An oil spill monitoring method and system incorporating image analysis processing techniques

Abstract

The present application relates to the technical field of marine oil spill pollution monitoring, and particularly relates to an oil spill monitoring method and system combined with image analysis and processing technology. The method comprises the following steps: S1: obtaining an initial oil spill diffusion path through a remote sensing monitoring device, extracting a floe distribution image in a coverage area of the initial oil spill diffusion path, selecting the floe distribution image at the end of the initial oil spill diffusion path as a reference image, and establishing a correlation relationship of adjacent floe spatial distribution characteristics; S2: based on the correlation relationship, screening a floe distribution image with the lowest edge structure similarity and the weakest texture distribution consistency as a minimum feature sample; and realizing dynamic matching of oil-ice complex based on the edge structure attribute of the minimum feature sample. The present application establishes a quantitative relationship of edge structure and texture consistency based on the multi-modal correlation of floes, realizes accurate identification of oil film state and period analysis of morphological evolution, and improves the monitoring accuracy and environmental adaptability of polar oil spills.

Description

Technical Field

[0001] This invention relates to the field of marine oil spill pollution monitoring technology, and in particular to an oil spill monitoring method and system that combines image analysis and processing technology. Background Technology

[0002] With the increasing frequency of navigation along the Arctic shipping routes and the deepening of marine resource development activities, oil spill pollution monitoring in polar waters faces unprecedented technical challenges. These unique environments simultaneously exhibit multi-physics coupling effects, including dynamic displacement of sea ice, low-temperature solidification of oil slicks, and ocean current disturbances. Traditional single-point monitoring methods are insufficient to effectively capture the spatiotemporal evolution of oil-ice complexes. Particularly in the periglacial region, the nonlinear interaction between the sea ice fragmentation and reorganization process and the oil spill diffusion path leads to significant biases in pollution extent assessment.

[0003] The current monitoring system faces three major technical bottlenecks: First, the continuity of floating ice in remote sensing images is easily affected by polar day / polar night lighting conditions, leading to inaccurate boundary determination in modeling; second, the interaction effect between oil film and floating ice lacks quantitative evaluation standards, and existing threshold methods are insufficient to adapt to dynamic changes in ice porosity; third, phase transition processes in low-temperature environments accelerate the transformation of oil-ice morphology, and traditional static models cannot provide real-time feedback on the periodic characteristics of ice aggregation and fragmentation. These shortcomings severely restrict the timeliness of pollution diffusion prediction and the scientific validity of disposal plans. Summary of the Invention

[0004] To overcome the shortcomings of limited perception, slow response, and model lag, this invention provides an oil spill monitoring method and system that combines image analysis and processing technology.

[0005] The technical implementation scheme of the present invention is: an oil spill monitoring method combining image analysis and processing technology, comprising the following steps:

[0006] S1: Obtain the initial oil spill diffusion path through remote sensing monitoring equipment, extract the floating ice distribution image within the coverage area of ​​the initial oil spill diffusion path, select the floating ice distribution image at the end of the initial oil spill diffusion path as the reference image, and establish the correlation relationship of the spatial distribution characteristics of adjacent floating ice.

[0007] S2: Based on the aforementioned correlation, the ice distribution image with the lowest edge structure similarity and the weakest texture distribution consistency is selected as the minimum feature sample; and dynamic matching of oil-ice complex is achieved based on the edge structure attributes of the minimum feature sample.

[0008] S3: Combining spatiotemporal sequence analysis, we analyze the periodic changes in the ice body from fragmentation to aggregation during the morphological evolution of the oil-ice complex.

[0009] S4: Determine the oil spill diffusion monitoring method based on the aforementioned periodic variation characteristics.

[0010] Preferably, the step of acquiring the initial oil spill diffusion path through remote sensing monitoring equipment, extracting the distribution image of floating ice within the area covered by the initial oil spill diffusion path, selecting the floating ice distribution image at the end of the initial oil spill diffusion path as a reference image, and establishing the correlation relationship of spatial distribution characteristics of adjacent floating ice includes:

[0011] If the floating ice in the floating ice distribution image is continuously distributed, then the nearest neighbor monitoring path to the initial oil spill diffusion path is obtained, until a floating ice distribution image in which the floating ice is discontinuous is obtained;

[0012] The continuous distribution refers to the distribution of ice floes that are of uniform size and equidistant from each other;

[0013] Based on the discontinuous distribution image of the floating ice, a correlation relationship is established between the spatial distribution features of neighboring floating ice with the reference image as the center.

[0014] The adjacent monitoring path refers to a continuous monitoring trajectory that dynamically adjusts its extension distance along the vertical direction of the initial oil spill diffusion path based on the spatial distribution characteristics of floating ice, and is used to capture the discontinuous floating ice distribution area of ​​the actual oil spill diffusion boundary.

[0015] The correlation refers to the quantitative correlation between the edge structure similarity and texture distribution consistency between spatially adjacent ice floes.

[0016] Preferably, the correlation refers to the quantitative correlation between the edge structure similarity and texture distribution consistency among spatially adjacent ice floes, including:

[0017] The edge structure similarity is calculated by the dynamic time-normalized distance of the ice floe contour curvature sequence to obtain the edge structure similarity quantification value;

[0018] The texture distribution consistency is obtained by joint feature correlation analysis of the directional gradient histogram and the gray-level co-occurrence matrix to obtain a quantitative value of texture distribution consistency;

[0019] The quantitative correlation adopts adaptive weighted fusion and Sigmoid normalization to output normalized correlation strength coefficients.

[0020] Preferably, the step of selecting the ice floe distribution image with the lowest edge structure similarity and the weakest texture distribution consistency as the minimum feature sample based on the correlation includes:

[0021] Obtain the edge structure similarity quantization value and the texture distribution consistency quantization value of each ice floe distribution image in the association relationship;

[0022] All ice floe distribution images are sorted in ascending order based on the edge structure similarity quantization value to obtain the first ascending order sorting result;

[0023] Based on the first ascending sort result, select ice floe distribution images with a preset proportion from the first ascending sort result to form a candidate group;

[0024] The candidate groups are sorted in ascending order based on the texture distribution consistency quantization value to obtain a second ascending order sorting result.

[0025] Based on the second ascending sort result, the first ice floe distribution image in the second ascending sort result is selected as the smallest feature sample.

[0026] The minimum feature sample refers to the ice floe distribution image with the lowest edge structure similarity quantization value and texture distribution consistency quantization value selected through double sorting, representing the spatially isolated target with the greatest difference from neighboring ice floe distribution images.

