A method and system for extracting oil film from marine radar based on improved snow goose algorithm

Abstract

The present invention proposes a method and system for extracting oil film from a marine radar based on an improved Snow Goose algorithm, and relates to the field of image processing technology. The technical key points of the present invention include: first, preprocessing the acquired radar oil film image; then, performing feature extraction on the preprocessed image and generating an oil film region of interest, wherein the SIFT feature filtering algorithm based on intensity threshold is combined to effectively remove the interference of high-brightness clutter, and an oil film region of interest that better fits the actual shape of the oil film is generated through region expansion and boundary smoothing; then, segmenting and extracting the preprocessed image using the improved Snow Goose algorithm, which effectively avoids local optimality problems and achieves faster convergence speed, thereby achieving a more stable segmentation effect based on the optimal threshold obtained by solution; finally, fusing the oil film region of interest and the segmented and extracted image to obtain an oil film image. The present invention improves the accuracy of oil film extraction.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a method and system for extracting oil film from a marine radar based on an improved snow goose algorithm. Background Art

[0002] In the field of offshore oil pollution monitoring, traditional oil slick extraction methods are often affected by factors such as image quality, noise interference, and complex backgrounds, resulting in inaccurate or inefficient extraction results. Existing image processing algorithms struggle to effectively distinguish oil slicks from the seawater background when processing marine radar images. This is particularly true when the oil slick edges are blurred or the slick thickness is uneven, significantly reducing the extraction effect. Therefore, developing a method that can adapt to complex marine environments and efficiently and accurately extract the oil slick area is of great practical significance. Summary of the Invention

[0003] In view of the above problems, the present invention proposes a method and system for extracting oil film from marine radar based on an improved snow goose algorithm.

[0004] According to one aspect of the present invention, a method for extracting oil film from a marine radar based on an improved snow goose algorithm is proposed. The method comprises:

[0005] Preprocessing the acquired radar oil film image;

[0006] Perform feature extraction on the preprocessed image and generate the oil film interest region;

[0007] The improved snow goose algorithm is used to segment and extract the preprocessed image;

[0008] The oil film region of interest and the segmented and extracted image are fused to obtain an oil film image.

[0009] Furthermore, the preprocessing includes: coordinate transformation, grayscale transformation, contrast enhancement, and noise reduction processing; wherein the noise reduction processing adopts adaptive Gaussian filtering, and its two-dimensional Gaussian kernel is:

[0010] ;

[0011] in, is the coordinate offset relative to the center point within the Gaussian kernel; is the standard deviation.

[0012] Furthermore, the feature extraction of the pre-processed image and the generation of the oil film region of interest include:

[0013] Use SIFT feature extraction algorithm for feature extraction;

[0014] The extracted features are screened using a feature filtering algorithm based on intensity threshold to filter out high-brightness clutter in the image;

[0015] The oil film interest region is generated by performing convex hull calculation, peripheral point selection, region expansion and boundary smoothing on the filtered feature region.

[0016] Furthermore, the calculation formula of the feature filtering algorithm based on the intensity threshold is:

[0017] ;

[0018] in, Indicates the location of feature points in the image The pixel gray value; Indicates the intensity threshold.

[0019] Furthermore, the region is expanded by moving each boundary point on the convex hull outward by a fixed distance along its normal direction to generate a new extended boundary point;

[0020] The boundary smoothing is to optimize the boundary of the oil film interest area by using smoothing filtering. The formula of smoothing filtering is:

[0021] ;

[0022] in, represents the coordinates of the boundary points after smoothing filtering, is the coordinate of the current boundary point; Current point The coordinates of the previous and next adjacent boundary points; is the smoothing factor.

