Accurate continental shelf slope toe line extraction method considering terrain continuity

Abstract

The invention belongs to the technical field of marine surveying and mapping, and relates to a continental shelf slope toe line accurate extraction method considering terrain continuity. According to the method, on the basis of the precise recognition performance of rolling ball transformation on seabed negative topographic feature lines, two typical experimental groups are obtained by continuously adjusting the spatial scale threshold in the feature line extraction process and are named as a key group and a skeleton group respectively. On the basis, a concept of connected components in a graph theory is introduced to construct an importance grade division mode of a negative topographic feature line skeleton group, a multi-element continental shelf slope toe line screening process considering FOS geographic features is further fused, and finally, the accurate and efficient continental shelf slope toe line extraction method considering topographic continuity is realized. And a traditional method of collecting FOS construction connecting lines one by one is avoided through the thought of directly extracting the continental shelf slope toe lines, and the working efficiency of continental shelf external boundary delimiting work is improved to a certain extent.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of marine mapping, especially relates to the field of continental shelf outer limit delimitation and seabed topography synthesis, and particularly relates to a method for accurately extracting a continental shelf foot line by considering topographic continuity. BACKGROUND

[0002] The foot of the continental slope (FOS) is an important concept in geography and marine law. According to Article 76, paragraph 4 (b) of the United Nations Convention on the Law of the Sea, "The foot of the continental slope shall be taken to be the point at which the greatest change in gradient of the slope floor occurs, in the absence of evidence to the contrary." In a two-dimensional perspective, the degree of slope change of a topographic profile line is represented by the degree of curvature of the curve, which is represented by the curvature. In the three-dimensional terrain level, the ups and downs of the ground are represented by the terrain curvature. Therefore, from this perspective, the foot of the continental slope should correspond to the point with the maximum terrain curvature of the continental slope floor, which has the concave feature of the "negative topographic feature point" in geography.

[0003] Due to the complex geological structure in some areas, the seabed topography and geomorphology often vary abnormally, which makes it very difficult to identify and extract FOS using the definition of FOS in the Convention. Wu Ziyin et al. (A method for automatic identification and comprehensive judgment of the foot of the continental slope [J / OL]. Acta Geodaetica et Cartographica Sinica, 2014) started from the perspective of seabed topographic profile change, used two simplified methods of topographic second derivative extreme point and D-P algorithm to calculate the second derivative profile, and further identified and screened out FOS by combining multi-factor comprehensive judgment. However, there is no clear standard for seabed topographic profile division, and the data requirement is high, which cannot determine FOS using a single multi-beam profile. Zhuang Baojiang (Determination of the foot of the continental slope under complex geological background: A case study of the southern Mozambique continental margin [J]. Journal of Oceanography in China, 2024, 42(1): 13-22.) proposed a method based on average gradient change of water depth to calculate the position of FOS and combined with the comprehensive judgment of the opposite proof to complete the extraction of FOS with high precision. However, due to the limitations of seabed topographic profile objects, it is difficult to realize the global FOS identification and extraction work. Dong Jian, Zhang Zhiqiang et al. (TIN-DDM terrain feature line automatic extraction method based on rolling ball transformation [J]. Journal of National University of Defense Technology, 2024, 46(3): 167-179.) proposed the concept of "slope break point" in the process of TIN-DDM terrain feature line automatic extraction, which has certain inspiration for the extraction of the foot of the continental slope. However, this method is mainly applied to the extraction of terrain feature skeleton line and cannot be directly applied to the accurate extraction of the foot of the continental shelf. SUMMARY

[0004] To address the limitations of current methods for extracting continental shelf toe points, this invention builds upon the topographic feature line extraction method developed by Dong Jian et al., fully leveraging the accuracy advantage of the bidirectional rolling sphere transform method in terrain recognition. It constructs a negative topographic feature line skeleton and, by continuously changing the spatial scale threshold, obtains a set of candidate toe feature lines composed of FOS candidate points and a set of candidate skeleton feature lines that consider terrain continuity. Furthermore, it introduces the concept of connected components from graph theory, constructing connected components for both toe and skeleton feature lines. This allows for the calculation of the proportion of FOS candidate points in each skeleton feature line connected component, establishing a hierarchy of skeleton feature line importance, and integrating a multi-factor continental shelf toe line screening process, ultimately achieving accurate extraction of continental shelf toe lines. This invention, beyond finding candidate toe points, further facilitates the delineation of continental shelf toe lines, thus providing a solution to the continental shelf delimitation problem.