[0027] Preferably, the step of dynamically matching the oil-ice complex based on the edge structure attributes of the smallest feature sample to obtain the dynamic matching result includes:

[0028] Extract the edge structure attributes of the smallest feature sample;

[0029] Calculate the Hausdorff distance between the oil-ice complex and the minimum feature sample;

[0030] The comparison result is obtained by comparing the Hausdorff distance with the dynamic threshold, and feature matching is performed based on the comparison result.

[0031] The oil-ice complex refers to the mixture of oil and ice formed when oil spill pollutants envelop floating ice.

[0032] Preferably, the step of comparing the Hausdorff distance with a dynamic threshold to obtain a comparison result, and performing feature matching based on the comparison result, includes:

[0033] Calculate the relative deviation between the Hausdorff distance and the dynamic threshold;

[0034] The oil film diffusion state is determined based on the magnitude of the relative deviation: when the relative deviation is less than or equal to zero, the oil film is determined to be in a permeation-dominated state; when the relative deviation is greater than zero, the oil film is determined to be in a flow-around-dominated state.

[0035] The dynamic threshold is dynamically generated based on the product of the perimeter of the smallest feature sample and the average distance between neighboring ice floes.

[0036] Preferably, the step of combining spatiotemporal sequence analysis to analyze the periodic changes in the ice body from fragmentation to aggregation during the morphological evolution of the oil-ice complex includes:

[0037] Based on the dynamic matching results, N floating ice distribution images collected at consecutive time intervals and the dynamic matching results of oil-ice complexes corresponding to the floating ice distribution images are extracted to form a spatiotemporal sequence image dataset.

[0038] The projected area of ​​the oil-ice complex at each time point is extracted from the spatiotemporal sequence image data and arranged in chronological order to form a time series of projected areas.

[0039] In the projected area time series, the continuous process of the projected area changing from a local minimum to a local maximum and back to a local minimum or from a local maximum to a local minimum is detected. The sequence segments that satisfy this change rule and whose time span does not exceed N1 time intervals Δt are identified as complete morphological evolution cycles.

[0040] In each complete morphological evolution cycle, the local minimum point of the projected area is marked as a small-scale node, the local maximum point is marked as a large-scale node, and the inflection point of the rate of change of the projected area is marked as a medium-scale node.

[0041] Extract the oil film diffusion state data corresponding to the small, medium, and large nodes, and establish a mapping relationship between node type and diffusion state;

[0042] The distribution pattern of oil film diffusion state at each node in multiple continuous complete morphological evolution cycles was statistically analyzed, and the periodic transformation characteristics of the permeation-dominated state and the flow-around-dominated state in the process of oil-ice complex from fragmentation to polymerization were output.

[0043] Preferably, the step of extracting the oil film diffusion state data corresponding to the small-sized, medium-sized, and large-sized nodes, and establishing a mapping relationship between node type and diffusion state, includes:

[0044] Extract the oil film diffusion state data corresponding to each of the small-scale, medium-scale, and large-scale nodes;

[0045] The frequency of occurrence of the infiltration-dominated state and the flow-around-dominated state under each node type was statistically analyzed;

[0046] Establish a deterministic mapping rule or probabilistic mapping relationship table between node type and diffusion state based on frequency distribution;

[0047] The node type refers to the morphological stages of the oil-ice complex divided according to its projected area, including small, medium, and large states.

[0048] The diffusion state refers to the movement pattern of the oil film in the gaps between ice floes, which is divided into the permeation-dominated state and the flow-around-dominated state.

[0049] Preferably, the method for determining oil spill diffusion monitoring based on the periodic variation characteristics includes:

[0050] If the frequency of the penetration-dominant state in the small-scale node exceeds a preset threshold, the oil film is determined to be in the high-speed diffusion stage, triggering a rapid containment command.

[0051] If the dominant flow state continues to appear at large-state nodes in two consecutive cycles, a predicted diffusion path map of the oil spill bypassing the large ice floe area will be generated.

[0052] If the deviation between the diffusion state and the mapping rule of the medium-sized node exceeds the tolerance limit, the diffusion state mapping rule recalibration process will be initiated.

[0053] An oil spill monitoring system combining image analysis and processing technology includes:

[0054] The floating ice boundary modeling module acquires oil spill diffusion path and floating ice distribution data, dynamically identifies the continuity status of floating ice, and automatically tracks adjacent monitoring paths until it captures non-continuous floating ice images as modeling boundaries.

[0055] The feature quantization and matching module calculates the quantization values ​​of the similarity of the edge structure and the consistency of the texture distribution of adjacent floating ice, and filters the smallest feature sample through double sorting to achieve dynamic matching of oil-ice complex based on its edge structure attributes;

[0056] The morphological evolution analysis module extracts the characteristics of the change in the projected area of ​​the oil-ice complex in the spatiotemporal sequence, identifies small-sized, medium-sized, and large-sized nodes within the complete morphological evolution cycle, and establishes a mapping relationship between node type and infiltration / flow diffusion state.

[0057] The decision calibration module triggers high-speed diffusion containment commands or detour path predictions based on periodic change characteristics. When the diffusion state of a morphological node deviates from the mapping rule, the diffusion state mapping rule recalibration process is initiated.

[0058] Beneficial Effects: This invention utilizes the multimodal correlation characteristics of ice floe spatial distribution to achieve quantitative analysis of the consistency of edge structure and texture among adjacent ice floes; it selects the spatially isolated target as the smallest feature sample through dual sorting, and accurately distinguishes oil film permeation and flow states by combining Hausdorff distance dynamic matching; it constructs morphological evolution cycle features based on projected area time series to analyze the phase transition law from fragmentation to aggregation; and it triggers a closed-loop decision-making mechanism through the mapping relationship between node type and diffusion state. This method significantly improves the accuracy and environmental adaptability of polar oil spill diffusion monitoring, effectively suppresses the interference of ice floe dynamic displacement and low-temperature phase transition, achieves real-time tracking and early warning of pollution paths, and provides reliable technical support for ecological protection in ice-covered areas. Attached Figure Description

[0059] Figure 1 This is a flowchart of the oil spill monitoring method combining image analysis and processing technology of the present invention;

[0060] Figure 2 This is a schematic diagram of the oil spill monitoring system based on the combined image analysis and processing technology of the present invention. Detailed Implementation

[0061] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0062] Example 1: An oil spill monitoring method combining image analysis and processing technology, such as Figure 1 As shown, it includes the following steps:

[0063] S1: Obtain the initial oil spill diffusion path through remote sensing monitoring equipment, extract the floating ice distribution image within the coverage area of ​​the initial oil spill diffusion path, select the floating ice distribution image at the end of the initial oil spill diffusion path as the reference image, and establish the correlation relationship of the spatial distribution characteristics of adjacent floating ice.