[0023] Furthermore, the segmentation and extraction of the pre-processed image using the improved snow goose algorithm includes:

[0024] Set two segmentation thresholds for the preprocessed image and ,in Indicates the boundary between the oil film area and the transition area; Indicates the boundary between the transition area and the background area; the two segmentation thresholds are and Perform optimization solutions, including:

[0025] The fitness function is the improved three-region inter-class variance, where the three regions are background region, transition region, and oil film region. The fitness value is calculated as follows:

[0026] ;

[0027] Where, Indicates a dynamic reward mechanism, with a value range of [0, 0.25]; represents the inter-class variance of the three regions; 、 Represents the penalty terms and penalty items The coefficient of

[0028] The formula for position update is:

[0029] ;

[0030] Where, is the new candidate solution position; is the current individual position; is the adaptive weight coefficient, used to control the moving step size; is a standard normal distribution random number; is the current iteration number; are the upper and lower bounds of the search space respectively;

[0031] The preprocessed image is segmented and extracted based on the two optimal segmentation thresholds obtained.

[0032] Furthermore, the adaptive weight coefficient for:

[0033] ;

[0034] Where, Indicates the maximum number of iterations; 、 is a constant, and =1.

[0035] Furthermore, the inter-class variance of the three regions for:

[0036] ;

[0037] Where, 、 、 They are the pixel proportions corresponding to the three regions; are the grayscale means corresponding to the three regions respectively; is the global average grayscale.

[0038] Furthermore, the improved snow goose algorithm is used to divide the two segmentation thresholds and Optimization solutions also include:

[0039] A mixed strategy is used when the population is initialized: a part of the population individuals are Gaussian distributed around the preset central area, and the other part of the population individuals are uniformly randomly distributed in the search space; so that the two segmentation thresholds of the candidate solution are and Always keep < The ordered relationship;

[0040] When calculating the fitness value, if the pixel ratios of the three regions are 、 、 If any pixel ratio is less than the preset minimum threshold, then the inter-class variance of the three regions is =0.

[0041] Furthermore, the fusing process of the oil film region of interest and the segmented and extracted image includes:

[0042] Assigning the oil film interest region a value of white to obtain a processed oil film interest region;

[0043] The segmented and extracted image is processed with double threshold three-classification and the noise area is removed to obtain the processed segmented image;

[0044] Multiply the processed oil film interest region with the processed segmented image to obtain the oil film image.

[0045] According to another aspect of the present invention, a marine radar oil film extraction system based on an improved snow goose algorithm is proposed, the system comprising:

[0046] a preprocessing module configured to preprocess the acquired radar oil film image;

[0047] an interest region extraction module configured to perform feature extraction on the preprocessed image and generate an oil film interest region;

[0048] a segmentation module configured to perform segmentation extraction on the preprocessed image using an improved snow goose algorithm;

[0049] The oil film extraction module is configured to fuse the segmented and extracted image with the oil film region of interest to obtain an oil film image.

[0050] The beneficial technical effects of the present invention are:

[0051] The present invention proposes a method and system for extracting oil slicks from marine radars based on an improved Snow Goose algorithm. First, the acquired radar oil slick image is preprocessed. Through refined image preprocessing, the contrast between the oil slick and the seawater background is significantly enhanced, making the oil slick features more prominent in the image, thereby improving the accuracy and reliability of subsequent feature extraction. Feature extraction is then performed on the preprocessed image, and an oil slick region of interest (ROI) is generated. This method, combined with an intensity threshold-based SIFT feature filtering algorithm, effectively reduces computational complexity and removes interference from high-brightness clutter, further enhancing the robustness of feature extraction. Adjustable peripheral region extraction and boundary smoothing are used to generate a ROI that better fits the actual shape of the oil slick, significantly improving the accuracy of oil slick extraction. The preprocessed image is then segmented and extracted using the improved Snow Goose algorithm, making the image segmentation process more efficient and accurate, effectively avoiding local optimality, and achieving faster convergence and more stable segmentation results. Finally, the oil slick region of interest and the segmented and extracted image are fused to obtain an oil slick image. The present invention improves the accuracy of oil slick extraction. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] The above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily apparent by reading the following detailed description with reference to the accompanying drawings, in which several embodiments of the present invention are shown by way of example and not limitation, in which:

[0053] Figure 1 This is a flow chart of a method for extracting oil film from a marine radar based on an improved snow goose algorithm according to an embodiment of the present invention;

[0054] Figure 2 is an example of an original radar oil film image in an embodiment of the present invention;