[0005] The technical solution adopted by the present invention to achieve the above objectives is as follows:

[0006] Methods for accurately extracting continental shelf slope toe lines that take into account topographic continuity include:

[0007] Step a: Rolling ball terrain recognition and negative terrain feature line extraction.

[0008] Step b: Extract negative terrain feature lines based on spatial scale threshold.

[0009] Step c: Classify the importance level of the candidate continental shelf toe line set.

[0010] Step d: Screening of multi-factor continental shelf toe lines.

[0011] In step a, rolling ball terrain recognition refers to using the maximum rolling ball radius at the point of contact state transition during the rolling ball transformation process as the critical rolling ball radius. The corresponding terrain location at this point is the terrain feature point, thus realizing the terrain feature recognition characteristic of rolling ball transformation. Specifically, as follows... Figure 2 The contact point states corresponding to (a), (b), and (c) are demonstrated. From sampling point A to sampling point D, the terrain slope gradually increases, the terrain curvature gradually increases, and the positive critical rolling ball radius r gradually decreases. The positive critical rolling ball radius is the smallest at sampling point D, which corresponds to the "negative terrain feature point" in the terrain.

[0012] Therefore, the principle of negative terrain feature line extraction is based on the terrain recognition characteristics of rolling sphere transformation. First, the positive critical rolling sphere radius and negative critical rolling sphere radius of each water depth point in the DDM are calculated. Then, the positive critical rolling sphere radius is sorted using the "reverse engineering" approach. The maximum point of the positive critical rolling sphere radius is used as the expansion center of the clustering subdivision unit. Then, the negative terrain feature points are continuously clustered and subdivision units are constructed using this as the expansion center. Finally, the intersection operation is performed between the positive terrain subdivision units that have been clustered. The intersection of the boundary lines of the subdivision units is the negative terrain feature line.

[0013] Step b is based on the negative terrain feature lines extracted in step a. For the extracted negative terrain feature lines, the terrain type is first determined by using a pre-set spatial scale threshold and combining the positive and negative rolling ball radii of the water depth point (formula (1)). Then, negative terrain feature points that meet the requirements are selected, and the spatial scale threshold is gradually reduced. Multiple experiments are conducted to select feature points with greater slope changes.

[0014] In step c, the steps for classifying the importance levels of the candidate continental shelf toe line set are as follows:

[0015] Step 1: Based on the principle of negative terrain feature line extraction according to spatial scale threshold, extract the "key group" negative terrain feature lines when the spatial scale threshold is low (the overall shape is short and scattered, but close to the hard requirement of "maximum slope change"). Then, extract the "skeleton group" negative terrain feature lines when the spatial scale threshold is high (the overall shape is a continuous network, but can still distinguish different terrain categories to some extent) (e.g. Figure 3 (a) and (b)).

[0016] Step 2: Introduce the concept of connected components in graph theory, and denote the feature lines of the skeleton group as S and the feature lines of the key group as I (Equation (2)). Construct the undirected graph G of the skeleton group feature lines S as follows (Equation (3)).

[0017] Step 3, let each edge (u,v)∈S, and create the adjacency list Adj(u) for each feature point u in the skeleton feature line as follows (Formula (4)).

[0018] Step 4: Use depth-first search to construct the connected components of the "skeleton group" feature lines and record the connected component array.

[0019] Step 5: Calculate the intersection ratio of each connected component array in the feature lines of the "skeleton group" with all feature points in the "key group".

[0020] Step 6: Set the importance level of each connected component according to the intersection ratio of each connected component (Formula (5)).

[0021] Step 7: Then, based on the size of the grade classification, further filter out the key negative terrain feature lines that take into account the continuity of the terrain.

[0022] In step d, the multi-factor continental shelf toe line screening process mainly includes the slope method, water depth method, and "noise reduction" method:

[0023] Step 1, Slope Method. The continental shelf toe line, as the starting line of the outer boundary of the continental shelf, should be located at the bottom of the continental slope geographically. As the transition zone between continental crust and oceanic crust, the continental slope often has a slope between 3 and 6°, with an average of 4°. First, record the natural neighborhood points of each depth point in the seafloor topography data, that is, the set of neighborhood depth points directly connected to the sampled depth point after constructing an irregular triangular network for the DDM. Then calculate the average slope of each depth point with respect to the natural neighborhood points, and on this basis, calculate the total average slope of each connected component. Only when the total average slope exceeds a certain threshold can it be retained (Formula (6)).