[0064] It should be noted that the initial oil spill diffusion path refers to the continuous dynamic diffusion trajectory of oil spill pollution captured in real time by remote sensing equipment; the floating ice distribution image refers to remote sensing image data containing information on the location, size, and spacing of floating ice; the reference image refers to a reference image located at the end of the oil spill diffusion path and having discontinuous floating ice distribution characteristics; and the spatial distribution characteristics of adjacent floating ice refers to the quantitative representation of the location distribution, size, shape, and spacing combination of floating ice within a spatial range centered on the reference image.

[0065] S2: Based on the aforementioned correlation, the ice distribution image with the lowest edge structure similarity and the weakest texture distribution consistency is selected as the minimum feature sample; and dynamic matching of oil-ice complex is achieved based on the edge structure attributes of the minimum feature sample.

[0066] S3: Combining spatiotemporal sequence analysis, we analyze the periodic changes in the ice body from fragmentation to aggregation during the morphological evolution of the oil-ice complex.

[0067] S4: Determine the oil spill diffusion monitoring method based on the aforementioned periodic variation characteristics.

[0068] The initial oil spill diffusion path is obtained using remote sensing monitoring equipment. Images of floating ice distribution within the area covered by the initial oil spill diffusion path are extracted. The floating ice distribution image at the end of the initial oil spill diffusion path is selected as a reference image. A correlation relationship is established between the spatial distribution characteristics of adjacent floating ice, including:

[0069] If the floating ice in the floating ice distribution image is continuously distributed, then the nearest neighbor monitoring path to the initial oil spill diffusion path is obtained, until a floating ice distribution image in which the floating ice is discontinuous is obtained;

[0070] The continuous distribution refers to the distribution of ice floes that are of uniform size and equidistant from each other;

[0071] Based on the discontinuous distribution image of the floating ice, a correlation relationship is established between the spatial distribution features of neighboring floating ice with the reference image as the center.

[0072] The adjacent monitoring path refers to a continuous monitoring trajectory that dynamically adjusts its extension distance along the vertical direction of the initial oil spill diffusion path based on the spatial distribution characteristics of floating ice, and is used to capture the discontinuous floating ice distribution area of ​​the actual oil spill diffusion boundary.

[0073] The correlation refers to the quantitative correlation between the edge structure similarity and texture distribution consistency between spatially adjacent ice floes.

[0074] It should be noted that in polar oil spill monitoring, remote sensing equipment (such as satellites or drones) first captures the initial spread trajectory of the oil spill pollution and extracts images of the floating ice distribution in the area covered by that path. If the floating ice in this area exhibits a continuous distribution characteristic—that is, the floating ice is of uniform size and regularly spaced—it indicates that the structure of this area is simple and cannot truly reflect the complex boundaries of the oil spill spread. In this case, the scanning distance needs to be dynamically adjusted along the vertical direction of the initial path to generate a neighboring monitoring path. This path expands in real time according to the actual distribution characteristics of the floating ice until it captures a discontinuous floating ice area, such as a broken floating ice band of mixed sizes and irregular spacing.

[0075] It should be noted that the quantitative determination of continuous distribution uses a relative standard: a coefficient of variation (standard deviation / mean) of ice floes below a set threshold is considered 'uniform in size', and a fluctuation range (standard deviation / mean distance) of the distance between adjacent ice floes less than a threshold is considered 'equal in distance'. The threshold is set based on historical marine data; for example, a coefficient of variation threshold of 0.15-0.25 is typically used in the Arctic region. The determination threshold must be configured during system initialization.

[0076] Example Explanation: Assuming the initial path image shows ice floes distributed uniformly in a grid pattern (continuous distribution), the scan will be offset to both sides. When a discontinuous distribution of "large ice floes interspersed with ice shards" is detected in a certain area, the image at the end of that area is selected as the reference image. Using this as the center, the correlation between the spatial distribution characteristics of neighboring ice floes is constructed, and the similarity of edge shapes between ice floes (through dynamic matching of contour curvature) and the consistency of surface texture (through gradient and grayscale feature fusion calculation) are quantitatively analyzed. This design ensures that the monitored target is located at the true physical boundary of the oil spill diffusion, avoiding identification errors in a uniform ice floe environment.

[0077] Ice floe image acquisition method: The remote sensing equipment is equipped with a multispectral or synthetic aperture radar sensor to identify the outline of the ice floe by the difference in reflectivity of ice and water. Combined with image segmentation technology, the location, size and spacing information of the ice floe are extracted, and finally a digital image of the ice floe distribution is generated.

[0078] Association refers to the quantitative correlation between the edge structure similarity and texture distribution consistency among spatially adjacent ice floes, including:

[0079] The edge structure similarity is calculated by the dynamic time-normalized distance of the ice floe contour curvature sequence to obtain the edge structure similarity quantification value;

[0080] The texture distribution consistency is obtained by joint feature correlation analysis of the directional gradient histogram and the gray-level co-occurrence matrix to obtain a quantitative value of texture distribution consistency;

[0081] The quantitative correlation adopts adaptive weighted fusion and Sigmoid normalization to output normalized correlation strength coefficients.

[0082] It should be noted that in edge structure similarity calculations, the shape of the ice floe profile undergoes non-linear changes due to ice fragmentation. Dynamic Time Warping (DTW) can align curvature sequences of different lengths, adapt to local profile stretching and shrinking, and quantify shape differences. The formula is: Where A =

[0083] [a1,a2,…,a TI ]: The curvature sequence of the baseline ice floe profile (coordinates of curvature points), B = [b1, b2, ..., b n []: Curvature sequence of neighboring ice floes' contours; π: Optimal alignment path, minimizing cumulative distance. Texture distribution consistency is calculated using the Histogram of Oriented Gradients (HOG) to capture the directional distribution of ice floe surface edges, and the Gray-Level Co-occurrence Matrix (GLCM) to describe gray-level spatial relationships. Joint analysis overcomes the limitations of single features and comprehensively evaluates texture consistency. The formula is: Where H = [h1, h2, ..., h k ]: HOG eigenvectors (gradient orientation histogram), G = [g1, g2, ..., g k ]: GLCM feature vectors (contrast, energy, etc.) Feature mean. Quantization association processing, adaptive weighted fusion dynamically allocates weights based on illumination or ice thickness to highlight dominant features; Sigmoid standardization compresses to [0,1] to eliminate the influence of dimensions and output interpretable association strength. The formula is: S = w e ·E+w t ·T, Among them, S norm : Standardized output, E: Edge similarity quantization value (DTW output), T: Texture consistency quantization value (Corr output), we ,w t Adaptive weights (e.g.) w t =1-w e σ is the characteristic variance, k, θ: Sigmoid parameters (k = 1, θ = 0.5 control the slope and translation).

