A Ship Target Segmentation Method and System Based on an Improved Harris Eagle Optimization Algorithm

Abstract

This invention proposes a ship target segmentation method and system based on an improved Harris Eagle optimization algorithm, belonging to the field of image processing and pattern recognition technology. The key technical points of this invention include: processing an airborne grayscale image containing the sky, sea area, ships, and coastline to obtain a region of interest (ROI) image; using the improved Harris Eagle optimization algorithm to obtain an optimal segmentation threshold, and using this optimal threshold to segment the airborne grayscale image to obtain a segmented image based on the optimal threshold; and fusing the ROI image and the segmented image based on the optimal threshold to obtain the final segmented image. This invention accurately locates high-response regions of ship targets using SIFT features, and combines the improved Harris Eagle optimization algorithm to achieve adaptive threshold optimization within the target grayscale band, significantly improving the accuracy and robustness of ship target segmentation under complex sea conditions, providing highly reliable technical support for maritime monitoring and other scenarios.

Description

Technical Field

[0001] This invention relates to the field of image processing and pattern recognition technology, specifically to a ship target segmentation method and system based on an improved Harris Eagle optimization algorithm. Background Technology

[0002] With the increasing application of infrared imaging technology in marine monitoring, airborne infrared imagery, with its all-weather sensing capabilities, has become a key means of ship target identification. However, under complex sea and sky conditions, it faces core challenges such as weak differences in target and background radiation, strong sea clutter interference, and significant fog attenuation, leading to bottlenecks in the adaptability and accuracy of traditional segmentation methods. In recent years, Scale-Invariant Feature Transform (SIFT) has demonstrated unique advantages in low-contrast scenes due to its stable representation of ship structural features; at the same time, the innovative application of metaheuristic optimization algorithms has provided a new solution for adaptive thresholding techniques. It is worth noting that traditional optimization algorithms are prone to getting trapped in local optima in dynamic sea surface interference environments, and it is urgent to combine them with feature space constraint mechanisms to improve robustness. Summary of the Invention

[0003] In view of the above problems, this invention proposes a ship target segmentation method and system based on an improved Harris Eagle optimization algorithm.

[0004] According to one aspect of the present invention, a ship target segmentation method based on an improved Harris Eagle optimization algorithm is proposed, the method comprising:

[0005] Process airborne grayscale images containing the sky, sea, ships, and coastline to obtain region of interest images;

[0006] An optimal segmentation threshold is obtained by using an improved Harris Hawk optimization algorithm, and the airborne grayscale image is segmented using the optimal segmentation threshold to obtain a segmented image based on the optimal segmentation threshold.

[0007] The region of interest image and the segmented image based on the optimal segmentation threshold are fused to obtain the final segmented image.

[0008] Furthermore, the process of processing the airborne grayscale image, which includes the sky, sea, ships, and coastline, to obtain the region of interest image includes:

[0009] The SIFT algorithm is used to extract multiple key points from the airborne grayscale image, including: constructing a Gaussian difference pyramid and detecting local extrema; filtering multiple local extrema based on contrast and edge response to obtain multiple key points;

[0010] Ellipse fitting is performed on multiple key points to obtain the smallest bounding ellipse containing all key points.

[0011] The image region corresponding to the smallest bounding ellipse is assigned white, and the remaining image regions are assigned black, thus generating a preliminary region of interest mask feature map.

[0012] Edge detection and land-sea segmentation line extraction are performed on the airborne grayscale image; the lower part of the land-sea segmentation line is assigned white and the upper part is assigned black to generate a land-sea segmentation mask;

[0013] The region of interest image is obtained by multiplying the initial region of interest mask feature map and the land-sea segmentation mask map pixel by pixel.

[0014] Furthermore, obtaining the optimal segmentation threshold using the improved Harris Eagle optimization algorithm includes:

[0015] During the exploration phase, the formula for updating the position of individuals in the population is:

[0016] ;

[0017] In the formula, This represents the position of the i-th individual in the (t+1)-th iteration; This represents the position of a randomly selected individual in the population; r and q are uniformly random numbers between [0,1]. is a uniformly distributed random number in [0,1]; lb and ub are the lower and upper bounds of the search space, respectively;

[0018] During the transition from exploration to development, the energy loss model for eagles hunting prey, and the energy consumed by the prey during its escape, is as follows:

[0019] ;

[0020] In the formula, Indicates a nonlinear decay exponent; This represents the initial escape energy of the prey; Indicates the exponent of the decay curve; Indicates the disturbance intensity coefficient; The noise is a standard normal distribution; t represents the current iteration number; T represents the maximum number of iterations.