[0024] Step two, the slope method. Based on the average slope used to filter the connected components of the slope toe line, the average slope within the natural neighborhood of each sampling point is further set to not exceed the topographic slope range of the continental base slope, thereby further eliminating characteristic line segments from the continental shelf slope region from the results.

[0025] Step 3, Water Depth Method. Since the continental slope bottom should be located at the bottom of the continental slope from a geographical perspective, the topographic trend of the DDM data can be analyzed from the perspective of water depth to set an appropriate water depth threshold.

[0026] Step four, the "denoising" method. Negative terrain feature lines extracted by the rolling ball transform should be carefully selected. Each feature line segment in the key group must be greater than a certain number of water depth points to prevent the presence of "noise segments" in small terrain features from affecting the experimental results. Similarly, the segments in the skeleton group must also be greater than or equal to a certain number of water depth points to prevent some "noise segments" in the skeleton group from having fewer nodes, resulting in high values ​​when calculating the proportion of key points, and their connected components achieving high importance levels, thus leading to inaccurate final continental shelf toe line selection results.

[0027] Step five, the second method of "denoising". Since the algorithm of this invention is oriented towards water depth data in a spatial coordinate system, the problem of "small branch" segments inevitably appears when expanding the continental shelf toe line extraction results. Therefore, the line segment processing tool in MATLAB software can be used to delete the "small branch" segments in the continental shelf toe line and retain the "main line segments" in the continental shelf toe line, taking into account the aesthetics of the continental shelf toe line identification results to a certain extent, in accordance with the relevant provisions of Article 76 of the United Nations Convention on the Law of the Sea.

[0028]

[0029] Adj(u)={v|(u,v)∈S} (4)

[0030]

[0031] In the formula: R is the spatial scale threshold, r' is the positive critical rolling sphere radius, r” is the negative critical rolling sphere radius, Q(P i ) represents the water depth point P i The corresponding terrain type attribute;

[0032] The skeleton group feature lines generated based on the rolling ball transformation are S, and the key group feature lines are I. and Represents the two endpoints of the i-th negative terrain feature line segment. and Represents the two endpoints of the j-th negative terrain feature line segment;

[0033] V is the set of points in an undirected graph, containing all feature points, and E (or S) is the set of lines in an undirected graph.

[0034] Adj(u) is This forms an adjacency list with the feature point u;

[0035] R n t represents the importance level of the connected components. n m represents the number of terrain feature points contained in a connected component. n Indicates the number of feature points in the key group;

[0036] Slope(i,j) represents the water depth point P. i =(x i ,y i ,z i Relative to its neighboring point P j =(x j ,y j ,z j The slope value of n i Let P be the water depth. i The number of neighboring points, T i Let P be the water depth. i Average slope, N k For connected component C K The number of mid-water depth points, S k For connected component C K The average slope value.

[0037] The beneficial effects of this invention are:

[0038] Compared to traditional methods for automatic identification of continental shelf toe points, this invention creatively proposes a direct extraction of continental shelf toe lines from a three-dimensional perspective. Building upon previous methods that used rolling sphere transformation to extract negative topographic feature lines, this invention constructs a feature line hierarchy model based on spatial scale thresholds and integrates a continental shelf toe line screening process to ensure that the final extracted continental shelf toe lines both comply with the Commission's legal requirements for the FOS (Front Side of the Shelf) and relatively consider the continuity of the continental shelf's topography. Furthermore, this invention's approach of directly extracting continental shelf toe lines avoids the traditional method of collecting FOS points one by one to construct lines, thus improving the operational efficiency of continental shelf outer boundary delineation to a certain extent. Attached Figure Description

[0039] Figure 1 This is a schematic diagram of the method and analysis process of the present invention;

[0040] Figure 2 This is a schematic diagram illustrating the terrain feature identification characteristics based on the critical rolling ball radius;

[0041] Figure 3 This is a schematic diagram of terrain slope polygon extraction based on spatial scale threshold, where (a) is the "skeleton group" and (b) is the "key group".