[0084] It should be noted that the explanation and use of the normalized correlation strength coefficient: the normalized correlation strength coefficient (S) calculated in claim 3 norm This comprehensively quantifies the overall similarity between neighboring ice floe images in terms of edge structure and texture distribution, with values ​​ranging from [0,1]. norm The closer the value is to 0, the greater the difference in overall features between the ice floe image and its neighboring images, that is, the stronger the spatial isolation.

[0085] It should be noted that in the adaptive weight calculation, the feature variance σ E and σ T σ is calculated based on the quantized values ​​of all neighboring ice floes within the current monitoring window. E Let σ be the standard deviation of the quantized values ​​of edge structure similarity for all ice floes. T The standard deviation of the texture distribution consistency quantization value is used. The default values ​​of Sigmoid parameters k and θ are k=1 and θ=0.5, which are suitable for typical lighting conditions. When the ambient lighting or ice thickness changes significantly, the sensitivity of nonlinear response is enhanced by adjusting k, and θ is offset compensation based on the mean ice thickness. The specific parameters are calibrated through experiments in typical scenarios.

[0086] In the process of selecting the minimum feature samples in claim 4, this coefficient provides an effective measure of the overall isolation of the ice floe image. It can serve as an auxiliary reference or verification metric in the dual-ranking selection process, helping to confirm that the selected targets (i.e., images with the lowest edge and texture quantization values) are indeed the regions with the strongest spatial isolation, thereby reinforcing the rationality of the selection results. For example, by calculating the average S for each candidate image... norm The value further corroborates the degree of difference between it and its surrounding environment.

[0087] Based on the aforementioned correlation, images of ice floes with the lowest edge structure similarity and the weakest texture distribution consistency are selected as the minimum feature samples, including:

[0088] Obtain the edge structure similarity quantization value and the texture distribution consistency quantization value of each ice floe distribution image in the association relationship;

[0089] All ice floe distribution images are sorted in ascending order based on the edge structure similarity quantization value to obtain the first ascending order sorting result;

[0090] Based on the first ascending sort result, select ice floe distribution images with a preset proportion from the first ascending sort result to form a candidate group;

[0091] The candidate groups are sorted in ascending order based on the texture distribution consistency quantization value to obtain a second ascending order sorting result.

[0092] Based on the second ascending sort result, the first ice floe distribution image in the second ascending sort result is selected as the smallest feature sample.

[0093] The minimum feature sample refers to the ice floe distribution image with the lowest edge structure similarity quantization value and texture distribution consistency quantization value selected through double sorting, representing the spatially isolated target with the greatest difference from neighboring ice floe distribution images.

[0094] To further explain, obtaining the minimum feature sample is crucial in oil spill monitoring because it represents the spatially isolated region in the ice floe distribution image that differs most significantly from neighboring targets. Its unique edge and texture characteristics provide a highly discriminative benchmark for subsequent dynamic matching of oil-ice complexes. The selection of this sample employs a dual sorting mechanism: First, all images are sorted in ascending order based on edge structure similarity quantification values, generating a first ascending order result, with the image with the lowest value (the one with the greatest edge difference) prioritized. Second, a predetermined proportion (e.g., 20%) of images is selected from the first result to form a candidate group, which is then sorted in ascending order based on texture distribution consistency quantification values ​​to obtain a second ascending order result. Finally, the first image with the lowest texture value is selected as the minimum feature sample.

[0095] It should be noted that the preset ratio should balance the isolation of the target and the complexity of the environment: a lower ratio (e.g., 10%-20%) is suitable for areas with high ice floe density to improve screening specificity, while a higher ratio (e.g., 30%-40%) is suitable for low-density areas to avoid missed detections. The specific ratio is dynamically adjusted based on the dispersion of the spatial distribution of ice floes in historical monitoring data to ensure that the candidate group covers potentially isolated targets. The initial ratio is set at 20%, and then dynamically adjusted according to density.

[0096] This sequential design stems from the physical characteristics of ice floes: edge structure reflects the macroscopic deformation of the ice body's outline and is a core influencing factor in the oil spill diffusion path; prioritizing these edges allows for rapid identification of targets with abnormal shapes. Texture distribution describes surface microscopic features and serves as a secondary indicator in the candidate group, refining the selection process to ensure that samples are highly isolated from their surrounding environment in both structure and texture. For example, in Arctic ocean monitoring, analyzing 100 ice floe images: first, they are sorted by edge similarity, with the top 20 candidate images (such as ice bodies with severely broken outlines) selected; then, they are sorted by texture consistency, with the top image (such as the image with the most irregular surface cracks) chosen. This sample will accurately identify isolated ice areas easily permeable by oil spills. This process balances efficiency and accuracy, enhancing the robustness of monitoring in complex environments.

[0097] It should be noted that edge enhancement processing is performed on the selected smallest feature sample ice targets: the Canny operator is used to detect sub-pixel level contours and remove noise interference; curvature feature points are sampled along the contour line to generate an ordered coordinate sequence (point set A); based on the average distance D between neighboring ice floes, the contour point set is adaptively downsampled to ensure that the point distance is ≤D / 3.

[0098] Based on the edge structure attributes of the minimum feature sample, dynamic matching of the oil-ice complex is achieved to obtain dynamic matching results, including:

[0099] Extract the edge structure attributes of the smallest feature sample;

[0100] Calculate the Hausdorff distance between the oil-ice complex and the minimum feature sample;

[0101] The comparison result is obtained by comparing the Hausdorff distance with the dynamic threshold, and feature matching is performed based on the comparison result.

[0102] The oil-ice complex refers to the mixture of oil and ice formed when oil spill pollutants envelop floating ice.

