A hatchway edge recognition method for port large machinery

Abstract

This invention relates to a method for hatch edge recognition of large port machinery, belonging to the technical field of ship operation process calculation methods. The technical solution of this invention is as follows: First, all collected point clouds are collected as a FULL complete point cloud; the deck portion is extracted as an AREA region point cloud; the point cloud extending five meters from the middle of the deck towards both sides of the ship, coinciding with the AREA region point cloud, is taken as the CORE core point cloud; the point cloud obtained by calculation and filtering from the CORE core point cloud is taken as the EDGE boundary point cloud; the boundary point information of each point cloud cluster is calculated through point cloud segmentation, thereby obtaining the coordinates of the hatch edge intersection points corresponding to each hatch number. The beneficial effects of this invention are: fast calculation speed, strong applicability, and the ability to quickly and effectively identify and calculate ship hatch coordinates, providing a new key technology for the automation transformation of large port machinery.

Description

Technical Field

[0001] This invention relates to a method for identifying hatch edges of large port machinery, belonging to the technical field of ship operation process calculation methods. Background Technology

[0002] Currently, the construction of smart ports is mainly focused on the container loading and unloading field. The technological development of automation for large port machinery is relatively lagging behind. Large port machinery relies heavily on manual operation, and there is no mature automation solution for large port machinery. Summary of the Invention

[0003] The purpose of this invention is to provide a method for hatch edge recognition of large port machinery. By using point cloud extraction and clustering segmentation, and through the input of ship data before operation and the collection of lidar point cloud data during operation, the working relationship between the radar and the shore machine is established, enabling the shore machine to monitor the ship hatch status in real time. This method has fast calculation speed, strong applicability, and can quickly and effectively identify and calculate the coordinates of ship hatches. It provides a new key technology for the automation transformation of large port machinery and effectively solves the above-mentioned problems existing in the background technology.

[0004] The technical solution of this invention is: a method for identifying hatch edges of large port machinery, comprising the following steps:

[0005] (1) Establish a spatial rectangular coordinate system. The initial position of the ship loader is the bow of the ship. The radar's own coordinate system is used as the standard. The ground direction is the positive x-axis direction, the land side direction is the positive y-axis direction, and the bow direction is the positive z-axis direction.

[0006] (2) Extract point clouds in a spatial rectangular coordinate system and take all collected point clouds as FULL complete point clouds;

[0007] (3) By slicing the point cloud, the deck portion in the full point cloud is extracted as the point cloud of the AREA region;

[0008] (4) Take the intersection of the point cloud. Extend five meters from the middle of the deck towards both sides of the ship to form an intersection with the AREA area. Take points in the intersection area to form the CORE core point cloud.

[0009] (5) Point cloud comparison: The CORE core point cloud is used as the EDGE boundary point cloud by calculation and filtering.

[0010] (6) By segmenting the point cloud, the point cloud at the EDGE boundary is divided to obtain the point cloud cluster at the ship hatch;

[0011] (7) Calculate the boundary point information of each point cloud cluster, and then obtain the coordinates of the hatch edge intersection points corresponding to each cabin number.

[0012] In step (3), the FULL point cloud is divided into slice intervals along the positive z-axis according to the set step size value. Each point in the FULL point cloud is sequentially placed into different slice intervals. Each point is only in one corresponding slice interval, forming an AREA region point cloud belonging to that slice interval.

[0013] In step (5), the maximum and minimum values ​​of the AREA region point cloud along the x, y, and z directions are calculated, the maximum and minimum values ​​of the CORE core point cloud along the x, y, and z directions are calculated, and the EDGE boundary point cloud is extracted; specifically, it includes the following three judgments:

[0014] (a) To address the situation where the sliced ​​area is at the edge of the cabin, the difference between the maximum value and the minimum value of the CORE core point cloud in the x-axis direction is less than the set height difference, ensuring that the bottom point cloud can be effectively removed;

[0015] (b) Ensure that the point cloud data of the ship has small x-axis values ​​on both sides, i.e., high height, and large x-axis values ​​in the middle hull, i.e., low height; and that the difference between the minimum value of the CORE core point cloud and the minimum value of the AREA area point cloud in the x direction is greater than the set height difference.