[0042] Figure 4 The images are extracted step by step based on the negative topographic feature lines according to their importance level. From (a) to (d), the continental shelf toe lines that are retained have a higher importance level;

[0043] Figure 5 The "denoising" method preserves the "main line segment" of the continental shelf toe line, where (a) is the continental shelf toe line image before processing, and (b) is the continental shelf toe line image after processing. Detailed Implementation

[0044] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings and technical solutions.

[0045] To achieve the goal of accurately extracting continental shelf toe lines that take into account topographic continuity, the overall design of this invention includes three parts: (1) a topographic identification and negative topographic feature line extraction model; (2) a continental shelf toe line candidate line set evaluation model; and (3) a multi-element continental shelf toe line screening model. Specifically, as follows... Figure 1 As shown, the specific steps are as follows:

[0046] Step (1) processes the original DDM data through terrain identification and negative terrain feature line extraction models, identifies the rolling ball terrain, and extracts negative terrain feature lines. The specific process is as follows:

[0047] (1.1) Read the water depth point information in the digital depth model (DDM) and record the positive critical rolling ball radius and negative critical rolling ball radius of each water depth point based on the rolling ball transformation method. In mathematical form, the rolling ball transformation is equivalent to the combination of the positive and negative buffer surface transformations of the DDM, that is, construct the upper and lower buffer surfaces of the DDM, then take the upper and lower buffer surface points as the center of the rolling ball, and then construct equidistant spheres in sequence, and use mathematical methods to realize the physical rolling process of the rolling ball transformation. By continuously analyzing the numerical correlation between the rolling ball contact point and the rolling ball radius, the contact state between each sampling point and the dynamic rolling ball at different radii is determined, and then the positive critical rolling ball radius and negative critical rolling ball radius of each water depth point are obtained. (For the specific process, please refer to the literature "Zhang Zhiheng et al. Topographic morphology division and continuous scale expression of TIN-DDM based on rolling ball transformation [J]. Acta Geodaetica et Cartographica Sinica, 2020, 49(05): 644-655.").

[0048] (1.2) Reorder the water depth points based on the positive critical rolling sphere radius of each water depth point. Construct a Delaunay triangulation of the original water depth data using MATLAB and record the natural neighborhood points of each water depth point.

[0049] (1.3) Based on the sorting information of the positive critical rolling sphere radius of the water depth point and the information of natural neighborhood points, the center of the topographic subdivision unit is determined and clustered continuously until the seabed topographic region is completely subdivided. The center of the subdivision unit is the extreme point of the positive rolling sphere radius, and the boundary point of the subdivision unit is the negative topographic feature point.

[0050] (1.4) Extract the boundary lines of the subdivided units, which are all the negative terrain feature lines in the DDM. The nodes of each feature line segment in the negative terrain feature lines are the negative terrain feature points.

[0051] (1.5) For the extracted negative terrain feature lines, a spatial scale threshold is set. Each node on the feature line needs to determine the terrain type based on its critical rolling sphere radius, thereby filtering out terrain feature points with corresponding geographical features. As shown in formula (1).

[0052] The spatial scale threshold is shown in equation (1):

[0053]

[0054] In the formula: R is the spatial scale threshold, r' is the positive critical rolling sphere radius, and r” is the negative critical rolling sphere radius; Q(P i ) represents the water depth point P i The corresponding terrain type attribute, Q(P) i Q(P) represents positive terrain, where 1 represents positive terrain. i -1 represents negative terrain, Q(P) iQ(P) = 0 represents flat terrain. i A value of -2 represents a nested negative terrain within a positive terrain, Q(P) i ) represents a positive terrain nested within a negative terrain.

[0055] Using the criteria in formula (1), all negative terrain feature lines that meet the requirements are selected (including Q(P)). i (The terrain type attribute corresponding to -1 or -2). Because the clustering method used in the extraction of negative terrain feature lines follows the approach of extending outwards from the positive critical rolling sphere maxima of local terrain as the cluster center, the extracted feature lines do not meet the "maximum / minimum" requirement. Therefore, it is necessary to successively reduce the spatial scale threshold to extract a set of negative terrain feature lines with a smaller positive rolling sphere radius that better matches the maximum / minimum requirement of terrain curvature.

