Channel boundary and centerline mining method and system for AIS ship trajectory big data

Abstract

The application provides a waterway boundary and center line mining method and system for AIS ship trajectory big data, and comprises the following steps: S1, carrying out Delaunay triangulation on the track points of the AIS ship trajectory data, and screening the triangles in the track point triangulation to obtain the original boundary of the track points; S2, carrying out smoothing processing on the obtained original boundary of the track points, further carrying out Delaunay triangulation on the vertex point set of the original boundary of the track points after the smoothing processing, and screening the vertex point set triangulation to obtain the waterway contour; and S3, constructing the waterway graph and the center line of the waterway based on the waterway contour.

Description

Technical Field

[0001] This invention relates to the field of information technology, specifically to a method and system for mining channel boundaries and centerlines based on AIS ship trajectory big data, and more specifically, to a channel boundary and centerline mining technology based on the shape reconstruction of track point sets for AIS ship trajectory big data. Background Technology

[0002] Land traffic network identification based on image recognition is a mature technology, but waterway area identification based on marine imagery is difficult to achieve using similar methods.

[0003] Currently, maritime channel identification mainly relies on manual measurement and high-resolution remote sensing imagery. Its main drawbacks are the high cost of manual measurement and remote sensing equipment, as well as poor real-time performance. Automatic Identification Systems (AIS) provide abundant crowdsourced vessel trajectory data, offering advantages such as low cost and high update frequency. Compared to channel identification technologies relying on manual measurement or high-resolution remote sensing imagery, maritime channel mining technologies based on AIS vessel trajectory big data offer significantly greater usability.

[0004] Current channel mining techniques for AIS trajectory big data are mainly based on spatial density noise clustering, using the tracks at the center and edge of the trajectory clusters as approximations of the channel centerline and channel boundary, respectively. The effectiveness of this type of mining method is quite sensitive to the parameter selection of the clustering algorithm and is easily affected by noise points in the cluster center and edge tracks, making it difficult to accurately characterize the geometric features of the actual channel.

[0005] Patent document CN112487116A (application number: 202011298608.9) discloses a ship track analysis method based on AIS big data. The data cleaning submodule corrects obviously abnormal data. The single channel stop point submodule calculates the single channel stop point based on the cleaned data. The multi-channel stop point clustering submodule clusters based on the single channel stop point to obtain multiple channel stop points, ensuring the accuracy of the original channel analysis data. After the data is compressed by the Douglas-Peucker algorithm, the LCSS algorithm is used to find the similar parts of two trajectories. At the same time, the similarity between tracks is calculated by Dynamic Time Warping (DTW). The DBScan algorithm is used to cluster similar tracks. Track mining is carried out by combining the graphic outlining algorithm and the LCSS similarity algorithm with actual business scenarios. The integrated use of multiple algorithms improves the speed and accuracy of track mining. Summary of the Invention

[0006] To address the shortcomings of existing technologies, the purpose of this invention is to provide a method and system for mining channel boundaries and centerlines based on AIS ship trajectory big data.

[0007] A method for mining channel boundaries and centerlines based on AIS ship trajectory big data, provided by the present invention, includes:

[0008] Step S1: Perform Delaunay triangulation on the track points of the AIS ship trajectory data, and filter the original boundaries of the track points based on the triangles in the track point triangulation.

[0009] Step S2: Smooth the original boundaries of the acquired track points, and further perform Delaunay triangulation on the vertex set of the original boundary of the track points after smoothing. The channel outline is obtained by filtering the triangulation of the vertex set.

[0010] Step S3: Construct a waterway map and the centerline of the waterway based on the waterway contour.

[0011] Preferably, step S1 employs:

[0012] Step S1.1: Preprocess the original AIS ship track big data, and uniformly sample the preprocessed AIS ship track big data at fixed time intervals;

[0013] Step S1.2: Convert latitude and longitude coordinates into geohash values;

[0014] Step S1.3: Based on the frequency distribution of geohash values, use the maximum inter-class variance algorithm to obtain the frequency screening threshold of hash values, retain the coordinates corresponding to geohash values ​​whose frequency is higher than the screening threshold, and use them as approximate track boundary candidate points. Map the approximate track boundary candidate points to Cartesian plane coordinates to obtain the planar point set of track points.