[0021] During the development phase, the formula for updating the position of individuals in the population is:

[0022] ;

[0023] In the formula, This represents the optimal prey position in the t-th iteration; This represents the energy required for prey to escape. Indicates the jump range; This indicates an improved version of the Levi flight item. , This is the step size coefficient. For a standard normally distributed random variable, It's the Lévy Index. λ is the scaling factor, and λ is the iterative decay exponent.

[0024] Furthermore, the method of obtaining the optimal segmentation threshold using the improved Harris Eagle optimization algorithm also includes: after each iteration in each stage, trunculating out-of-bounds solutions using the following formula:

[0025] .

[0026] According to another aspect of the present invention, a ship target segmentation system based on an improved Harris Eagle optimization algorithm is proposed, the system comprising:

[0027] The region of interest acquisition module is configured to process airborne grayscale images containing the sky, sea, ships at sea, and coastline to acquire region of interest images.

[0028] The optimal segmentation image acquisition module is configured to obtain the optimal segmentation threshold using the improved Harris Eagle optimization algorithm, and to segment the airborne grayscale image using the optimal segmentation threshold to obtain a segmented image based on the optimal segmentation threshold.

[0029] The ship target segmentation module is configured to fuse the region of interest image and the segmented image based on the optimal segmentation threshold to obtain the final segmented image.

[0030] Furthermore, the process of processing the airborne grayscale image containing the sky, sea, ships, and coastline in the region of interest acquisition module to acquire the region of interest image includes:

[0031] The SIFT algorithm is used to extract multiple key points from the airborne grayscale image, including: constructing a Gaussian difference pyramid and detecting local extrema; filtering multiple local extrema based on contrast and edge response to obtain multiple key points;

[0032] Ellipse fitting is performed on multiple key points to obtain the smallest bounding ellipse containing all key points.

[0033] The image region corresponding to the smallest bounding ellipse is assigned white, and the remaining image regions are assigned black, thus generating a preliminary region of interest mask feature map.

[0034] Edge detection and land-sea segmentation line extraction are performed on the airborne grayscale image; the lower part of the land-sea segmentation line is assigned white and the upper part is assigned black to generate a land-sea segmentation mask;

[0035] The region of interest image is obtained by multiplying the initial region of interest mask feature map and the land-sea segmentation mask map pixel by pixel.

[0036] Furthermore, the step of obtaining the optimal segmentation threshold using the improved Harris Eagle optimization algorithm in the optimal segmentation image acquisition module includes:

[0037] During the exploration phase, the formula for updating the position of individuals in the population is:

[0038] ;

[0039] In the formula, This represents the position of the i-th individual in the (t+1)-th iteration; This represents the position of a randomly selected individual in the population; r and q are uniformly random numbers between [0,1]. is a uniformly distributed random number in [0,1]; lb and ub are the lower and upper bounds of the search space, respectively;

[0040] During the transition from exploration to development, the energy loss model for eagles hunting prey, and the energy consumed by the prey during its escape, is as follows:

[0041] ;

[0042] In the formula, Indicates a nonlinear decay exponent; This represents the initial escape energy of the prey; Indicates the exponent of the decay curve; Indicates the disturbance intensity coefficient; The noise is a standard normal distribution; t represents the current iteration number; T represents the maximum number of iterations.

[0043] During the development phase, the formula for updating the position of individuals in the population is:

[0044] ;

[0045] In the formula, This represents the optimal prey position in the t-th iteration; This represents the energy required for prey to escape. Indicates the jump range; This indicates an improved version of the Levi flight item. , This is the step size coefficient. For a standard normally distributed random variable, It's the Lévy Index. λ is the scaling factor, and λ is the iterative decay exponent.

[0046] Furthermore, the optimal segmentation threshold obtained using the improved Harris Eagle optimization algorithm in the optimal segmentation image acquisition module further includes: after each iteration in each stage, truncating out-of-bounds solutions using the following formula:

[0047] .