[0103] To further explain, in oil spill monitoring, dynamic matching based on the edge structure attributes of the smallest feature sample can accurately identify morphological changes in oil-ice complexes. The smallest feature sample, as a spatially isolated target, has an edge contour (such as sharp edges or broken shapes) that is sensitive to oil film diffusion and can effectively reflect the physical characteristics of oil-ice interaction. The matching process requires unified processing: first, the contour of the smallest feature sample and the oil film boundary of the oil-ice complex are extracted into a binary edge point set to eliminate the influence of material differences on geometric calculations.

[0104] Hausdorff distance calculation formula: Where A: the set of edge points of the minimum feature sample, B: the set of boundary points of the oil film in the oil-ice complex, d(a,b): the Euclidean distance between point pairs, a∈A: sample edge points, b∈B: oil film boundary points. Maximum and minimum distance in one direction.

[0105] Dynamic threshold setting and matching are implemented. The dynamic threshold is generated by the product of the perimeter L of the smallest feature sample contour and the average distance D between neighboring ice floes: T dynamic =α·L·D, (T dynamic The threshold is dynamic, and α is an environmental correction factor (default 1.2). Calculate the Hausdorff distance H and T. dynamic The relative deviation δ = (HT dynamic ) / T dynamic δ is the relative deviation: δ≤0: oil film penetration is dominant (small distance, oil film penetrates into ice gaps), δ>0: oil film flow around the ice is dominant (large distance, oil film diffuses around the ice).

[0106] It should be noted that the coefficient α is used to fine-tune the dynamic threshold according to actual environmental conditions. Its default value of 1.2 is set based on empirical data in typical Arctic monitoring scenarios. In practical applications, this value is appropriately adjusted according to changes in environmental factors such as wind speed and temperature. For example, α can be appropriately increased under strong wind or extremely low temperature conditions so that the dynamic threshold can better reflect the impact of environmental changes on oil film diffusion behavior.

[0107] It should be noted that the default value of the environmental correction factor α is 1.2 for classic scenario verification. In actual applications, an environmental compensation term is added: when the wind speed increases, the α value is increased as needed to cope with the accelerated oil film diffusion; in low-temperature environments, it is finely adjusted according to changes in oil viscosity. The compensation factor is calibrated through wind tunnel or low-temperature water tank experiments, and the system provides a manual calibration interface.

[0108] It should be noted that the oil film boundary of the oil-ice complex is extracted through the following steps: a. Gray-scale thresholding is performed on the remote sensing image, and the backscattering characteristics of synthetic aperture radar (SAR) are used to distinguish the oil film, floating ice, and seawater regions; b. Morphological closure operation is performed on the oil film region to fill the pores and smooth the boundary. It should be noted that the oil film boundary extraction adopts an adaptive segmentation strategy: the gray-scale threshold is automatically calculated using the Otsu algorithm, and the SAR backscattering characteristics are combined with the typical oil film scattering intensity range (e.g., -15 to -10 dB) to assist in the discrimination. The oil film scattering intensity range is calibrated according to the actual image. The size of the structuring element of the morphological closure operation is determined according to the image resolution and the minimum feature scale of the local floating ice, for example, 3-5 times the resolution scale; c. The closed contour of the oil film region is extracted and converted into a binary edge point set coordinate sequence.

[0109] Example: The perimeter of the smallest feature sample contour is L = 20m, and the average distance between adjacent ice floes is D = 2m, then Tdynamic =48m. If the measured H = 35m (δ = -0.27), it is determined to be a seepage state, triggering a rapid containment command. This mechanism quantifies the oil film movement pattern through geometric features, adapting to the dynamic environment of floating ice.

[0110] The comparison result is obtained by comparing the Hausdorff distance with a dynamic threshold, and feature matching is performed based on the comparison result, including:

[0111] Calculate the relative deviation between the Hausdorff distance and the dynamic threshold;

[0112] The oil film diffusion state is determined based on the magnitude of the relative deviation: when the relative deviation is less than or equal to zero, the oil film is determined to be in a permeation-dominated state; when the relative deviation is greater than zero, the oil film is determined to be in a flow-around-dominated state.

[0113] The dynamic threshold is dynamically generated based on the product of the perimeter of the smallest feature sample and the average distance between neighboring ice floes.

[0114] To further explain, the relative deviation is essentially a measure of the spatial fit between the oil film boundary and the ice floe profile. Its value is calculated by subtracting the dynamic threshold from the Hausdorff distance and then dividing by the dynamic threshold, reflecting the physical strength of the obstruction to oil film diffusion by the ice floe. When the relative deviation is less than or equal to zero, it indicates that the oil film boundary is highly fitted to the edge of the ice floe profile, and the oil is rapidly spreading through ice gaps. This state is called the penetration-dominated state, requiring immediate containment measures. When the relative deviation is greater than zero, it indicates that the oil film boundary is far from the ice floe profile, and the oil is spreading around the surface of large ice bodies. This state is called the flow-around-dominated state, and the pollution will form branch paths along the ice edge.

[0115] The zero-value threshold is set based on the critical point of the ice floe barrier effect: when the deviation is zero, the oil film exactly covers the outline of the ice floe, at which point the infiltration and flow around the floe reach equilibrium. The infiltration-dominated state represents the capillary movement of oil within the gaps in the ice floe, easily leading to deep-seated pollution; the flow-dominated state characterizes the lateral spread of oil along the ice edge, expanding the pollution area. For example, in Arctic monitoring, a negative relative deviation indicates that the oil film has infiltrated into the gaps in the ice floe, triggering emergency response; a positive value indicates that the oil film will bypass the iceberg, requiring an update to the diffusion path prediction method. This judgment mechanism accurately captures the dynamic characteristics of oil-ice interaction through geometric relationships.

[0116] By combining spatiotemporal sequence analysis, the periodic changes in the morphological evolution of the oil-ice complex, from ice fragmentation to aggregation, are analyzed, including:

[0117] Based on the dynamic matching results, N floating ice distribution images collected at consecutive time intervals and the dynamic matching results of oil-ice complexes corresponding to the floating ice distribution images are extracted to form a spatiotemporal sequence image dataset.

[0118] The projected area of ​​the oil-ice complex at each time point is extracted from the spatiotemporal sequence image data and arranged in chronological order to form a time series of projected areas.

[0119] In the projected area time series, the continuous process of the projected area changing from a local minimum to a local maximum and back to a local minimum or from a local maximum to a local minimum is detected. The sequence segments that satisfy this change rule and whose time span does not exceed N1 time intervals Δt are identified as complete morphological evolution cycles.