[0015] Step S1.4: Use Delaunay triangulation to reconstruct the shape of the planar track point set, and perform triangular void index filtering on the triangles in the obtained track point triangulation. Extract the external boundary from the remaining triangulation to obtain the original boundary of the track points.

[0016] Preferably, the preprocessing of the original AIS ship trajectory big data involves: using geohashing technology to grid the ship trajectory coordinates of the original AIS ship trajectory big data to obtain gridded AIS ship trajectory big data.

[0017] Preferably, step S2 employs:

[0018] Step S2.1: Smooth the original boundaries of the track points to obtain refined channel polygon boundaries;

[0019] Step S2.2: Extract the vertex set of the channel boundary based on the refined channel polygon boundary. Further extract the Delaunay triangulation from the extracted channel boundary vertex set and remove the triangles outside the channel boundary to obtain the channel outline.

[0020] Preferably, step S2.1 involves smoothing the original boundary of the waypoint using burr removal and median filtering techniques to obtain a refined channel polygon boundary.

[0021] Preferably, step S2.2 involves: using Delaunay triangulation to obtain a new triangulation network from the extracted channel boundary vertex set; and using the centroid coordinates of the triangles to filter out the triangles inside the channel boundary from the new triangulation network.

[0022] The triangles inside the waterway include single-adjacent triangulation networks, double-adjacent triangulation networks, and triple-adjacent triangulation networks.

[0023] Preferably, step S3 involves: the vertices of the waterway map being formed by the centroids of specific triangles, and the edges of the waterway map being formed by the medians of the other triangles on the private edges; the waterway centerline is an edge in the extracted waterway map.

[0024] The specific triangle includes a single-adjacent triangular network or a three-adjacent triangular network;

[0025] The channel map extraction algorithm starts from any single neighboring triangle and performs a depth-first traversal of the channel triangulation. A channel edge is created from the direction of the common edge of the single neighboring triangles. The initial line segment of the edge is the line segment from the channel vertex to the midpoint of the common edge.

[0026] Different operations are performed depending on the type of triangle encountered: if it is an edge triangle, the median line on its private edge is extracted and added to the current path edge; if it is a vertex triangle, the current path edge is closed and a new path edge is created in the direction of other shared edges; until all triangles have been traversed.

[0027] The side triangle is a triangle whose number of adjacent triangles is even;

[0028] The vertex triangle is a triangle whose number of adjacent triangles is odd.

[0029] A system for mining channel boundaries and centerlines based on AIS ship trajectory big data, provided by the present invention, includes:

[0030] Module M1: Performs Delaunay triangulation on the track points of AIS ship trajectory data, and obtains the original boundaries of the track points by filtering the triangles in the track point triangulation.

[0031] Module M2: Smooths the original boundaries of the acquired track points, and further performs Delaunay triangulation on the vertex set of the original boundary of the track points after smoothing. The channel outline is obtained by filtering the triangulation of the vertex set.

[0032] Module M3: Constructs a waterway map and the centerline of the waterway based on the waterway outline.

[0033] Preferably, module M1 adopts:

[0034] Module M1.1: Preprocesses the original AIS ship track big data and uniformly samples the preprocessed AIS ship track big data at fixed time intervals;

[0035] Module M1.2: Converts latitude and longitude coordinates into geohash values;

[0036] Module M1.3: Based on the frequency distribution of geohash values, the algorithm of maximizing inter-class variance is used to obtain the frequency screening threshold of hash values. The coordinates corresponding to geohash values ​​with a frequency higher than the screening threshold are retained and used as approximate track boundary candidate points. The approximate track boundary candidate points are mapped to Cartesian plane coordinates to obtain the planar point set of track points.

[0037] Module M1.4: Uses Delaunay triangulation to reconstruct the shape of the planar track point set, and performs triangular void index filtering on the triangles in the obtained track point triangulation. Extracts the external boundary from the remaining triangulation to obtain the original boundary of the track points.

[0038] The preprocessing of the original AIS ship trajectory big data involves: using geohashing technology to grid the ship trajectory coordinates of the original AIS ship trajectory big data to obtain gridded AIS ship trajectory big data.

[0039] The module M2 adopts:

[0040] Module M2.1: Smooths the original boundaries of the waypoints to obtain refined channel polygon boundaries;

[0041] Module M2.2: Extracts the vertex set of the channel boundary based on the refined channel polygon boundary. The extracted vertex set of the channel boundary is further used to extract the Delaunay triangulation and remove the triangles outside the channel boundary to obtain the channel outline.