[0048] The beneficial technical effects of this invention are:

[0049] This invention proposes a method and system for segmenting ship targets in airborne infrared imagery based on an improved Harris Eagle optimization algorithm. By accurately identifying high-response regions of ship targets using SIFT features, and combining this with the improved Harris Eagle optimization algorithm to achieve adaptive threshold optimization within the target's grayscale band, targeted optimization segmentation is achieved, effectively solving the convergence stability problem under complex radiation environments. Specifically, the improved Harris Eagle optimization algorithm enhances its search adaptability under strong background interference by expanding the global search space; optimizes the exploration and development strategy to enhance its response to weak targets and improve the accuracy of local detail segmentation; incorporates an improved Levy flight mechanism to improve the ability to escape local optima, ensuring the algorithm's convergence performance in complex images; and optimizes the continuity of individual search paths, maintaining the integrity of the image's grayscale structure. This invention significantly suppresses the negative impacts of wave reflection and fog interference; significantly improves segmentation accuracy and robustness under complex sea conditions; and overcomes the challenge of balancing segmentation accuracy and real-time performance under dynamic sea conditions, providing highly reliable technical support for scenarios such as maritime monitoring and oil spill emergency response. Attached Figure Description

[0050] The above and other objects, features, and advantages of exemplary embodiments of the present invention will become readily apparent from the following detailed description taken in conjunction with the accompanying drawings. Several embodiments of the invention are illustrated in the drawings by way of example and not limitation, wherein:

[0051] Figure 1 This is a flowchart of the ship target segmentation method based on the improved Harris Eagle optimization algorithm described in an embodiment of the present invention;

[0052] Figure 2 This is an example image of the original grayscale image in an embodiment of the present invention;

[0053] Figure 3 This is an example of a SIFT feature point image preserved in an embodiment of the present invention;

[0054] Figure 4 This is an example of a minimum-fit ellipse image containing all SIFT feature points in an embodiment of the present invention;

[0055] Figure 5 This is an example of a preliminary region of interest mask feature map in an embodiment of the present invention;

[0056] Figure 6 This is an example of a Canny edge detection map in an embodiment of the present invention;

[0057] Figure 7 This is an example of extracting coastline images in an embodiment of the present invention;

[0058] Figure 8 This is an example of a land-sea segmentation mask in an embodiment of the present invention;

[0059] Figure 9 This is an example of a region of interest image in an embodiment of the present invention;

[0060] Figure 10 This is an example of image segmentation based on the optimal threshold in an embodiment of the present invention;

[0061] Figure 11 This is an example of the final ship segmentation image in this embodiment of the invention. Detailed Implementation

[0062] The principles and spirit of the invention will now be described with reference to several exemplary embodiments. It should be understood that these embodiments are given merely to enable those skilled in the art to better understand and implement the invention, and are not intended to limit the scope of the invention in any way. Rather, these embodiments are provided to make this disclosure more thorough and complete, and to fully convey the scope of this disclosure to those skilled in the art.

[0063] Those skilled in the art will recognize that embodiments of the present invention can be implemented as a system, apparatus, device, method, or computer program product. Therefore, this disclosure can be specifically implemented in the following forms: entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software. It should be understood herein that any number of elements in the accompanying drawings is for illustrative purposes only and not as a limitation, and any naming is for distinction only and has no limiting meaning.

[0064] This invention combines SIFT features with an improved Harris Eagle optimization algorithm to propose a ship target segmentation method and system based on SIFT features and the improved Harris Eagle optimization algorithm. This invention effectively characterizes the local differences between ship targets and the sea-sky background by extracting SIFT features from airborne infrared images. Simultaneously, it dynamically adjusts the segmentation threshold using the improved Harris Eagle optimization algorithm to adapt to target segmentation requirements under different sea conditions.

[0065] This invention proposes a ship target segmentation method based on an improved Harris Eagle optimization algorithm, such as... Figure 1 As shown, the method includes:

[0066] S1. Process the airborne grayscale image containing the sky, sea, ships and coastline to obtain the region of interest image;

[0067] S2. Obtain the optimal segmentation threshold using the improved Harris Eagle optimization algorithm, and segment the airborne grayscale image using the optimal segmentation threshold to obtain a segmented image based on the optimal segmentation threshold.

[0068] S3. Fuse the region of interest image and the segmented image based on the optimal segmentation threshold to obtain the final segmented image.

