vertical position (i,j); m represents the order of the polynomial; q n represents the nth longitudinal fitting coefficient.

3. The method for denoising an oil spill detection image based on an X-band marine radar according to claim 1, characterized in that: The expression of the transverse trigonometric function fitting is: T(i,j)=p1+p2×sin(2πj)+p3×cos(2πj) (2) Where T(i, j) represents the fitting grayscale value at position (i, j) in the longitudinal direction; p1, p2, and p3 represent the horizontal fitting coefficients respectively.

4. The method for denoising an oil spill detection image based on an X-band marine radar according to claim 1, characterized in that: The expression for the environmental Gaussian distribution fitting is: Where E(i, j) represents the grayscale value at image (i, j); μ represents the mean of the surrounding area; σ represents the standard deviation of the fluctuation.

5. The method for denoising an oil spill detection image based on an X-band marine radar according to claim 1, characterized in that: The expression for generating the repair value by weighted fusion calculation is: P(i,j)=W v ×V(i,j)+W t ×T(i,j)+Env(i,j) (4) Where W v and W t Represent the weights corresponding to the longitudinal fitting value and the horizontal fitting value respectively; Env(i,j) is a function value belonging to f(E(i,j))=, representing the environment fitting value at image (i,j).

Images