[0055] Figure 3 1 is a schematic diagram of the preprocessing process in an embodiment of the present invention;

[0056] Figure 4 is an example of a radar image after coordinate system conversion in an embodiment of the present invention;

[0057] Figure 5 is an example of an image after preprocessing in an embodiment of the present invention;

[0058] Figure 6 is an example diagram of feature points after screening in an embodiment of the present invention;

[0059] Figure 7 This is an example of a ROI mask image generated by SIFT feature extraction in an embodiment of the present invention;

[0060] Figure 81 is a flow chart of image segmentation and extraction using the improved snow goose algorithm in an embodiment of the present invention;

[0061] Figure 9 1 is an example diagram of the result of image segmentation and extraction using the improved snow goose algorithm in an embodiment of the present invention;

[0062] Figure 10 is an example of an image after region of interest processing in an embodiment of the present invention;

[0063] Figure 11 This is an example of the final oil film image in the embodiment of the present invention;

[0064] Figure 12 Schematic diagram of the oil film position on the radar image in the plane coordinate system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0065] The principles and spirit of the present invention will be described below with reference to several exemplary embodiments. It should be understood that these embodiments are provided solely to enable those skilled in the art to better understand and implement the present invention, and are not intended to limit the scope of the present invention in any way. Rather, these embodiments are provided to make this disclosure more thorough and complete, and to fully convey the scope of the present disclosure to those skilled in the art.

[0066] Those skilled in the art will appreciate that embodiments of the present invention may be implemented as a system, apparatus, device, method, or computer program product. Therefore, the present disclosure may be implemented in the following forms: entirely in hardware, entirely in software (including firmware, resident software, microcode, etc.), or in a combination of hardware and software. It should be understood that any number of elements in the figures is for illustrative purposes only and not limiting, and any nomenclature is for distinction only and does not have any limiting meaning.

[0067] The present invention proposes a method and system for extracting oil slicks from marine radars based on an improved Snow Goose Algorithm. The method enhances the contrast between the oil slick and the background while reducing noise interference by improving image preprocessing steps. SIFT feature filtering based on intensity thresholding is introduced to reduce the amount of computation and improve the accuracy of feature extraction. Adjustable peripheral region extraction and boundary smoothing techniques are used to optimize the generation of regions of interest (ROIs), making the ROIs more closely aligned with the oil slick area. Finally, an improved Snow Goose Algorithm (ISGA) is used to segment and extract images, enabling efficient and accurate identification of oil slick areas.

[0068] The embodiment of the present invention proposes a method for extracting oil film from marine radar based on an improved snow goose algorithm. Figure 1 As shown, the method includes:

[0069] S1. Preprocessing the acquired radar oil film image;

[0070] S2, extracting features from the preprocessed image and generating an oil film region of interest;

[0071] S3, using the improved snow goose algorithm to segment and extract the preprocessed image;

[0072] S4. Fusion processing is performed on the oil film region of interest and the segmented and extracted image to obtain an oil film image.

[0073] The method begins with S1, preprocessing the acquired radar oil film image.

[0074] According to an embodiment of the present invention, the radar oil spill image obtained is as follows: Figure 2 The preprocessing process is as follows. Figure 3 As shown, it includes coordinate transformation, grayscale transformation, contrast enhancement, and noise reduction processing.

[0075] Specifically, the original marine radar image is firstly subjected to coordinate transformation (the polar coordinate system of the image is transformed into the Cartesian coordinate system). The image after coordinate transformation is as follows: Figure 4 To facilitate SIFT feature extraction, in addition to grayscale conversion of the original image, further contrast enhancement is performed: the grayscale range of the specific area [0.4, 1] is stretched to [0, 1], which can highlight the oil film features and suppress background interference. It can be expressed as follows:

[0076]

[0077] in, is the grayscale value of the output image after transformation at (x, y); is the original grayscale value of the input image at the coordinate point (x, y); Indicates the lower limit of the grayscale range that needs to be stretched; Indicates the upper limit of the grayscale range that needs to be stretched; 255 is the maximum grayscale value of an 8-bit image. Then, the image is processed for co-channel interference noise reduction and speckle noise suppression. To suppress speckle noise, a two-dimensional Gaussian kernel convolution denoising with σ=1.2 is performed; the two-dimensional Gaussian kernel is:

[0078]

[0079] in, is the coordinate offset relative to the center point within the Gaussian kernel; is the standard deviation. The preprocessed image is as follows Figure 5 shown.