[0069] The method begins with S1, in which an airborne grayscale image containing the sky, sea, ships and coastline is processed to obtain the region of interest image.

[0070] According to an embodiment of the present invention, the airborne infrared image grayscale image is as follows: Figure 2 As shown. The specific steps of S1 include: S11, extracting multiple key points of the airborne grayscale image using the SIFT algorithm; including: constructing a Gaussian difference pyramid and detecting local extrema; filtering multiple local extrema based on contrast and edge response to obtain multiple key points; S12, performing ellipse fitting on multiple key points to obtain the minimum bounding ellipse containing all key points; assigning the image region corresponding to the minimum bounding ellipse to white and the remaining image regions to black to generate a preliminary region of interest mask feature map; S13, performing edge detection and land-sea segmentation line extraction on the airborne grayscale image; assigning the lower sea area of ​​the land-sea segmentation line to white and the upper area to black to generate a land-sea segmentation mask map; S14, multiplying the preliminary region of interest mask feature map and the land-sea segmentation mask map pixel by pixel to obtain the region of interest image.

[0071] First, in S11, the SIFT feature extraction method is used to extract key points from the original image, specifically including:

[0072] 1) Scale-space extremum detection: Potential keypoints are detected using the Difference of Gaussians (DoG) pyramid.

[0073]

[0074] in, It is a Gaussian kernel. These are Gaussian scales and parameters, where k is the scale factor, typically 1. I(m, n) is the image. The output is the difference between the blurred images at two different scales after calculation. In each layer of the DoG pyramid, local extrema are detected by comparing each pixel with its 26 neighboring pixels (8 neighbors at the same scale and 9×2 neighbors at the upper and lower scales).

[0075] 2) Precise key point localization:

[0076] 21) Taylor expansion fitting: At the candidate key point ( ), locally fit the DoG function D(x, y, σ) using a two-dimensional Taylor expansion:

[0077]

[0078] Where, , is a continuous variable in the scale space, representing any position and scale.

[0079]

[0080] is the offset of the key point (obtained by taking the derivative and setting it to zero).

[0081] 22) Reject low-contrast points: Substitute into the Taylor expansion formula to calculate the response (contrast) value of the key point:

[0082]

[0083] If |D( )| < tc, it is considered that the contrast of this key point is too low and it is rejected. In the embodiment of the present invention, tc is 0.01.

[0084] D( ) reflects the strength of the response near the key point and is the core area for judging whether this point is a valid key point. If D( ) is large, it means that this point is a significant extreme value in the image (such as a strong response at a corner or an edge); if D( ) is close to zero, it means that this point may be a weak response in noise or a flat area and has no distinguishability.

[0085] 23) Edge response filtering: Filter points with strong edge responses through the eigenvalue ratio of the Hessian matrix. The Hessian matrix is:

[0086]

[0087] Where, , , are the second-order derivatives of DoG. The eigenvalues α and β of the Hessian matrix (assuming α ≥ β) respectively represent the curvatures of the image in the two main directions at this point. The eigenvalue ratio γ:

[0088]

[0089] Among them, if If the value is not specified, it is considered an edge point and is removed. In this embodiment, ET represents the edge threshold, and for example, ET is set to 10. The image after extracting SIFT feature points is shown below. Figure 3 As shown.

[0090] Then, in S12, ellipse fitting is performed on multiple key points to construct a minimum bounding ellipse. This ellipse is centered on the data mean, and its size and tilt angle accurately reflect the overall distribution of the data and can cover all the feature points within it, such as... Figure 4 As shown, the image region corresponding to the smallest bounding ellipse is assigned white, and the remaining image regions are assigned black, generating a preliminary region of interest mask feature map, as shown. Figure 5 As shown.

[0091] Then, in S13, edge detection and land-sea segmentation line extraction are performed on the airborne grayscale image; the lower part of the land-sea segmentation line is assigned white, and the upper part is assigned black, generating a land-sea segmentation mask; among them, the Canny edge detection algorithm is used to extract edge features from the original image, such as... Figure 6 As shown; the longest edge is selected as the land-sea dividing line, such as... Figure 7 As shown; the generated land-sea segmentation mask is as follows. Figure 8 As shown.

[0092] Then, in S14, the initial region of interest mask feature map and the land-sea segmentation mask map are multiplied pixel by pixel to obtain the region of interest image, such as... Figure 9 As shown.