[0056] Multiple extraction experiments were conducted by progressively decreasing the spatial scale threshold. The "key group" of negative terrain feature lines (with the number of line segments controlled within 10% of the total number of negative terrain feature line segments) was selected when the spatial scale threshold was lower. The "skeleton group" of negative terrain feature lines (with the number of line segments controlled within 90% of the total number of negative terrain feature line segments) was selected when the spatial scale threshold was higher. Two experimental groups with typical feature line distribution characteristics were retained. (e.g.) Figure 3 In Figures (a) and (b), the red lines represent negative topographic feature lines extracted based on spatial scale thresholds. Figure (a) shows the negative topographic feature lines selected when the spatial scale threshold is high (hereinafter referred to as the skeleton group), while Figure (b) shows the negative topographic feature lines selected when the spatial scale threshold is low (hereinafter referred to as the key group). It can be seen that when the spatial scale threshold is low, the negative topographic feature lines extracted by the key group approach the rigid requirement of maximum slope variation, but the line segments are short and scattered. When the spatial scale threshold is high, the negative topographic feature lines extracted by the skeleton group appear as a continuous network, but different topographic categories can still be distinguished. Based on this, it is necessary to consider the differences between the two groups to achieve complementary advantages in the final results, extracting accurate, continuous continental shelf slope toe lines that take into account topographic variations.

[0057] Step (2) introduces connected component calculation, classifies the importance level of negative terrain feature lines through the continental shelf toe line candidate line set evaluation model, and selects the continental shelf toe lines. The specific steps are as follows:

[0058] (2.1) Depth-First Search (DFS), a commonly used graph search method, has low space complexity and is easy to implement, often used to solve path connectivity problems. Introducing the concept of connected components from graph theory, this paper uses DFS functions in Visual Studio 2019 to construct the connected components of the skeleton group feature lines, and indexes the depth sampling points in the connected components of the "skeleton group" feature lines, denoted as the connected array. The steps include:

[0059] (2.1.1) Record the skeleton group feature lines generated based on the rolling ball transformation as S, and generate the key group feature lines as I, as shown in formula (2).

[0060]

[0061] In the formula, and Represents the two endpoints of the i-th negative terrain feature line segment. and These represent the two endpoints of the j-th negative terrain feature line segment.

[0062] (2.1.2) Let each edge (u,v)∈S, indicating that a feature line segment can be formed between feature points u and v. The undirected graph G of the skeleton group feature lines S is constructed as follows, as shown in formula (3). Each line segment consists of a pair of feature points, and E contains all feature line segments.

[0063]

[0064] In the formula, V is the set of points in the undirected graph, containing all feature points, and E (or S) is the set of lines in the undirected graph.

[0065] (2.1.3) Traverse each edge (u,v)∈S, adding feature points u and v to the adjacency list of graph G. If feature point u or feature point v is not in the adjacency list of graph G, create a new adjacency list. Thus... The adjacency list Adj(u) formed with the feature point u is shown in formula (4).

[0066] Adj(u)={v|(u,v)∈S} (10)

[0067] In the formula: Adj(u) is An adjacency list is formed with the feature point u.

[0068] (2.1.4) Use the DFS function (Depth-First Search) to traverse the adjacency list of all feature points until all feature points connected to the given feature point have been visited. At this point, the skeleton feature lines are clustered into k connected components C, C = {C1, C2, ... C2}. K The feature points within each connected component are interconnected, forming a feature line (network).

[0069] (2.2) Calculate the number of intersection points between each connected component in the feature lines of the "skeleton group" and all feature points in the "key group". Let C n Let I be the nth connected component (1≤n≤k) in the skeleton group feature line S. Given that I is a key group feature line containing all slope toe candidate points, then the number of key group feature points m in the nth connected component is... n Represented as: m n =|C n ∩I|.

[0070] (2.3) Further calculate the proportion of key group negative terrain feature points contained in each connected component of the skeleton group negative terrain feature line under a larger spatial scale threshold, and then set the C of each connected component according to the intersection ratio of each connected component. n Importance level R n As shown in formula (5). The higher this ratio, the higher the importance level of the negative terrain feature line.

[0071]

[0072] In the formula, R n t represents the importance level of the connected components. n m represents the number of terrain feature points contained in a connected component. n This indicates the number of feature points in the key group.