[0042] The module M2.1 employs the following technique: smoothing the original boundary of the track point using burr removal and median filtering to obtain a refined channel polygon boundary.

[0043] The module M2.2 employs the following method: the extracted channel boundary vertex set is used to obtain a new triangulation network using Delaunay triangulation technology, and the new triangulation network is used to filter out triangles inside the channel boundary by using the centroid coordinates of the triangles.

[0044] The triangles inside the waterway include single-adjacent triangulation networks, double-adjacent triangulation networks, and triple-adjacent triangulation networks.

[0045] Preferably, module M3 adopts the following configuration: the vertices of the waterway chart are formed by the centroids of specific triangles, and the edges of the waterway chart are formed by the medians of the other triangles on the private edges; the waterway centerline is an edge in the extracted waterway chart.

[0046] The specific triangle includes a single-adjacent triangular network or a three-adjacent triangular network;

[0047] The channel map extraction algorithm starts from any single neighboring triangle and performs a depth-first traversal of the channel triangulation. A channel edge is created from the direction of the common edge of the single neighboring triangles. The initial line segment of the edge is the line segment from the channel vertex to the midpoint of the common edge.

[0048] Different operations are performed depending on the type of triangle encountered: if it is an edge triangle, the median line on its private edge is extracted and added to the current path edge; if it is a vertex triangle, the current path edge is closed and a new path edge is created in the direction of other shared edges; until all triangles have been traversed.

[0049] The side triangle is a triangle whose number of adjacent triangles is even;

[0050] The vertex triangle is a triangle whose number of adjacent triangles is odd.

[0051] Compared with the prior art, the present invention has the following beneficial effects:

[0052] 1. This invention utilizes AIS ship trajectory big data for channel mining, which is cheaper, more efficient, and more usable than traditional channel identification methods based on manual mapping or high-resolution remote sensing images.

[0053] 2. This invention utilizes track point contour extraction to mine waterway boundaries and centerlines. Compared with methods based on spatial density noise clustering, the waterway boundaries and centerlines mined by this invention are smoother and their geometric features are closer to the real natural water environment. This provides an effective method and reference for studying water traffic patterns, water environment analysis, and waterway width estimation.

[0054] 3. This invention uses geohashing technology to perform grid-based encoding on the original trajectory coordinates, which can effectively suppress noise in the original coordinate data and make the frequency term mining algorithm more robust.

[0055] 4. The spatial granularity of the waterway area in this invention benefits from the dynamic nature of geohashing technology, making it scalable (compared to traditional manual grid coding). That is, the granularity of the waterway area can be made finer or coarser by increasing or decreasing the length of the geohashing values ​​according to actual usage needs.

[0056] 5. The inter-class variance maximization algorithm used in this invention for frequency filtering of geohash values ​​is a data-driven method. It effectively avoids the difficulty in selecting filtering thresholds caused by inconsistent AIS data sampling density in different environments such as nearshore and offshore areas.

[0057] 5. This invention performs post-processing work such as removing short sides and burrs on the initially extracted channel boundary to avoid phenomena such as the excavated channel boundary being too narrow or dangerous and not conforming to natural conditions.

[0058] 6. The channel boundaries and centerlines extracted by this invention have consistency. The results of this invention can be directly used to further mine channel geometric features, such as calculating the navigable width of the channel (see...). Figure 8 The height of the private edge of the triangle in the triangular mesh can be directly extracted. Attached Figure Description

[0059] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0060] Figure 1 This is a schematic diagram of the technical process of the present invention.

[0061] Figure 2 This is a schematic diagram of the triangular void index.

[0062] Figure 3 This is a schematic diagram of the original triangulation network in the waterway area.

[0063] Figure 4 A schematic diagram of a triangular mesh for filtering void index.

[0064] Figure 5 This is a diagram showing the smoothing effect of the channel boundary.

[0065] Figure 6 This is a schematic diagram of the triangular classification of waterways.

[0066] Figure 7 A schematic diagram for constructing a waterway chart.

[0067] Figure 8 This is a rendering of the excavation process for the boundary and centerline of the Bohai Sea waterway. Detailed Implementation

[0068] The present invention will now be described in detail with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the present invention, but do not limit the invention in any way. It should be noted that those skilled in the art can make several changes and improvements without departing from the concept of the present invention. These all fall within the protection scope of the present invention.