[0016] (c) Remove point clouds that are below a certain height of the deck. The value of the point cloud in the AREA region on the x direction is less than the sum of the minimum value of the point cloud in the AREA region on the x direction and the set cabin height.

[0017] By traversing all intervals and using the three criteria mentioned above, the resulting point cloud set is the EDGE boundary point cloud.

[0018] In step (6), the EDGE boundary point cloud is divided, and the point cloud cluster at the ship hatch is output through the Euclidean clustering principle; the search radius of the point cloud cluster is set to be smaller than the interval between the hatches; the point cloud clusters are numbered sequentially, with the point cloud clusters in the positive y direction numbered as odd numbers and the point cloud clusters in the negative y direction numbered as even numbers, and the numbering increases sequentially along the negative z-axis direction.

[0019] In step (7), the formula is used.

[0020] h = (i + i % 2) / 2

[0021] The hatch number corresponding to each point cloud cluster is calculated, where h is the hatch number, i is the point cloud cluster number, and % is the remainder.

[0022] In point cloud clustering, the intersection of the outer and rear olives is hatch point 1, the intersection of the front and outer olives is hatch point 2, the intersection of the inner and front olives is hatch point 3, and the intersection of the rear and inner olives is hatch point 4. Each hatch is marked with the coordinates of four points according to this rule to obtain the hatch coordinate value.

[0023] The beneficial effects of this invention are: by using point cloud extraction and clustering segmentation, and through the input of ship data before operation and the collection of lidar point cloud data during operation, the working relationship between the radar and the shore machine is established, enabling the shore machine to monitor the ship hatch situation in real time. This method has fast calculation speed, strong applicability, and can quickly and effectively identify and calculate the coordinates of ship hatches, providing a new key technology for the automation transformation of large port machinery. Attached Figure Description

[0024] Figure 1 This is a schematic diagram of the spatial rectangular coordinate system of the present invention;

[0025] Figure 2 This is a schematic diagram of the full point cloud of the present invention;

[0026] Figure 3 This is a schematic diagram of a ship point cloud slice according to the present invention;

[0027] Figure 4 This is a schematic diagram of the point cloud of the AREA region in this invention;

[0028] Figure 5 This is a schematic diagram of the point cloud of the core CORE of this invention;

[0029] Figure 6 This is a schematic diagram of the EDGE boundary point cloud of the present invention;

[0030] Figure 7 This is a schematic diagram of the de-clustering method of the present invention;

[0031] Figure 8 This is a schematic diagram of the hatch at four points of the present invention;

[0032] The diagram is labeled as follows: FULL (Full Point Cloud), Point Cloud Slice Interval, AREA (Regional Point Cloud), CORE (Core Point Cloud), EDGE (Boundary Point Cloud), Point Cloud Cluster 1, 11, 12, 13, 14, 15, 16, 17, 18, 18, 21, 22, 23, 24. Detailed Implementation

[0033] To make the purpose, technical solutions, and advantages of the invention's embodiments clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described are only a small part of the embodiments of the present invention, not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the protection scope of the present invention.

[0034] A method for identifying hatch edges of large port machinery includes the following steps:

[0035] (1) Establish a spatial rectangular coordinate system. The initial position of the ship loader is the bow of the ship. The radar's own coordinate system is used as the standard. The ground direction is the positive x-axis direction, the land side direction is the positive y-axis direction, and the bow direction is the positive z-axis direction.

[0036] (2) Extract point clouds in a spatial rectangular coordinate system and take all collected point clouds as FULL complete point clouds;

[0037] (3) By slicing the point cloud, the deck portion in the full point cloud is extracted as the point cloud of the AREA region;

[0038] (4) Take the intersection of the point cloud. Extend five meters from the middle of the deck towards both sides of the ship to form an intersection with the AREA area. Take points in the intersection area to form the CORE core point cloud.