[0120] In each complete morphological evolution cycle, the local minimum point of the projected area is marked as a small-scale node, the local maximum point is marked as a large-scale node, and the inflection point of the rate of change of the projected area is marked as a medium-scale node.

[0121] Extract the oil film diffusion state data corresponding to the small, medium, and large nodes, and establish a mapping relationship between node type and diffusion state;

[0122] The distribution pattern of oil film diffusion state at each node in multiple continuous complete morphological evolution cycles was statistically analyzed, and the periodic transformation characteristics of the permeation-dominated state and the flow-around-dominated state in the process of oil-ice complex from fragmentation to polymerization were output.

[0123] To further explain, in oil spill monitoring, continuously timed data acquisition is fundamental to analyzing the dynamic evolution of oil-ice complexes, ensuring the consistency of morphological changes. The spatiotemporal sequence image dataset integrates images of floating ice distribution and their corresponding oil film diffusion states (provided by dynamic matching results) to form a multidimensional data matrix with temporal and spatial correlation. The specific technical principle is as follows: remote sensing equipment periodically photographs the floating ice area, and the system synchronously records the oil-ice complex matching results (permeation or flow around the surface) for each frame of the image, constructing a timestamped sequential dataset.

[0124] The projected area of ​​the oil-ice complex is extracted through binarization segmentation: first, the oil film boundary identified by dynamic matching is converted into a closed contour; then, the pixel area within the contour is calculated and converted into the actual spatial size. The time series of the projected area is a sequence of area values ​​arranged in chronological order, reflecting the contraction and expansion patterns of oil film diffusion.

[0125] The determination of a complete morphological evolution cycle is based on the physical process: when the projected area undergoes a continuous process from a local minimum (peak of ice fragmentation) to a local maximum (complete ice aggregation), and then back to a local minimum, or vice versa, and the duration does not exceed a preset threshold (e.g., N1 = 5 Δt), it is considered a complete cycle. This oscillation process is essentially the periodic splitting and reorganization of floating ice under the influence of ocean currents.

[0126] Node labeling rules are derived from morphological features:

[0127] Small-scale nodes: local minimum points of projected area, representing the highly fragmented state of ice floes;

[0128] Large-scale nodes: local maximum points of projected area, representing the fully aggregated state of floating ice;

[0129] Medium-type nodes: Inflection points in the rate of change of projected area (zero point of the first derivative), i.e., the critical moments when the area growth rate changes from positive to negative or vice versa. It should be noted that the inflection point detection of the rate of change of projected area includes a noise suppression mechanism: first, the area time-series data is smoothed using a Savitzky-Golay filter, and then the rate of change is calculated using the central difference method. The inflection point must simultaneously satisfy the sign change condition (zero point of the first derivative) and the change magnitude threshold (e.g., setting a minimum rate of change threshold) to eliminate false inflections caused by noise interference.

[0130] By statistically analyzing the oil film diffusion state at various node locations across multiple cycles (e.g., small-scale nodes exhibiting permeation dominance in 80% of cases), a mapping relationship between node type and diffusion state is established, ultimately outputting periodic transformation characteristics. For example, Arctic monitoring data shows that small-scale nodes exhibit a permeation dominance probability exceeding 75%, indicating accelerated oil film infiltration during the ice-breaking period; large-scale nodes exhibit a flow-around dominance probability of 90%, revealing oil film diffusion along the ice edge during the aggregation period. This mechanism quantifies the phase transition law of oil-ice interaction.

[0131] Extract the oil film diffusion state data corresponding to the small, medium, and large nodes, and establish a mapping relationship between node type and diffusion state, including:

[0132] Extract the oil film diffusion state data corresponding to each of the small-scale, medium-scale, and large-scale nodes;

[0133] The frequency of occurrence of the infiltration-dominated state and the flow-around-dominated state under each node type was statistically analyzed;

[0134] Establish a deterministic mapping rule or probabilistic mapping relationship table between node type and diffusion state based on frequency distribution;

[0135] The node type refers to the morphological stages of the oil-ice complex divided according to its projected area, including small, medium, and large states.

[0136] The diffusion state refers to the movement pattern of the oil film in the gaps between ice floes, which is divided into the permeation-dominated state and the flow-around-dominated state.

[0137] To further explain, the logic behind establishing the mapping relationship between node types and diffusion states is as follows: First, extract the oil film diffusion state data corresponding to each morphological node. This data originates from the dominance of permeation or flow around the node determined by the dynamic matching process. For example, for small-scale nodes, the moment when the projected area of ​​the oil-ice complex is smallest in the time series is recorded simultaneously with the oil film diffusion state value at that moment. These data represent the physical pattern of oil spill movement at a specific morphological stage, reflecting the constraint effect of the floating ice structure on diffusion.

[0138] The frequency analysis aims to uncover patterns: by analyzing data from multiple complete evolution cycles (e.g., 10 cycles), the proportion of the two diffusion states under each node type is calculated. For example, if permeation-dominant occurs 85 times among 100 small-scale nodes, it indicates that oil film permeation during the ice-breaking period is a high-probability event. Frequency distribution reveals a stable correlation between morphology and diffusion, based on which mapping rules are constructed: Deterministic mapping rule: when the frequency of a single state of a node type exceeds a threshold (e.g., ≥90%), a forced correlation is established. For example, if the frequency of flow-around dominance reaches 95% for large-scale nodes, then the rule "when the oil-ice complex is in the large-scale stage, the diffusion state is determined to be flow-around dominant" is generated. Probability mapping relationship table: if the state distribution is balanced (e.g., 55% permeation / 45% flow-around for medium-scale nodes), probability values ​​are output for decision-making reference.

[0139] Example: Arctic monitoring data shows that over 30 consecutive cycles: small-scale nodes: permeability-dominated occurrence rate 82%; medium-scale nodes: permeability-dominated 58% and flow-around-dominated 42%; large-scale nodes: flow-around-dominated 93%. Therefore, the following rule is established: small-scale nodes trigger a default permeability warning; large-scale nodes trigger flow-around-path prediction; and medium-scale nodes use a probability distribution decision method. This mechanism links morphological evolution with oil film dynamics, achieving accurate state prediction.

[0140] The method for monitoring oil spill diffusion based on the aforementioned periodic variation characteristics includes:

[0141] If the frequency of the penetration-dominant state in the small-scale node exceeds a preset threshold, the oil film is determined to be in the high-speed diffusion stage, triggering a rapid containment command.