[0069] To address the shortcomings of existing technologies, this invention provides a channel boundary and centerline mining technique based on the shape reconstruction of track point sets for AIS ship trajectory big data. The aim is to improve the accuracy and efficiency of channel boundary and centerline mining and identification, as well as reduce the cost of channel identification.

[0070] This invention discloses a channel boundary and centerline mining technique for AIS ship trajectory big data. This technique performs Delaunay triangulation on the track points of AIS ship trajectory data, and extracts the triangulation contour to mine the polygonal boundaries and channel centerlines of the waterway navigation area.

[0071] Example 1

[0072] A method for mining channel boundaries and centerlines based on AIS ship trajectory big data, provided by the present invention, includes:

[0073] Step S1: Perform Delaunay triangulation on the track points of the AIS ship trajectory data, and filter the original boundaries of the track points based on the triangles in the track point triangulation.

[0074] Step S2: Smooth the original boundaries of the acquired track points, and further perform Delaunay triangulation on the vertex set of the original boundary of the track points after smoothing. The channel outline is obtained by filtering the triangulation of the vertex set.

[0075] Step S3: Construct a waterway map and the centerline of the waterway based on the waterway contour.

[0076] Specifically, step S1 employs the following:

[0077] Step S1.1: Preprocess the original AIS ship track big data, and uniformly sample the preprocessed AIS ship track big data at fixed time intervals;

[0078] Step S1.2: Convert latitude and longitude coordinates into geohash values;

[0079] Step S1.3: Based on the frequency distribution of geohash values, the frequency filtering threshold is obtained using the maximum inter-class variance algorithm. The coordinates corresponding to geohash values ​​with a frequency higher than the filtering threshold are retained to suppress data noise and serve as approximate candidate points for the track boundary. These approximate candidate points are then mapped to Cartesian plane coordinates to obtain a set of planar points for the track points. The maximum inter-class variance algorithm is a commonly used method in grayscale image binarization. It achieves image binarization by traversing the binarization threshold and maximizing the variance between the target and background regions. This invention utilizes the maximum inter-class variance algorithm to maximize the frequency variance of the eliminated and retained geohash values, thus achieving the purpose of geohash value filtering.

[0080] Step S1.4: Reconstruct the shape of the planar track point set using Delaunay triangulation, and perform triangular hole index filtering on the triangles in the obtained track point triangulation. Extract the external boundary from the remaining triangulation to obtain the original boundary of the track points. Delaunay triangulation is a graphical analysis technique that transforms a planar discrete point set into triangular mesh data. Delaunay triangulation has two characteristics: maximizing the minimum angle to make the triangulation as close to regular as possible, and the uniqueness that no four points can be concyclic. The triangulation algorithm constructs a convex hull on the planar point set and continuously connects adjacent convex hull edges to form triangles to build a triangulation until no usable edges remain.

[0081] Specifically, the preprocessing of the original AIS ship trajectory big data involves: using geohashing technology to grid the ship trajectory coordinates, resulting in gridded AIS ship trajectory big data. Geohashing is a process of encoding two-dimensional coordinates into a string. It uses a binary search method to narrow down the longitude and latitude intervals for binary encoding, then merges the odd and even bits of the longitude and latitude codes, finally converting the binary string into a base32 string.

[0082] Specifically, step S2 employs the following:

[0083] Step S2.1: Smooth the original boundaries of the track points to obtain refined channel polygon boundaries;

[0084] Step S2.2: Extract the vertex set of the channel boundary based on the refined channel polygon boundary. Further extract the Delaunay triangulation from the extracted channel boundary vertex set and remove the triangles outside the channel boundary to obtain the channel outline.

[0085] Specifically, step S2.1 involves smoothing the original boundary of the waypoint using burr removal and median filtering techniques to obtain a refined wayway polygon boundary.

[0086] Specifically, step S2.2 involves: using Delaunay triangulation to obtain a new triangulation network from the extracted set of vertex points on the channel boundary; and using the centroid coordinates of the triangles to filter out the triangles inside the channel boundary from the new triangulation network.

[0087] The triangles inside the waterway include single-adjacent triangulation networks, double-adjacent triangulation networks, and triple-adjacent triangulation networks.