[0093] Then, S2 is executed. In S2, the improved Harris Eagle optimization algorithm is used to obtain the optimal segmentation threshold, and the airborne grayscale image is segmented using the optimal segmentation threshold to obtain a segmented image based on the optimal segmentation threshold.

[0094] According to an embodiment of the present invention, the existing Harris Hawks Optimization (HHO) algorithm includes the following steps:

[0095] 1) Population initialization: During the initialization phase, N initial solutions are randomly generated within the search space. (The eagle's position), each solution represents a possible candidate threshold value. This uniform random initialization ensures population diversity, and the calculation formula is:

[0096]

[0097] Where lb and ub are the lower and upper bounds of the search space, respectively; It is a uniformly distributed random number in [0,1].

[0098] 2) Exploration Phase: When prey is energetic, the eagle flock employs two random exploration strategies: cooperative surveillance based on random individual position updates (q≥0.5); and random exploration based on the population average position and random vector updates (q<0.5). This phase encourages global search and avoids premature convergence. Individual position updates use a formula based on randomly selecting individual positions from the population. The selected individuals are all screened based on image grayscale features, ensuring that the position update process is guided by grayscale information, thus making the search direction and step size more inclined towards regions with better grayscale feature performance. The calculation formula is:

[0099]

[0100] in, Let represent the position of the i-th individual in the eagle flock at generation t+1. The position of the eagle individual randomly selected in generation t; This represents the current optimal solution (prey location). The average position of the population; r1, r2, r3, r4, q are random numbers in [0, 1].

[0101] 3) The transition from exploration to development: During this period, while the flock of eagles hunts prey, the prey consumes energy during its escape. In the energy model, energy... , The range ∈ [-1,1] simulates the escape ability of the prey (optimal solution). |E|≥1: the prey is energetic, and the flock is in the exploration phase (global search); |E|<1: the prey is tired, and the flock enters the development phase (local fine search). Energy decreases linearly with iteration, ensuring that the algorithm focuses on exploration in the early stages and development in the later stages. The energy model is as follows:

[0102]

[0103] In the formula, As initial energy, ; t is the current iteration number; T is the maximum iteration number; rand() is a uniform random number in [0,1].

[0104] 4) Development Phase: Based on the prey escape energy E and the random number r (r∈[0,1]), the Harris Eagle optimization algorithm employs four capture strategies:

[0105] 41) Strategy 1 - Soft Encirclement: When 0.5 ≤ |E| < 1 and r ≥ 0.5, the prey has enough energy to jump away, and the flock of eagles uses soft encirclement to deplete its energy.

[0106]

[0107] in, It is the distance difference between the individual and the prey (optimal solution), h=2(1-r5) (r5∈[0,1] is a random number), used to control the jumping intensity of the prey's behavior.

[0108] 42) Strategy 2 - Hard Attack: When |E| < 0.5 and r ≥ 0.5, the prey is exhausted, and the flock of eagles launches a rapid attack:

[0109]

[0110] 43) Strategy 3 - Asymptotic rapid dive soft encirclement: When 0.5 ≤ |E| < 1 and r < 0.5, the eagle flock uses Levy flight to perform random walks, expanding the search area:

[0111]

[0112] in, , ; Jump amplitude J = 1.8(1-rand()); S is a D-dimensional random vector (elements ∈ [0,1]). For the Lévy flight function; Let be the fitness of the current individual i in generation t (i.e., the objective function value of the current solution).

[0113] In this embodiment of the invention, the objective function, i.e., the fitness function, is an improved method based on calculating the weighted inter-class variance using the image grayscale histogram, and its expression is:

[0114]

[0115] In the formula, The fitness function value; For class 0 weight (the proportion of pixels below the threshold), Weight for Class 1 (proportion of pixels above the threshold); The average degree is 0. ;reward represents the threshold reward coefficient, used to adjust the fitness weights according to the threshold size;T represents the control segmentation threshold;B represents the gray level range;p(i) represents the probability distribution of gray level i in the image.

[0116] Position updates are guided by the currently determined objective function value, which strengthens the gray-level direction of the local search and enables the population to quickly converge to the threshold solution with the best gray-level segmentation effect.