[0080] The existing preprocessing uses fixed parameter histogram equalization. The present invention adds dynamic range adjustment and adaptive filtering in the preprocessing process, which can enhance the contrast between the oil film and the background and retain weak features.

[0081] Then, S2 is executed to extract features from the preprocessed image and generate an oil film region of interest; this includes: extracting features using the SIFT feature extraction algorithm; filtering the extracted features using a feature filtering algorithm based on an intensity threshold to filter out high-brightness clutter in the image; and generating an oil film region of interest by performing convex hull calculation, peripheral point selection, region expansion, and boundary smoothing on the filtered feature region.

[0082] According to an embodiment of the present invention, based on traditional SIFT feature extraction, feature filtering based on intensity thresholds is added to reduce initial computational complexity and errors. This is because oil films are typically dark (low grayscale values) in images. By performing segmentation extraction based on the dark threshold, high-brightness clutter can be directly filtered out. The formula is:

[0083]

[0084] in, is a Boolean mask that marks which feature points meet the intensity condition; is the position of the feature point in the image The pixel gray value; is the intensity threshold, which is used to distinguish the oil film characteristics (dark) from the background noise (bright). The feature points retained after screening are as follows Figure 6 shown.

[0085] The oil film region of interest is generated by performing convex hull calculation, peripheral point selection, region expansion, and boundary smoothing on the filtered feature area. Convex hull calculation is a prior art and will not be described here in detail. Since the oil film edge feature points are more representative, the present invention incorporates adjustable peripheral region extraction, which can extract peripheral feature points according to the input ratio, so that the final ROI area is more closely aligned with the oil film portion. The extraction formula is:

[0086]

[0087] in, is the number of peripheral feature points, is the total number of feature points, is the preset ratio of peripheral feature points, and its value range is [0, 1].

[0088] In order to prevent errors in feature point extraction, which may cause the oil film edge to be ignored by the final ROI and thus omit the peripheral area of ​​the feature points, the present invention adds a fixed pixel buffer as follows:

[0089]

[0090] in, is the buffer extension distance (pixels), is the unit normal vector at the boundary point, pointing to the outside of the convex hull; The coordinates of the new boundary points after expansion. Move outward a fixed distance along its normal direction and finally generate new extended boundary points to ensure that the final ROI does not ignore the oil film area.

[0091] In order to make the ROI boundary fit the oil film area better, the present invention introduces smoothing filtering to optimize the ROI boundary. The smoothing formula is:

[0092]

[0093] in, is the coordinate of the current boundary point; For the current point The coordinates of the previous and next adjacent boundary points; is the smoothing factor, The smaller it is, the more it retains the original shape (the less smoothing it is). Larger values ​​result in greater smoothing (but may result in excessive distortion).

[0094] The present invention improves the extraction accuracy of the oil film ROI through the above process, enhances its noise resistance, and ensures the morphological authenticity of the generated area. The ROI mask generated image under SIFT feature extraction is as follows Figure 7 shown.

[0095] Then, S3 is executed to segment and extract the preprocessed image using the improved snow goose algorithm.

[0096] According to an embodiment of the present invention, Figure 8 As shown in the figure, the process of segmenting and extracting the pre-processed image using the improved snow goose algorithm includes: setting two segmentation thresholds for the pre-processed image and ,in Indicates the boundary between the oil film area and the transition area; Indicates the boundary between the transition area and the background area; through two thresholds and The preprocessed image is divided into three regions, namely the oil film region: pixel value ∈ [0, ); Transition zone: pixel value∈[ , ); Background area: pixel value∈[ ,255]; using the improved snow goose algorithm to calculate the two segmentation thresholds and Perform optimization solutions, including:

[0097] S31. Initialize ISGA parameters: the snow goose population size is N geese = 80; the maximum number of iterations is =350; the lower limit lb of the threshold search space is [65, 100], the upper limit ub is [85, 120], and the adaptive weight coefficient is , is the current iteration number, 、 is a constant and =1;

[0098] The adaptive weight coefficient Additional The fixed weight of can make the algorithm still maintain a certain exploration ability in the later stage and avoid falling into the local optimum too early; as an example, =0.95, =0.05.