[0088] Specifically, step S3 employs the following method: the vertices of the waterway map are formed by the centroids of specific triangles, and the edges of the waterway map are formed by the medians of the other triangles on the private edges; the waterway centerline is the edge in the extracted waterway map.

[0089] The specific triangle includes a single-adjacent triangular network or a three-adjacent triangular network;

[0090] The overall framework of the waterway map construction algorithm is described in Algorithm 1:

[0091]

[0092]

[0093] The channel map extraction algorithm starts with any single neighbor triangle and performs a depth-first traversal of the vertex triangles of the channel triangulation (Algorithm 1, lines 1-4);

[0094] The graph vertex triangle traversal algorithm recursively traverses all vertex triangles in the graph, as described in Algorithm 2:

[0095]

[0096] Create a channel edge from the direction of the shared edge of the adjacent triangles (Algorithm 2, lines 4-5). The initial line segment of the edge is the line segment from the channel vertex to the midpoint of the shared edge.

[0097] Recursively extract the edges and vertices of the current channel edge, as described in Algorithm 3:

[0098]

[0099]

[0100] Different operations are performed based on the type of triangle encountered: if it is an edge triangle (Algorithm 3, lines 4-7), the median line on its private edge is extracted and added to the current channel edge; if it is a vertex triangle (Algorithm 3, lines 1-3), the current channel edge is closed and a new channel edge is created in the direction of other shared edges; until all triangles have been traversed.

[0101] The side triangle is a triangle whose number of adjacent triangles is even;

[0102] The vertex triangle is a triangle whose number of adjacent triangles is odd.

[0103] A system for mining channel boundaries and centerlines based on AIS ship trajectory big data, provided by the present invention, includes:

[0104] Module M1: Performs Delaunay triangulation on the track points of AIS ship trajectory data, and obtains the original boundaries of the track points by filtering the triangles in the track point triangulation.

[0105] Module M2: Smooths the original boundaries of the acquired track points, and further performs Delaunay triangulation on the vertex set of the original boundary of the track points after smoothing. The channel outline is obtained by filtering the triangulation of the vertex set.

[0106] Module M3: Constructs a waterway map and the centerline of the waterway based on the waterway outline.

[0107] Specifically, module M1 adopts:

[0108] Module M1.1: Preprocesses the original AIS ship track big data and uniformly samples the preprocessed AIS ship track big data at fixed time intervals;

[0109] Module M1.2: Converts latitude and longitude coordinates into geohash values;

[0110] Module M1.3: Based on the frequency distribution of geohash values, the frequency filtering threshold is obtained using the maximum inter-class variance algorithm. The coordinates corresponding to geohash values ​​with a frequency higher than the filtering threshold are retained, which helps suppress data noise and serves as approximate candidate points for track boundaries. These approximate candidate points are then mapped to Cartesian plane coordinates to obtain a set of planar points for the track points. The maximum inter-class variance algorithm is a commonly used method in grayscale image binarization. It achieves image binarization by traversing the binarization threshold and maximizing the variance between the target and background regions. This invention utilizes the maximum inter-class variance algorithm to maximize the frequency variance of the eliminated and retained geohash values, thus achieving the purpose of geohash value filtering.

[0111] Module M1.4: Reconstructs the shape of a planar track point set using Delaunay triangulation, and filters the triangles in the resulting track point triangulation using the triangular void index. The outer boundaries are then extracted from the remaining triangulation to obtain the original boundaries of the track points. Delaunay triangulation is a graphical analysis technique that transforms a planar discrete point set into triangular mesh data. Delaunay triangulation has two key characteristics: maximizing the minimum angle to achieve the closest possible regularity of the triangulation, and the uniqueness that no four points can be concyclic. The triangulation algorithm constructs a convex hull on the planar point set and continuously connects adjacent convex hull edges to form triangles, building the triangulation until no more edges are available.

[0112] Specifically, the preprocessing of the original AIS ship trajectory big data involves: using geohashing technology to grid the ship trajectory coordinates, resulting in gridded AIS ship trajectory big data. Geohashing is a process of encoding two-dimensional coordinates into a string. It uses a binary search method to narrow down the longitude and latitude intervals for binary encoding, then merges the odd and even bits of the longitude and latitude codes, finally converting the binary string into a base32 string.