[0117] 44) Strategy 4 - Gradual, rapid dive and hard encirclement: When |E| < 0.5 and r < 0.5, the eagle flock constructs a hard encirclement to capture prey:

[0118]

[0119] in, .

[0120] 5) Boundary handling: Truncate out-of-bounds solutions to ensure all solutions are within the feasible region; the calculation formula is:

[0121]

[0122] The embodiments of the present invention improve upon the traditional Harris Eagle optimization algorithm as follows:

[0123] a) During the exploration phase, to expand the exploration range, simplify calculations, focus on the effective area, and adjust parameters, the population individual location update formula is modified as follows:

[0124]

[0125] Among them, X k r is the location of a random individual selected from the population; r,q~U(0,1) are uniform random numbers, r is used to control the perturbation amplitude during the search process; q is used to determine whether to execute a specific strategy.

[0126] b) During the transition from exploration to development, to retain more exploration space and ensure a rapid transition to the development phase when t approaches T, the escape energy model is modified as follows:

[0127]

[0128] in, It is a non-linear decay exponent; It is the exponent of the decay curve; It is the disturbance intensity coefficient; This is standard normally distributed noise.

[0129] c) During the development phase, to enhance local search and reduce unnecessary perturbations, the calculation formula is selected based on development efficiency, instead of choosing from four strategies. The position update formula in the original algorithm's four strategy selection is improved as follows:

[0130]

[0131] The above formula increases development efficiency from 50% to 85%; For the improved Levi flight project.

[0132] The Lévy flight algorithm generates a random step size that alternates between long hops and short hovers, simulating the dive pattern of an eagle flock during a surprise attack. This heavy-tailed distribution enhances the algorithm's ability to escape local optima. To avoid late-stage oscillations and optimize the step size variance, this embodiment of the invention further optimizes the Lévy flight formula, i.e., the improved Lévy flight term is:

[0133]

[0134] Where k is the step size coefficient (0.004 in this embodiment of the invention). λ is the Lévy index (1.48 in this embodiment); λ is the iteration decay index (0.68 in this embodiment); For a standard normally distributed random variable, This is the scaling factor.

[0135] d) In boundary handling, to preserve gradient information and improve convergence speed, elastic bounce is used instead of innovative interception. The improved calculation formula is:

[0136]

[0137] The improved Harris Eagle optimization algorithm described above is used to obtain the optimal segmentation threshold. This optimal threshold is then used to segment the airborne grayscale image, resulting in a segmented image based on the optimal threshold, such as... Figure 10 As shown.

[0138] Then, S3 is executed, in which the region of interest image and the segmented image based on the optimal segmentation threshold are fused to obtain the final segmented image.

[0139] According to an embodiment of the present invention, Figure 10 and Figure 9 Perform pixel-by-pixel multiplication and remove blemishes to obtain the final ship segmentation result, such as... Figure 11 As shown.

[0140] This invention also proposes a ship target segmentation system based on an improved Harris Eagle optimization algorithm, the system comprising:

[0141] The region of interest acquisition module is configured to process airborne grayscale images containing the sky, sea, ships at sea, and coastline to acquire region of interest images.

[0142] The optimal segmentation image acquisition module is configured to obtain the optimal segmentation threshold using the improved Harris Eagle optimization algorithm, and to segment the airborne grayscale image using the optimal segmentation threshold to obtain a segmented image based on the optimal segmentation threshold.

[0143] The ship target segmentation module is configured to fuse the region of interest image and the segmented image based on the optimal segmentation threshold to obtain the final segmented image.

[0144] In this embodiment, preferably, the process of processing the airborne grayscale image containing the sky, sea, ships, and coastline in the region of interest acquisition module to acquire the region of interest image includes:

[0145] The SIFT algorithm is used to extract multiple key points from the airborne grayscale image, including: constructing a Gaussian difference pyramid and detecting local extrema; filtering multiple local extrema based on contrast and edge response to obtain multiple key points;

[0146] Ellipse fitting is performed on multiple key points to obtain the smallest bounding ellipse containing all key points.

[0147] The image region corresponding to the smallest bounding ellipse is assigned white, and the remaining image regions are assigned black, thus generating a preliminary region of interest mask feature map.

[0148] Edge detection and land-sea segmentation line extraction are performed on the airborne grayscale image; the lower part of the land-sea segmentation line is assigned white and the upper part is assigned black to generate a land-sea segmentation mask;

[0149] The region of interest image is obtained by multiplying the initial region of interest mask feature map and the land-sea segmentation mask map pixel by pixel.