[0073] (2.4) Based on the importance level classification, further filter out the continental shelf slope toe lines with higher importance levels that take into account topographic continuity. This process is as follows: Figure 4 As shown in (a), (b), (c), and (d), the importance level of the negative topographic feature lines retained from (a) to (d) gradually increases, which better reflects the geographical characteristics of the continental shelf toe lines. This not only further narrows down the selection range of continental shelf toe lines, but also ensures that the extracted continental shelf toe lines fully take into account topographic changes and their own continuity as line elements.

[0074] Step (3) further eliminates unreasonable negative terrain feature lines and feature points using a multi-element continental shelf toe line screening model. The specific steps are as follows:

[0075] (3.1) Index the neighboring water depth points of each water depth point, calculate the average slope of each water depth point, and calculate the total average slope of each continental shelf toe line in the candidate line set. Only those with a total average slope exceeding a certain threshold can be retained. This can, to some extent, reduce the influence of boundary points on toe line extraction in the digital depth model and help eliminate the influence of some rugged small terrains on the experimental results. As shown in formula (6).

[0076]

[0077] In the formula, Slope(i,j) represents the water depth point P. i =(x i ,y i ,z i Relative to its neighboring point P j =(x j ,y j ,z j The slope value of n i Let P be the water depth. i The number of neighboring points, T i Let P be the water depth. i Average slope, N k For connected component C K The number of mid-water depth points, S k For connected component C K The average slope value.

[0078] (3.2) Since the extraction quality depends to some extent on the spatial scale threshold setting of the negative topographic feature line skeleton group and key group, it is inevitable that the continental shelf slope toe line will "converge" towards the continental slope due to the wide threshold setting in the final extraction result. Therefore, under the premise of using the total average slope method to screen negative topographic feature lines, it should be ensured that the average slope of each feature point within the setting does not exceed the topographic slope range of the continental base slope, so as to further eliminate feature line segments that tend to the continental shelf slope area and achieve a second screening.

[0079] (3.3) Based on the slope screening, analyze the seabed topography distribution of DDM, set the range of continental base slope water depth values, and the average water depth value of the continental shelf toe line or the water depth value of each continental shelf toe point should be concentrated within the water depth threshold range, thereby filtering out some unreasonable negative topographic feature points and realizing the third screening.

[0080] (3.4) Based on slope and water depth screening, analyze the slope break line extraction, set restrictions on the number of nodes for each feature line in the "key group" and "skeleton group", eliminate small terrain interference segments, and achieve four screenings. After multiple experiments, it is recommended that each feature line in the key group must have more than or equal to 3 nodes, and each feature line in the skeleton group must have more than or equal to 10 nodes.

[0081] (3.5) Using the line segment processing tool in MATLAB software, and referring to the relevant provisions of Article 76 of the United Nations Convention on the Law of the Sea, and considering the aesthetics of the continental shelf toe line identification results, the "small branch" line segments in the continental shelf toe line are deleted, while the "main line segments" are retained, thus completing the fifth screening step in the multi-element continental shelf toe line screening process. A schematic diagram is shown below.Figure 5 .

[0082] This invention, starting from the current need for delineating the outer limits of the continental shelf, and referencing the legal definition of continental shelf toe points in Article 76 of the United Nations Convention on the Law of the Sea, proposes a method for accurately extracting continental shelf toe lines while considering topographic continuity. Addressing the current problems in the accurate identification of continental shelf toe points, it proposes a new approach directly addressing the accurate extraction of continental shelf toe lines based on three-dimensional seafloor topography. The basic idea is to build upon the accurate identification performance of rolling sphere transformation for negative seafloor topographic feature lines, and obtain two typical experimental groups by continuously adjusting the spatial scale threshold during feature line extraction, named the "key group" and the "skeleton group," respectively. Based on this, the concept of connected components from graph theory is introduced to construct an importance ranking model for the negative topographic feature line skeleton group, and further integrates a multi-element continental shelf toe line screening process that considers FOS geographical features, ultimately achieving a continental shelf toe line extraction method that is both accurate and efficient while considering topographic continuity.