[0099] S32. Population initialization: Combining prior knowledge (center point initialization) and random exploration (uniform distribution) to balance convergence speed and global search capability. In this invention, population initialization adopts the following hybrid strategy: 1) Hard constraint: 70% of the individuals are Gaussian distributed around the preset center [75, 110] to make the initial solution concentrated near the possible optimal solution, improving convergence efficiency; 30% of the individuals are uniformly and randomly distributed in the search space [lb, ub] to ensure population diversity and avoid excessive concentration of initial solutions; 2) Forced sorting: Ensure that the two threshold parameters of the candidate solution are and Always keep < The ordered relationship between the two thresholds ensures the effectiveness of the dual threshold segmentation logic and avoids ≥ The invalid solutions will affect the convergence of the algorithm.

[0100] The population initialization strategy of traditional ISGA is completely random initialization. The present invention increases the concentration of 70% individuals in the optimal interval (near [75,110]) and keeps 30% random, which can accelerate the convergence speed.

[0101] S33, perform iterative calculation:

[0102] The fitness function uses the three-region inter-class variance calculation to maximize the inter-class difference between foreground and background, and adapts to complex scenes through three-region division (background area, transition area, oil film area). The formula for calculating the three-region inter-class variance is:

[0103]

[0104] in, is the inter-class variance of the three regions; (k=0,1,2) is the pixel ratio of each area (background area / transition area / oil film area); is the grayscale mean of the three regions; is the global average grayscale.

[0105] The inter-class variance of the three regions is further improved, and the improved inter-class variance formula is:

[0106]

[0107] Where, is the final fitness value, which is used to comprehensively evaluate the threshold [ , ] The segmentation quality, the larger the value, the better the segmentation effect; It is a dynamic reward mechanism with a value range of [0, 0.25]; for Normalized distance from the empirical value of 70, for The normalized distance deviates from the empirical value of 110. The purpose is to normalize the value to the range of [0,1] to prevent the difference in numerical dimensions from affecting the penalty intensity. 、 Represents the penalty terms and penalty items The coefficient of ; as an example, =0.2, =0.1.

[0108] Experiments show that The impact on segmentation is greater, so the penalty is stronger, while It is less sensitive to the segmentation results, has weaker penalties, and allows for greater flexibility.

[0109] Further, a validity check is performed: if the pixel ratio of the three regions If any pixel ratio is less than the preset minimum threshold (for example, the threshold is 0.01), then the inter-class variance of the three regions is =0 (i.e. marked as an invalid solution) to avoid a certain type of area having too few pixels, which would make the segmentation result meaningless.

[0110] The fitness function of existing ISGA is a single inter-class variance, which is calculated as a single or double threshold, without any regional validity check. The improved three-region inter-class variance algorithm in this paper can adapt to multi-target segmentation, and the fitness function can improve the rationality of segmentation, achieving more robust segmentation and faster convergence.

[0111] After calculating the fitness value, use the leader to guide the update and move the individual towards the current optimal solution. The position update formula is:

[0112]

[0113] in, is the new candidate solution position; is the current individual position; w is the adaptive weight coefficient, which is used to control the moving step size; Avoid straight-line approximation to introduce randomness; is the global optimal solution position.

[0114] Furthermore, to avoid local optimality, random exploration updates are added, and the position update formula is improved as follows:

[0115]

[0116] in, Standard normally distributed random numbers.

[0117] The existing ISGA weight decay is linear, but this invention improves it to nonlinear decay to retain the ability of later fine-tuning and avoid premature convergence. In addition, the boundary processing link of the original algorithm adopts simple truncation, while this invention adds forced sorting + hard constraints + validity checks to 100% guarantee the validity of the solution. Figure 9 The segmentation result diagram is shown.