[0113] Specifically, module M2 adopts:

[0114] Module M2.1: Smooths the original boundaries of the waypoints to obtain refined channel polygon boundaries;

[0115] Module M2.2: Based on the refined channel polygon boundary, extract the channel boundary vertex set. The extracted channel boundary vertex set is further used to extract the Delaunay triangulation and remove the triangles outside the channel boundary to obtain the channel outline.

[0116] Specifically, module M2.1 employs a technique of removing burrs and using median filtering to smooth the original boundaries of the waypoints, resulting in refined channel polygon boundaries.

[0117] Specifically, module M2.2 employs the following method: the extracted set of vertex points on the channel boundary is used to obtain a new triangulation network using Delaunay triangulation technology, and the new triangulation network is used to filter out triangles inside the channel boundary by using the centroid coordinates of the triangles;

[0118] The triangles inside the waterway include single-adjacent triangulation networks, double-adjacent triangulation networks, and triple-adjacent triangulation networks.

[0119] Specifically, module M3 adopts the following approach: the vertices of the waterway chart are formed by the centroids of specific triangles, and the edges of the waterway chart are formed by the medians of the other triangles on the private edges; the waterway centerline is the edge in the extracted waterway chart.

[0120] The specific triangle includes a single-adjacent triangular network or a three-adjacent triangular network;

[0121] The overall framework of the waterway map construction algorithm is described in Algorithm 1:

[0122]

[0123]

[0124] The channel map extraction algorithm starts with any single neighbor triangle and performs a depth-first traversal of the vertex triangles of the channel triangulation (Algorithm 1, lines 1-4);

[0125] The graph vertex triangle traversal algorithm recursively traverses all vertex triangles in the graph, as described in Algorithm 2:

[0126]

[0127] Create a channel edge from the direction of the shared edge of the adjacent triangles (Algorithm 2, lines 4-5). The initial line segment of the edge is the line segment from the channel vertex to the midpoint of the shared edge.

[0128] Recursively extract the edges and vertices of the current channel edge, as described in Algorithm 3:

[0129]

[0130]

[0131] Different operations are performed based on the type of triangle encountered: if it is an edge triangle (Algorithm 3, lines 4-7), the median line on its private edge is extracted and added to the current channel edge; if it is a vertex triangle (Algorithm 3, lines 1-3), the current channel edge is closed and a new channel edge is created in the direction of other shared edges; until all triangles have been traversed.

[0132] The side triangle is a triangle whose number of adjacent triangles is even;

[0133] The vertex triangle is a triangle whose number of adjacent triangles is odd.

[0134] Example 2

[0135] Example 2 is a preferred example of Example 1.

[0136] Please refer to the overall implementation process. Figure 1 This embodiment uses AIS trajectory big data from the Bohai Sea region of my country, spanning from March to August 2021, as an example to introduce the implementation process for excavating the channel boundaries and centerlines in this sea area:

[0137] Geohashing technology is used to grid the original AIS track coordinates. The geohashing sets of the tracks are merged, and the frequency of each geohashing is counted.

[0138] A hash value frequency filtering threshold is obtained using the maximum inter-class variance algorithm. The coordinates corresponding to hash values ​​with relatively high occurrence frequencies are retained as approximate candidate track boundary points. The selection result is based on the maximum inter-class variance principle. Gridded coordinates with high occurrence frequencies are retained.

[0139] The retained track latitude and longitude coordinates were mapped to the EPSG 3857 plane coordinate system. The planar point set was triangulated using the Delaunay triangulation technique, and the hole index of each triangle in the mesh was calculated. Please refer to the schematic diagram of the original triangulation of the track points. Figure 3 .like Figure 2 As shown, the hole index of a triangle is defined as the radius of its inscribed circle. Triangles with high hole indices are removed from the triangulation; in this example, a hole index of 2e3 is chosen. Please refer to the filtered triangulation network. Figure 4 .

[0140] The triangulation boundary is extracted as the original channel boundary. Techniques such as edge smoothing and median filtering are then used to smooth the original boundary. The smoothing effect of the channel boundary is shown below. Figure 5 As shown.

[0141] A new triangulation network is obtained by using Delaunay triangulation on the smoothed channel boundary point set. Triangles inside the channel boundary are then selected based on their centroid coordinates. The triangles in the new triangulation network are classified according to the number of their neighboring triangles. Triangles are thus divided into two categories: vertex triangles (odd number of neighboring triangles) and edge triangles (even number of neighboring triangles). The classification results in this embodiment can be found in [reference needed]. Figure 6 .