[0142] If the dominant flow state continues to appear at large-state nodes in two consecutive cycles, a predicted diffusion path map of the oil spill bypassing the large ice floe area will be generated.

[0143] If the deviation between the diffusion state and the mapping rule of the medium-sized node exceeds the tolerance limit, the diffusion state mapping rule recalibration process will be initiated.

[0144] To further explain, the frequency of penetration dominance at small-scale nodes exceeding a threshold triggers containment control. When statistical data shows that the frequency of penetration dominance at small-scale nodes (during the highly fragmented ice phase) exceeds a set threshold (e.g., 75%), it indicates that the oil film has a tendency to spread rapidly in the ice-fragmented area. At this point, a rapid containment control command is triggered. The principle is that the ice-fragmented environment significantly increases the oil-soil contact area, theoretically accelerating the penetration process. Example: Suppose that within a certain monitoring period, 16 out of 20 small-scale nodes exhibit penetration dominance (frequency 80%), exceeding the threshold, and the system activates the containment control plan.

[0145] Large-scale nodes continuously generate dominant flow path predictions. If a dominant flow state is detected in two consecutive complete evolution cycles of large-scale nodes (iceberg aggregation stage), it indicates that the oil slick is systematically bypassing a large iceberg group. Based on this, a bypass path prediction map is generated. The principle is that the aggregated ice forms a physical barrier that changes the direction of the oil spill flow. Example: Assuming that the large-scale nodes in the 3rd and 4th cycles both show dominant flow, a schematic diagram of the diffusion path bypassing the large iceberg area is automatically generated.

[0146] The diffusion path prediction map is based on the dominant state of the large ice floes detected in two consecutive cycles. It extracts the spatial distribution profile of the current ice floes, generates the minimum resistance path along the geometric tangent direction of their edges, and extrapolates the oil film diffusion trajectory by combining real-time ocean current data.

[0147] If the state of a medium-sized node deviates from the predetermined diffusion rule, recalibration is initiated. This occurs when the actual diffusion state of a medium-sized node (the critical point of morphological transformation) deviates from the predetermined mapping rule by more than the tolerance limit (e.g., the rule predicts a penetration probability of 70%, but the actual monitored value is only 50%), indicating a change in environmental parameters. The recalibration process is then initiated, adjusting the judgment threshold using a dynamic verification formula. Example: A medium-sized node has a preset penetration probability of 60%. If five consecutive measured values ​​are below 40% (deviation > 20%), the threshold recalculation mechanism is triggered.

[0148] The essence of the design logic is as follows: Small-scale response: graded early warning for the risk of accelerated infiltration in the ice-breaking zone; Large-scale prediction: spatial path extrapolation based on continuous flow phenomena; Medium-scale calibration: maintaining the dynamic adaptability of diffusion state judgment rules. This mechanism transforms periodic characteristics into a progressive decision chain, realizing a closed-loop analysis from risk warning to spatial extrapolation.

[0149] It should be noted that the thresholds in the decision-making rules are exemplary values ​​(e.g., the frequency threshold for small-sized nodes dominating the infiltration state, the number of cycles of continuous flow dominance, and the deviation tolerance for medium-sized nodes) set based on the analysis of the patterns in the monitoring data (e.g., set to 75%, 2 cycles, and 20%, respectively). These values ​​are intended to provide operable judgment criteria and can be optimized and adjusted according to the environmental characteristics and early warning needs of the specific monitoring area during actual system deployment.

[0150] Example 2: Based on Example 1, such as Figure 2 As shown, an oil spill monitoring system combining image analysis and processing technology includes:

[0151] The floating ice boundary modeling module acquires oil spill diffusion path and floating ice distribution data, dynamically identifies the continuity status of floating ice, and automatically tracks adjacent monitoring paths until it captures non-continuous floating ice images as modeling boundaries.

[0152] The feature quantization and matching module calculates the quantization values ​​of the similarity of the edge structure and the consistency of the texture distribution of adjacent floating ice, and filters the smallest feature sample through double sorting to achieve dynamic matching of oil-ice complex based on its edge structure attributes;

[0153] The morphological evolution analysis module extracts the characteristics of the change in the projected area of ​​the oil-ice complex in the spatiotemporal sequence, identifies small-sized, medium-sized, and large-sized nodes within the complete morphological evolution cycle, and establishes a mapping relationship between node type and infiltration / flow diffusion state.

[0154] The decision calibration module triggers high-speed diffusion containment commands or detour path predictions based on periodic change characteristics. When the diffusion state of a morphological node deviates from the mapping rule, the diffusion state mapping rule recalibration process is initiated.

[0155] The present application has been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of the present application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present application. Therefore, the content of this specification should not be construed as a limitation of the present application.