[0118] Then, S34 is executed to fuse the oil film region of interest and the segmented and extracted image to obtain an oil film image.

[0119] According to an embodiment of the present invention, first, the oil film interest region is assigned white, and the rest is changed to black, and the processed oil film interest region is obtained, such as Figure 10 Then, the segmented image is processed with double threshold three-classification and the noise area is removed, that is, the dark area in the image is set to 0, and the middle and bright areas are set to 255 to obtain the processed segmented image; then, the processed oil film interest area is multiplied with the processed segmented image to obtain the oil film image, as shown in Figure 11 shown.

[0120] In order to highlight the image, the oil film image can be combined with the original image, and the oil film area can be highlighted in red, such as Figure 12 Furthermore, the radar data in the plane coordinate system can be converted into radar data in the polar coordinate system to obtain the oil film position in the marine radar image.

[0121] The embodiment of the present invention further provides a marine radar oil film extraction system based on an improved snow goose algorithm, the system comprising:

[0122] a preprocessing module configured to preprocess the acquired radar oil film image;

[0123] an interest region extraction module configured to perform feature extraction on the preprocessed image and generate an oil film interest region;

[0124] a segmentation module configured to perform segmentation extraction on the preprocessed image using an improved snow goose algorithm;

[0125] The oil film extraction module is configured to fuse the segmented and extracted image with the oil film region of interest to obtain an oil film image.

[0126] It should be noted that the function of the marine radar oil film extraction system based on the improved Snow Goose algorithm described in this embodiment can be described by the aforementioned marine radar oil film extraction method based on the improved Snow Goose algorithm. For the parts not described in detail in the system embodiment, please refer to the above method embodiment.

[0127] It should be noted that although several units, modules, or submodules are mentioned in the detailed description above, such division is merely exemplary and not mandatory. In fact, according to embodiments of the present invention, the features and functions of two or more modules described above may be embodied in one module. Conversely, the features and functions of one module described above may be further divided and embodied by multiple modules.

[0128] Furthermore, although the operations of the method of the present invention are described in a particular order in the accompanying drawings, this does not require or imply that these operations must be performed in this particular order, or that all illustrated operations must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps.

[0129] Although the spirit and principles of the present invention have been described with reference to several specific embodiments, it should be understood that the present invention is not limited to the specific embodiments disclosed, and the division into various aspects does not mean that the features of these aspects cannot be combined to benefit. Such division is only for the convenience of expression. The present invention is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims.

Claims

1. A method for extracting oil film from marine radar based on an improved snow goose algorithm, characterized in that: include: Preprocessing the acquired radar oil film image; Perform feature extraction on the preprocessed image and generate the oil film interest region; The improved snow goose algorithm is used to segment and extract the preprocessed image, including: Set two segmentation thresholds for the preprocessed image and ,in Indicates the boundary between the oil film area and the transition area; Indicates the boundary between the transition area and the background area; the two segmentation thresholds are and Perform optimization solutions, including: The fitness function is the improved three-region inter-class variance, where the three regions are background region, transition region, and oil film region. The fitness value is calculated as follows: ； Where, Indicates a dynamic reward mechanism, with a value range of [0, 0.25]; represents the inter-class variance of the three regions; 、 Represents the penalty terms and penalty items The coefficient of The formula for position update is: ； Where, is the new candidate solution position; is the current individual position; is the adaptive weight coefficient, used to control the moving step size; is a standard normal distribution random number; is the current iteration number; are the upper and lower bounds of the search space respectively; Segment and extract the pre-processed image based on the two optimal segmentation thresholds obtained; The oil film region of interest and the segmented and extracted image are fused to obtain an oil film image.

2. The method for extracting oil film from marine radar based on the improved Snow Goose algorithm according to claim 1, characterized in that: The preprocessing includes: coordinate conversion, grayscale conversion, contrast enhancement, and noise reduction processing; wherein the noise reduction processing adopts adaptive Gaussian filtering.