Claims

1. A method for accurately extracting the toe line of a continental shelf while taking into account topographical continuity, characterized in that, The specific steps are as follows: Step (1) Process the original DDM data through terrain recognition and negative terrain feature line extraction model, identify the rolling ball terrain and extract negative terrain feature lines; Step (2) introduces connected component calculation, classifies the importance level of negative terrain feature lines through the continental shelf toe line candidate line set level evaluation model, and selects the continental shelf toe lines; Step (3) further eliminates unreasonable negative terrain feature lines and feature points by using a multi-element continental shelf toe line screening model.

2. The method for accurately extracting the continental shelf slope toe line while considering topographical continuity according to claim 1, characterized in that, Step (1) is as follows: (1.1) Read the water depth point information in the digital depth model (DDM), and record the positive critical rolling ball radius and negative critical rolling ball radius for each water depth point based on the rolling ball transformation method. In mathematical form, the rolling ball transformation is equivalent to the combination of the positive and negative buffer surface transformations of the DDM, that is, construct the upper and lower buffer surfaces of the DDM, then take the upper and lower buffer surface points as the center of the rolling ball, and then construct equidistant spheres in sequence. The physical rolling process of the rolling ball transformation is realized by mathematical methods. By continuously analyzing the numerical correlation between the rolling ball contact point and the rolling ball radius, the contact state between each sampling point and the dynamic rolling ball at different radii is determined, and then the positive critical rolling ball radius and negative critical rolling ball radius for each water depth point are obtained. (1.2) Reorder the water depth points based on the positive critical rolling sphere radius of each water depth point; construct the Delaunay triangulation of the original water depth data and record the natural neighborhood points of each water depth point; (1.3) Based on the sorting information of the positive critical rolling sphere radius of the water depth point and the information of natural neighborhood points, the center of the topographic subdivision unit is determined and the subdivision is continuously clustered until the seabed topographic region is completely subdivided; the center of the subdivision unit is the extreme point of the positive rolling sphere radius, and the boundary point of the subdivision unit is the negative topographic feature point. (1.4) Extract the boundary lines of the subdivided units, which are all the negative terrain feature lines in the DDM. The nodes of each feature line segment in the negative terrain feature lines are the negative terrain feature points. (1.5) For the extracted negative terrain feature lines, a spatial scale threshold is set. Each node on the feature line needs to determine the terrain type based on its critical rolling ball radius, thereby filtering out terrain feature points with corresponding geographical features; as shown in formula (1). The spatial scale threshold is shown in equation (1): In the formula: R is the spatial scale threshold, r' is the positive critical rolling sphere radius, and r” is the negative critical rolling sphere radius; Q(P i ) represents the water depth point P i The corresponding terrain type attribute, Q(P) i Q(P) represents positive terrain, where 1 represents positive terrain. i -1 represents negative terrain, Q(P) i Q(P) = 0 represents flat terrain. i A value of -2 represents a nested negative terrain within a positive terrain, Q(P) i ) 2 represents a positive terrain nested within a negative terrain; The negative terrain feature line segments that meet the requirements are selected from all negative terrain feature lines by using the judgment conditions of formula (1). Multiple extraction experiments were conducted by successively lowering the spatial scale threshold. The negative terrain feature lines of the "key group" with a lower spatial scale threshold were selected, and the negative terrain feature lines of the "skeleton group" with a higher spatial scale threshold were selected. The two experimental groups with typical feature line distribution characteristics were retained.

3. The method for accurately extracting the continental shelf slope toe line while considering topographical continuity, as described in claim 2, is characterized in that... The "key group" of negative terrain feature lines refers to line segments whose number is controlled within 10% of the total number of negative terrain feature line segments, and the "skeleton group" of negative terrain feature lines refers to line segments whose number is controlled within 90% of the total number of negative terrain feature line segments.