Claims

1. An oil spill monitoring method combining image analysis and processing technology, characterized in that, Includes the following steps: S1: Obtain the initial oil spill diffusion path through remote sensing monitoring equipment, extract the floating ice distribution image within the coverage area of ​​the initial oil spill diffusion path, select the floating ice distribution image at the end of the initial oil spill diffusion path as the reference image, and establish the correlation relationship of the spatial distribution characteristics of adjacent floating ice. The correlation refers to the quantitative correlation between the edge structure similarity and texture distribution consistency between spatially adjacent ice floes, including: the edge structure similarity is calculated by the dynamic time-normalized distance of the ice floe contour curvature sequence to obtain the quantitative value of edge structure similarity; the texture distribution consistency is obtained by the joint feature correlation analysis of the directional gradient histogram and the gray-level co-occurrence matrix to obtain the quantitative value of texture distribution consistency; the quantitative correlation adopts adaptive weighted fusion and Sigmoid normalization to output the normalized correlation strength coefficient. S2: Based on the aforementioned correlation, the ice distribution image with the lowest edge structure similarity and the weakest texture distribution consistency is selected as the minimum feature sample; and dynamic matching of the oil-ice complex is achieved based on the edge structure attributes of the minimum feature sample, including: extracting the edge structure attributes of the minimum feature sample; Calculate the Hausdorff distance between the oil-ice complex and the minimum feature sample; compare the Hausdorff distance with a dynamic threshold to obtain a comparison result; and perform feature matching based on the comparison result, including: calculating the relative deviation between the Hausdorff distance and the dynamic threshold; The oil film diffusion state is determined based on the magnitude of the relative deviation: when the relative deviation is less than or equal to zero, the oil film is determined to be in a permeation-dominated state; when the relative deviation is greater than zero, the oil film is determined to be in a flow-around-dominated state; the dynamic threshold is dynamically generated based on the product of the contour perimeter of the smallest feature sample and the average distance between adjacent floating ice; the oil-ice complex refers to the oil-ice mixture formed after oil spill pollutants envelop floating ice. S3: Combining spatiotemporal sequence analysis, analyze the periodic changes in the morphological evolution of the oil-ice complex from fragmentation to aggregation, including: based on the dynamic matching results, extracting N ice floe distribution images collected at consecutive time intervals and the corresponding dynamic matching results of the oil-ice complex to form a spatiotemporal sequence image dataset; extracting the projected area of ​​the oil-ice complex at each time point from the spatiotemporal sequence image data, arranging them in chronological order to form a projected area time series; in the projected area time series, detecting continuous processes where the projected area changes from a local minimum to a local maximum and back to a local minimum, or from a local maximum to a local minimum and back to a local maximum, and identifying processes that satisfy this change pattern and whose time span does not exceed N1 time intervals. The sequence fragments are identified as complete morphological evolution cycles; in each complete morphological evolution cycle, the local minimum point of the projected area is marked as a small-scale node, the local maximum point is marked as a large-scale node, and the turning point of the rate of change of the projected area is marked as a medium-scale node; the oil film diffusion state data corresponding to the small-scale node, medium-scale node, and large-scale node are extracted, and the mapping relationship between node type and diffusion state is established; the distribution law of oil film diffusion state at each node in multiple consecutive complete morphological evolution cycles is statistically analyzed, and the periodic transformation characteristics of the permeation-dominated state and the flow-around-dominated state of the oil-ice complex from fragmentation to aggregation are output; S4: Determine the oil spill diffusion monitoring method based on the periodic change characteristics, including: if the frequency of the occurrence of the permeation-dominant state at the small-scale node exceeds a preset threshold, then determine that the oil film is in the high-speed diffusion stage and trigger a rapid containment command; if the flow-around-dominant state continues to occur at the large-scale node for two consecutive cycles, then generate a diffusion path prediction map of the oil spill bypassing the large floating ice area; if the diffusion state of the medium-scale node deviates from the mapping rule beyond the tolerance, then initiate the diffusion state mapping rule recalibration process.

2. The oil spill monitoring method using combined image analysis and processing technology according to claim 1, characterized in that, The process of acquiring the initial oil spill diffusion path using remote sensing monitoring equipment, extracting images of floating ice distribution within the area covered by the initial oil spill diffusion path, selecting the floating ice distribution image at the end of the initial oil spill diffusion path as a reference image, and establishing a correlation between the spatial distribution characteristics of adjacent floating ice includes: If the floating ice in the floating ice distribution image is continuously distributed, then the nearest neighbor monitoring path to the initial oil spill diffusion path is obtained, until a floating ice distribution image in which the floating ice is discontinuous is obtained; The continuous distribution refers to the distribution of ice floes that are of uniform size and equidistant from each other; Based on the discontinuous distribution image of the floating ice, a correlation relationship is established between the spatial distribution features of neighboring floating ice with the reference image as the center. The adjacent monitoring path refers to a continuous monitoring trajectory that dynamically adjusts its extension distance along the vertical direction of the initial oil spill diffusion path based on the spatial distribution characteristics of floating ice, and is used to capture the discontinuous floating ice distribution area of ​​the actual oil spill diffusion boundary. The correlation refers to the quantitative correlation between the edge structure similarity and texture distribution consistency between spatially adjacent ice floes.

3. The oil spill monitoring method using combined image analysis and processing technology according to claim 1, characterized in that, The step of selecting the ice floe distribution image with the lowest edge structure similarity and the weakest texture distribution consistency as the minimum feature sample based on the aforementioned correlation includes: Obtain the edge structure similarity quantization value and the texture distribution consistency quantization value of each ice floe distribution image in the association relationship; All ice floe distribution images are sorted in ascending order based on the edge structure similarity quantization value to obtain the first ascending order sorting result; Based on the first ascending sort result, select ice floe distribution images with a preset proportion from the first ascending sort result to form a candidate group; The candidate groups are sorted in ascending order based on the texture distribution consistency quantization value to obtain a second ascending order sorting result. Based on the second ascending sort result, the first ice floe distribution image in the second ascending sort result is selected as the smallest feature sample. The minimum feature sample refers to the ice floe distribution image with the lowest edge structure similarity quantization value and texture distribution consistency quantization value selected through double sorting, representing the spatially isolated target with the greatest difference from neighboring ice floe distribution images.

4. The oil spill monitoring method using combined image analysis and processing technology according to claim 1, characterized in that, The step of extracting oil film diffusion state data corresponding to small, medium, and large nodes, and establishing a mapping relationship between node type and diffusion state, includes: Extract the oil film diffusion state data corresponding to each of the small-scale, medium-scale, and large-scale nodes; The frequency of occurrence of the infiltration-dominated state and the flow-around-dominated state under each node type was statistically analyzed; Establish a deterministic mapping rule or probabilistic mapping relationship table between node type and diffusion state based on frequency distribution; The node type refers to the morphological stages of the oil-ice complex divided according to its projected area, including small, medium, and large states. The diffusion state refers to the movement pattern of the oil film in the gaps between ice floes, which is divided into the permeation-dominated state and the flow-around-dominated state.

5. An oil spill monitoring system using combined image analysis and processing technology, for use in the oil spill monitoring method using combined image analysis and processing technology as described in any one of claims 1-4, characterized in that, Including: The floating ice boundary modeling module acquires oil spill diffusion path and floating ice distribution data, dynamically identifies the continuity status of floating ice, and automatically tracks adjacent monitoring paths until it captures non-continuous floating ice images as modeling boundaries. The feature quantization and matching module calculates the quantization values ​​of the similarity of the edge structure and the consistency of the texture distribution of adjacent floating ice, and filters the smallest feature sample through double sorting to achieve dynamic matching of oil-ice complex based on its edge structure attributes; The morphological evolution analysis module extracts the characteristics of the change in the projected area of ​​the oil-ice complex in the spatiotemporal sequence, identifies small-sized, medium-sized, and large-sized nodes within the complete morphological evolution cycle, and establishes a mapping relationship between node type and infiltration / flow diffusion state. The decision calibration module triggers high-speed diffusion containment commands or detour path predictions based on periodic change characteristics. When the diffusion state of a morphological node deviates from the mapping rule, the diffusion state mapping rule recalibration process is initiated.

Images