[0150] In this embodiment, preferably, the step of obtaining the optimal threshold using the improved Harris Eagle optimization algorithm in the optimal segmentation image acquisition module includes:

[0151] During the exploration phase, the formula for updating the position of individuals in the population is:

[0152] ;

[0153] In the formula, This represents the position of the i-th individual in the (t+1)-th iteration; This represents the position of a randomly selected individual in the population; r and q are uniformly random numbers between [0,1]. is a uniformly distributed random number in [0,1]; lb and ub are the lower and upper bounds of the search space, respectively;

[0154] During the transition from exploration to development, the energy loss model for eagles hunting prey, and the energy consumed by the prey during its escape, is as follows:

[0155] ;

[0156] In the formula, Indicates a nonlinear decay exponent; This represents the initial escape energy of the prey; It is the exponent of the decay curve; It is the disturbance intensity coefficient; This is standard normally distributed noise;

[0157] During the development phase, the formula for updating the position of individuals in the population is:

[0158] ;

[0159] In the formula, This represents the optimal prey position in the t-th iteration; This represents the energy required for prey to escape. Indicates the jump range; This indicates an improved version of the Levi flight item. , This is the step size coefficient. For a standard normally distributed random variable, It's the Lévy Index. λ is the scaling factor, and λ is the iterative decay exponent.

[0160] In this embodiment, preferably, the step of obtaining the optimal threshold using the improved Harris Eagle optimization algorithm in the optimal segmentation image acquisition module further includes: after each iteration in each stage, truncating out-of-bounds solutions using the following formula:

[0161] .

[0162] The functionality of the ship target segmentation system based on the improved Harris Eagle optimization algorithm described in this embodiment can be explained by the aforementioned ship target segmentation method based on the improved Harris Eagle optimization algorithm. Therefore, for the parts not detailed in the system embodiment, please refer to the above method embodiment, and they will not be repeated here.

[0163] It should be noted that although several units, modules, or sub-modules are mentioned in the detailed description above, this 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 can be embodied in one module. Conversely, the features and functions of one module described above can be further divided and embodied by multiple modules.

[0164] Furthermore, although the operations of the method of the present invention are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0165] While the spirit and principles of the invention have been described with reference to several specific embodiments, it should be understood that the invention is not limited to the disclosed specific embodiments, and the division of aspects does not imply that features in these aspects cannot be combined for benefit; such division is merely for ease of description. The invention is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims.

Claims

1. A ship target segmentation method based on an improved Harris Eagle optimization algorithm, characterized in that, include: Process airborne grayscale images containing the sky, sea, ships, and coastline to obtain region of interest images; An optimal segmentation threshold is obtained by using an improved Harris Hawk optimization algorithm, and the airborne grayscale image is segmented using the optimal segmentation threshold to obtain a segmented image based on the optimal segmentation threshold. The method of obtaining the optimal segmentation threshold using the improved Harris Eagle optimization algorithm includes: During the exploration phase, the formula for updating the position of individuals in the population is: ； In the formula, This represents the position of the i-th individual in the (t+1)-th iteration; This represents the position of a randomly selected individual in the population; r and q are uniformly random numbers between [0,1]. is a uniformly distributed random number in [0,1]; lb and ub are the lower and upper bounds of the search space, respectively; During the transition from exploration to development, the energy loss model for eagles hunting prey, and the energy consumed by the prey during its escape, is as follows: ； In the formula, Indicates a nonlinear decay exponent; This represents the initial escape energy of the prey; Indicates the exponent of the decay curve; Indicates the disturbance intensity coefficient; The noise is a standard normal distribution; t represents the current iteration number; T represents the maximum number of iterations. During the development phase, the formula for updating the position of individuals in the population is: ； In the formula, This represents the optimal prey position in the t-th iteration; This represents the energy required for prey to escape. Indicates the jump range; This indicates an improved version of the Levi flight item. , This is the step size coefficient. For a standard normally distributed random variable, It's the Lévy Index. λ is the scaling factor, and λ is the iteration decay exponent; The region of interest image and the segmented image based on the optimal segmentation threshold are fused to obtain the final segmented image.