[0142] Extracting the channel graph from a triangulation network. The vertices of the channel graph are formed by the centroids of the vertex triangles, and the edges are formed by the medians of the edge triangles on their private edges. The channel graph extraction algorithm starts with any single neighboring triangle and performs a depth-first traversal of the channel triangulation network. A channel edge is created from the direction of the shared edge (which exists only once) of the single neighboring triangles. The initial segment of the edge is the segment from the channel vertex to the midpoint of the shared edge. Different operations are performed depending on the type of triangle encountered: if it is an edge triangle, the median on its private edge (which exists only once) is extracted and added to the current channel edge; if it is a vertex triangle, the current channel edge is closed, and a new channel edge is created in the direction of other shared edges. This process continues until all triangles have been traversed. For a schematic diagram of the channel graph result in this example, please refer to [link to example]. Figure 7 .

[0143] The edges extracted from the waterway map represent the centerlines of the waterways to be extracted. For examples of waterway boundaries and centerline excavation results in the Bohai Sea region, please refer to [link to relevant documentation]. Figure 8 .

[0144] Those skilled in the art will understand that, in addition to implementing the system, apparatus, and their modules provided by this invention in purely computer-readable program code, the same program can be implemented in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers by logically programming the method steps. Therefore, the system, apparatus, and their modules provided by this invention can be considered a hardware component, and the modules included therein for implementing various programs can also be considered structures within the hardware component; alternatively, modules for implementing various functions can be considered both software programs implementing the method and structures within the hardware component.

[0145] Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Unless otherwise specified, the embodiments and features described in this application can be arbitrarily combined with each other.

Claims

1. A fairway boundary and centerline mining method for AIS ship trajectory big data, characterized in that, The method comprises the following steps: Step S1: Delaunay triangulation is performed on the track points of AIS ship trajectory data, and the original boundary of the track points is obtained by screening the triangles in the track point triangulation; Step S2: The original boundary of the track points obtained is smoothed, the vertex point set of the original boundary of the track points after smoothing is further subjected to Delaunay triangulation, and the channel profile is obtained by screening the vertex point set triangulation; Step S3: The channel profile is used to construct a channel map and a channel center line; In step S3, the vertices of the channel map are composed of the barycenters of specific triangles, and the edges of the channel map are composed of the midlines of the private edges of the remaining triangles; the channel center line is the edge of the extracted channel map; The specific triangle includes a single-neighbor triangle mesh or a three-neighbor triangle mesh; According to the channel map extraction algorithm, a single-neighbor triangle is started, a depth-first traversal is performed on the vertex triangle of the channel triangle mesh, a channel edge is created from the single-neighbor triangle common edge direction, and the initial line segment of the edge is the line segment from the channel vertex to the midpoint of the common edge; Different operations are performed according to the categories of the traversed triangles: if it is an edge triangle, the midline of the private edge thereof is extracted and added to the current channel edge; if it is a vertex triangle, the current channel edge is closed, and a new channel edge is created in the direction of the other common edge; until all triangles are traversed; The edge triangle is a triangle with an even number of adjacent triangles; The vertex triangle is a triangle with an odd number of adjacent triangles. 2.The AIS ship trajectory big data oriented waterway boundary and centerline mining method according to claim 1, characterized in that, In step S1, the following steps are performed: Step S1.1: The original AIS ship trajectory big data is preprocessed, and the preprocessed AIS ship trajectory big data is uniformly sampled at a fixed time interval; Step S1.2: The latitude and longitude coordinates are converted into geographic hash values; Step S1.3: The hash value frequency screening threshold is obtained by using the maximum inter-class variance algorithm according to the frequency distribution of the geographic hash values, the coordinates corresponding to the geographic hash values with a frequency higher than the screening threshold are retained, and the coordinates are used as approximate track boundary candidate points; the approximate track boundary candidate points are mapped to Cartesian plane coordinates to obtain a planar point set of the track points; Step S1.4: The shape of the planar track point set is reconstructed by using the Delaunay triangulation technique, the triangles in the obtained track point triangulation are screened according to the triangle hole index, and the external boundary is extracted from the remaining triangulation to obtain the original boundary of the track points. 3.The method of claim 2, wherein, In the preprocessing of the original AIS ship trajectory big data, the ship trajectory coordinates are gridded by using the geographic hash technology to obtain the gridded AIS ship trajectory big data. 4.The AIS ship trajectory big data oriented waterway boundary and centerline mining method according to claim 1, characterized in that, In step S2, the following steps are performed: Step S2.1: The original boundary of the track points is smoothed to obtain a refined channel polygon boundary; Step S2.2: The channel boundary vertex point set is extracted based on the refined channel polygon boundary, the extracted channel boundary vertex point set is further subjected to Delaunay triangulation, and the triangles outside the channel boundary are removed to obtain a channel profile.