3. The method for extracting oil film from marine radar based on the improved snow goose algorithm according to claim 1, characterized in that: The feature extraction of the pre-processed image and the generation of the oil film region of interest include: Use SIFT feature extraction algorithm for feature extraction; The extracted features are screened using a feature filtering algorithm based on intensity threshold to filter out high-brightness clutter in the image; The oil film interest region is generated by performing convex hull calculation, peripheral point selection, region expansion and boundary smoothing on the filtered feature region.

4. The method for extracting oil film from marine radar based on the improved Snow Goose algorithm according to claim 3, characterized in that: The calculation formula of the feature filtering algorithm based on intensity threshold is: ； in, Indicates the location of feature points in the image The pixel gray value; Indicates the intensity threshold.

5. The method for extracting oil film from marine radar based on the improved Snow Goose algorithm according to claim 3, characterized in that: The region expansion is to generate new extended boundary points by moving each boundary point on the convex hull outward by a fixed distance along its normal direction; The boundary smoothing is to optimize the boundary of the oil film interest area by using smoothing filtering. The formula of smoothing filtering is: ； in, represents the coordinates of the boundary points after smoothing filtering, is the coordinate of the current boundary point; Current point The coordinates of the previous and next adjacent boundary points; is the smoothing factor.

6. The method for extracting oil film from marine radar based on the improved Snow Goose algorithm according to claim 1, characterized in that: The adaptive weight coefficient for: ； Where, Indicates the maximum number of iterations; 、 is a constant, and =1; The three-region inter-class variance for: ； Where, 、 、 They are the pixel proportions corresponding to the three regions; are the grayscale means corresponding to the three regions respectively; is the global average grayscale.

7. The method for extracting oil film from marine radar based on the improved Snow Goose algorithm according to claim 1, characterized in that: The improved snow goose algorithm is used to segment the two thresholds and Optimization solutions also include: A mixed strategy is used when the population is initialized: a part of the population individuals are Gaussian distributed around the preset central area, and the other part of the population individuals are uniformly randomly distributed in the search space; so that the two segmentation thresholds of the candidate solution are and Always keep < The ordered relationship; When calculating the fitness value, if the pixel ratios corresponding to the three regions are 、 、 If any pixel ratio is less than the preset minimum threshold, then the inter-class variance of the three regions is =0.

8. The method for extracting oil film from marine radar based on the improved Snow Goose algorithm according to claim 1, characterized in that: The fusing process of the oil film region of interest and the segmented and extracted image comprises: Assigning the oil film interest region a value of white to obtain a processed oil film interest region; The segmented and extracted image is processed with double threshold three-classification and the noise area is removed to obtain the processed segmented image; Multiply the processed oil film interest region with the processed segmented image to obtain the oil film image.

9. A marine radar oil film extraction system based on an improved snow goose algorithm, characterized in that: include: a preprocessing module configured to preprocess the acquired radar oil film image; an interest region extraction module configured to perform feature extraction on the preprocessed image and generate an oil film interest region; The segmentation module is configured to perform segmentation extraction on the pre-processed image using an improved snow goose algorithm, comprising: Set two segmentation thresholds for the preprocessed image and ,in Indicates the boundary between the oil film area and the transition area; Indicates the boundary between the transition area and the background area; the two segmentation thresholds are and Perform optimization solutions, including: The fitness function is the improved three-region inter-class variance, where the three regions are background region, transition region, and oil film region. The fitness value is calculated as follows: ； Where, Indicates a dynamic reward mechanism, with a value range of [0, 0.25]; represents the inter-class variance of the three regions; 、 Represents the penalty terms and penalty items The coefficient of The formula for position update is: ； Where, is the new candidate solution position; is the current individual position; is the adaptive weight coefficient, used to control the moving step size; is a standard normal distribution random number; is the current iteration number; are the upper and lower bounds of the search space respectively; Segment and extract the pre-processed image based on the two optimal segmentation thresholds obtained; The oil film extraction module is configured to fuse the segmented and extracted image with the oil film region of interest to obtain an oil film image.

Images