2. The ship target segmentation method based on the improved Harris Eagle optimization algorithm according to claim 1, characterized in that, The process of processing an airborne grayscale image containing the sky, sea, ships, and coastline to obtain a region of interest image includes: The SIFT algorithm is used to extract multiple key points from the airborne grayscale image, including: constructing a Gaussian difference pyramid and detecting local extrema; filtering multiple local extrema based on contrast and edge response to obtain multiple key points; Ellipse fitting is performed on multiple key points to obtain the smallest bounding ellipse containing all key points. The image region corresponding to the smallest bounding ellipse is assigned white, and the remaining image regions are assigned black, thus generating a preliminary region of interest mask feature map. Edge detection and land-sea segmentation line extraction are performed on the airborne grayscale image; the lower part of the land-sea segmentation line is assigned white and the upper part is assigned black to generate a land-sea segmentation mask; The region of interest image is obtained by multiplying the initial region of interest mask feature map and the land-sea segmentation mask map pixel by pixel.

3. The ship target segmentation method based on the improved Harris Eagle optimization algorithm according to claim 2, characterized in that, The method of obtaining the optimal segmentation threshold using the improved Harris Eagle optimization algorithm also includes: after each iteration in each stage, trunculating out-of-bounds solutions using the following formula: 。 4. A ship target segmentation system based on an improved Harris Eagle optimization algorithm, characterized in that, include: The region of interest acquisition module is configured to process airborne grayscale images containing the sky, sea, ships at sea, and coastline to acquire region of interest images. The optimal segmentation image acquisition module is configured to obtain the optimal segmentation threshold using the improved Harris Eagle optimization algorithm, and to segment the airborne grayscale image using the optimal segmentation threshold to obtain a segmented image based on the optimal segmentation threshold. The method of obtaining the optimal segmentation threshold using the improved Harris Eagle optimization algorithm includes: During the exploration phase, the formula for updating the position of individuals in the population is: ； In the formula, This represents the position of the i-th individual in the (t+1)-th iteration; This represents the position of a randomly selected individual in the population; r and q are uniformly random numbers between [0,1]. is a uniformly distributed random number in [0,1]; lb and ub are the lower and upper bounds of the search space, respectively; During the transition from exploration to development, the energy loss model for eagles hunting prey, and the energy consumed by the prey during its escape, is as follows: ； In the formula, Indicates a nonlinear decay exponent; This represents the initial escape energy of the prey; Indicates the exponent of the decay curve; Indicates the disturbance intensity coefficient; The noise is a standard normal distribution; t represents the current iteration number; T represents the maximum number of iterations. During the development phase, the formula for updating the position of individuals in the population is: ； In the formula, This represents the optimal prey position in the t-th iteration; This represents the energy required for prey to escape. Indicates the jump range; This indicates an improved version of the Levi flight item. , This is the step size coefficient. For a standard normally distributed random variable, It's the Lévy Index. λ is the scaling factor, and λ is the iteration decay exponent; The ship target segmentation module is configured to fuse the region of interest image and the segmented image based on the optimal segmentation threshold to obtain the final segmented image.

5. The ship target segmentation system based on the improved Harris Eagle optimization algorithm according to claim 4, characterized in that, The process of processing an airborne grayscale image containing the sky, sea, ships, and coastline to obtain a region of interest image in the region of interest acquisition module includes: The SIFT algorithm is used to extract multiple key points from the airborne grayscale image, including: constructing a Gaussian difference pyramid and detecting local extrema; filtering multiple local extrema based on contrast and edge response to obtain multiple key points; Ellipse fitting is performed on multiple key points to obtain the smallest bounding ellipse containing all key points. The image region corresponding to the smallest bounding ellipse is assigned white, and the remaining image regions are assigned black, thus generating a preliminary region of interest mask feature map. Edge detection and land-sea segmentation line extraction are performed on the airborne grayscale image; the lower part of the land-sea segmentation line is assigned white and the upper part is assigned black to generate a land-sea segmentation mask; The region of interest image is obtained by multiplying the initial region of interest mask feature map and the land-sea segmentation mask map pixel by pixel.

6. The ship target segmentation system based on the improved Harris Eagle optimization algorithm according to claim 5, characterized in that, The optimal segmentation threshold obtained in the optimal segmentation image acquisition module further includes: after each iteration in each stage, truncating out-of-bounds solutions using the following formula: 。

Images