5. The method of claim 4, wherein, The step S2.1 adopts: using burr edge removal and median filtering technology to smooth the original boundary of the track point to obtain the refined channel polygon boundary. 6.The method of claim 4, wherein, The step S2.2 adopts: the extracted channel boundary vertex point set uses the Delaunay triangulation technology to obtain a new triangular net, and the new triangular net is screened through the triangle barycentric coordinates to obtain the triangle inside the channel boundary. The triangle inside the channel includes single-adjacent triangular net, double-adjacent triangular net and triple-adjacent triangular net.

7. A waterway boundary and centerline mining system for AIS ship trajectory big data, characterized in that, It comprises: Module M1: Delaunay triangulation of AIS ship trajectory data track points, filtering based on triangle in track point triangulation to obtain original boundary of track points; Module M2: smoothing the obtained original boundary of track points, further Delaunay triangulation based on the vertex point set of the smoothed original boundary of track points, filtering based on the vertex point set triangulation to obtain the channel profile; Module M3: constructing channel map and channel center line based on channel profile; The module M3 adopts: the vertex of the channel map is composed of the barycenter of a specific triangle, and the edge of the channel map is composed of the midline of the private edge of the remaining triangle; The channel center line is the edge extracted from the channel map; The specific triangle includes single-adjacent triangular net or triple-adjacent triangular net; Through the channel map extraction algorithm, starting from any single-adjacent triangle, the vertex triangle of the channel triangulation is traversed in depth-first manner, and a channel edge is created from the direction of the common edge of the single-adjacent triangle. The initial line segment of the edge is the line segment from the channel vertex to the midpoint of the common edge. According to the category of the triangle traversed, different operations are performed: if it is an edge triangle, the midline of the private edge is extracted and added to the current channel edge; if it is a vertex triangle, the current channel edge is closed, and a new channel edge is created in the direction of the other common edge; until all triangles are traversed. The edge triangle is a triangle with an even number of adjacent triangles; The vertex triangle is a triangle with an odd number of adjacent triangles.

8. The waterway boundary and centerline mining system for AIS ship trajectory big data of claim 7, wherein, The module M1 adopts: Module M1.1: preprocessing the original AIS ship trajectory big data, and uniformly sampling the preprocessed AIS ship trajectory big data at fixed time intervals; Module M1.2: converting latitude and longitude coordinates into geographic hash values; Module M1.3: obtaining a hash value frequency screening threshold using the maximum inter-class variance algorithm according to the frequency distribution of geographic hash values, retaining the coordinates corresponding to the geographic hash values with a frequency higher than the screening threshold, and mapping the approximate track boundary candidate points to Cartesian plane coordinates to obtain a plane point set of track points; Module M1.4: using Delaunay triangulation technology to reconstruct the shape of the plane track point set, and screening the triangles in the obtained track point triangulation according to the triangle hollow index, and extracting the external boundary in the remaining triangular net to obtain the original boundary of the track points; The preprocessing of the original AIS ship track big data adopts: the original AIS ship track big data is gridized by using a geographic hash technology to obtain gridized AIS ship track big data; The module M2 adopts: Module M2.1: smoothing the original boundary of the track point to obtain a refined channel polygon boundary; Module M2.2: extracting a channel boundary vertex point set based on the refined channel polygon boundary, and further extracting a Delaunay triangular network and removing triangles outside the channel boundary to obtain a channel contour; The module M2.1 adopts: using burr edge removal and median filtering technology to smooth the original boundary of the track point to obtain a refined channel polygon boundary; The module M2.2 adopts: using a Delaunay triangulation technology to obtain a new triangular network from the extracted channel boundary vertex point set, and screening triangles inside the channel boundary from the new triangular network through triangle barycentric coordinates; The triangles inside the channel include single-adjacent triangular network, double-adjacent triangular network and triple-adjacent triangular network.

Images