4. The method for accurately extracting the continental shelf slope toe line while considering topographical continuity according to claim 1, characterized in that, Step (2) is as follows: (2.1) Depth-First Search (DFS) introduces the concept of connected components from graph theory. It uses the DFS function to construct the connected components of the skeleton group's feature lines and indexes the depth sampling points within these connected components, denoted as the connected array. The steps include: (2.1.1) Record the skeleton group feature lines generated based on the rolling ball transformation as S, and generate the key group feature lines as I, as shown in formula (2); In the formula, and Represents the two endpoints of the i-th negative terrain feature line segment. and Represents the two endpoints of the j-th negative terrain feature line segment; (2.1.2) Let each edge (u,v)∈S, indicating that a feature line segment can be formed between feature point u and feature point v; construct the undirected graph G of the skeleton group feature line S as follows, as shown in formula (3); each line segment is composed of a pair of feature points, and E contains all feature line segments; In the formula, V is the set of points in the undirected graph, containing all feature points, and E (or S) is the set of lines in the undirected graph; (2.1.3) Traverse each edge (u,v)∈S and add feature points u and v to the adjacency list of graph G. If feature point u or feature point v is not in the adjacency list of graph G, create a new adjacency list. The adjacency list Adj(u) formed with the feature point u is shown in formula (4); Adj(u)={v|(u,v)∈S} (4) In the formula: Adj(u) is This forms an adjacency list with the feature point u; (2.1.4) Use the DFS function to traverse all feature point adjacency lists until all feature points connected to the given feature point have been visited; at this point, the skeleton feature lines are clustered into k connected components C, C = {C1, C2, ... C2}. K }; The feature points within each connected component are interconnected, forming a feature line; (2.2) Calculate the number of intersection points between each connected component in the feature lines of the "skeleton group" and all feature points in the "key group"; let C n Let I be the nth connected component in the skeleton group feature line S, where 1 ≤ n ≤ k. Given that I is a key group feature line containing all slope toe candidate points, then the number of key group feature points m in the nth connected component is... n Represented as: m n =|C n ∩I|; (2.3) Further calculate the proportion of key group negative terrain feature points contained in each connected component of the skeleton group negative terrain feature line under the large spatial scale threshold, and then set the C of each connected component according to the intersection ratio of each connected component. n Importance level R n As shown in formula (5); In the formula, R n t represents the importance level of the connected components. n m represents the number of terrain feature points contained in a connected component. n Indicates the number of feature points in the key group; (2.4) Based on the importance level classification, further filter out the continental shelf slope toe lines that take into account the topographic continuity and have a higher importance level.

5. The method for accurately extracting the continental shelf slope toe line while taking into account topographical continuity, as described in claim 1, is characterized in that... The specific steps of step (3) are as follows: (3.1) Index the neighboring water depths of each water depth point, calculate the average slope of each water depth point, and calculate the total average slope of each continental shelf toe line in the continental shelf toe line candidate line set. If the total average slope exceeds the threshold, it is retained. As shown in formula (6); In the formula, Slope(i,j) represents the water depth point P. i =(x i ,y i ,z i Relative to its neighboring point P j =(x j ,y j ,z j The slope value of n i Let P be the water depth. i The number of neighboring points, T i Let P be the water depth. i Average slope, N k For connected component C K The number of mid-water depth points, S k For connected component C K The average slope value; (3.2) Under the premise of using the total average slope method to screen negative topographic feature lines, it is necessary to ensure that the average slope of each feature point within the set does not exceed the topographic slope range of the continental base slope, so as to further eliminate feature line segments that tend to the continental shelf slope area and achieve a second screening. (3.3) Based on the slope screening, analyze the seabed topography distribution of DDM, set the range of continental base slope water depth values, and the average water depth value of the continental shelf toe line or the water depth value of each continental shelf toe point should be concentrated within the water depth threshold range, thereby filtering out some unreasonable negative topographic feature points and achieving the third screening. (3.4) Based on the slope screening and water depth screening, analyze the slope broken line extraction situation, set the limit conditions for the number of nodes of each feature line in the "key group" and "skeleton group", eliminate the interference line segments of small terrain, and achieve four screenings; after multiple experiments, it is recommended that each feature line in the key group must have more than or equal to 3 nodes, and each feature line in the skeleton group must have more than or equal to 10 nodes. (3.5) Using the line segment processing tool in MATLAB software, considering the aesthetics of the continental shelf toe line identification results, delete the "small branch" line segments in the continental shelf toe line and retain the "main line segments" in the continental shelf toe line, thus completing the fifth screening of the multi-element continental shelf toe line screening process.

6. The method for accurately extracting the continental shelf slope toe line while taking into account topographic continuity, as described in claim 1, is characterized in that... In step (3.4), after multiple experiments, each feature line in the key group has more than or equal to 3 nodes, and each feature line in the skeleton group has more than or equal to 10 nodes.

Images