[0039] (5) Point cloud comparison: The CORE core point cloud is used as the EDGE boundary point cloud by calculation and filtering.

[0040] (6) By segmenting the point cloud, the point cloud at the EDGE boundary is divided to obtain the point cloud cluster at the ship hatch;

[0041] (7) Calculate the boundary point information of each point cloud cluster, and then obtain the coordinates of the hatch edge intersection points corresponding to each cabin number.

[0042] In step (3), the FULL point cloud is divided into slice intervals along the positive z-axis according to the set step size value. Each point in the FULL point cloud is sequentially placed into different slice intervals. Each point is only in one corresponding slice interval, forming an AREA region point cloud belonging to that slice interval.

[0043] In step (5), the maximum and minimum values ​​of the AREA region point cloud along the x, y, and z directions are calculated, the maximum and minimum values ​​of the CORE core point cloud along the x, y, and z directions are calculated, and the EDGE boundary point cloud is extracted; specifically, it includes the following three judgments:

[0044] (a) To address the situation where the sliced ​​area is at the edge of the cabin, the difference between the maximum value and the minimum value of the CORE core point cloud in the x-axis direction is less than the set height difference, ensuring that the bottom point cloud can be effectively removed;

[0045] (b) Ensure that the point cloud data of the ship has small x-axis values ​​on both sides, i.e., high height, and large x-axis values ​​in the middle hull, i.e., low height; and that the difference between the minimum value of the CORE core point cloud and the minimum value of the AREA area point cloud in the x direction is greater than the set height difference.

[0046] (c) Remove point clouds that are below a certain height of the deck. The value of the point cloud in the AREA region on the x direction is less than the sum of the minimum value of the point cloud in the AREA region on the x direction and the set cabin height.

[0047] By traversing all intervals and using the three criteria mentioned above, the resulting point cloud set is the EDGE boundary point cloud.

[0048] In step (6), the EDGE boundary point cloud is divided, and the point cloud cluster at the ship hatch is output through the Euclidean clustering principle; the search radius of the point cloud cluster is set to be smaller than the interval between the hatches; the point cloud clusters are numbered sequentially, with the point cloud clusters in the positive y direction numbered as odd numbers and the point cloud clusters in the negative y direction numbered as even numbers, and the numbering increases sequentially along the negative z-axis direction.

[0049] In step (7), the formula is used.

[0050] h = (i + i % 2) / 2

[0051] The hatch number corresponding to each point cloud cluster is calculated, where h is the hatch number, i is the point cloud cluster number, and % is the remainder.

[0052] In point cloud clustering, the intersection of the outer and rear olives is hatch point 1, the intersection of the front and outer olives is hatch point 2, the intersection of the inner and front olives is hatch point 3, and the intersection of the rear and inner olives is hatch point 4. Each hatch is marked with the coordinates of four points according to this rule to obtain the hatch coordinate value.

[0053] In practical applications, to obtain hatch data, the entire collected point cloud is first used as the FULL complete point cloud; the deck portion is extracted as the AREA region point cloud; the point cloud extending five meters from the middle of the deck towards both sides of the ship, coinciding with the AREA region point cloud, is used as the CORE core point cloud; the point cloud obtained by calculation and filtering from the CORE core point cloud is used as the EDGE boundary point cloud. Boundary point information for each cluster is calculated through point cloud segmentation, thereby obtaining the coordinates of the hatch edge intersection points corresponding to each hatch number. This method is named the FACE method, taking the first letters of FULL, AREA, CORE, and EDGE.

[0054] 1. Point cloud extraction (calculation of full point cloud)

[0055] First, establish a spatial rectangular coordinate system. Before the program starts, the initial position of the ship loader is the bow. For ease of calculation, the program uses the radar's own coordinate system as the standard, with the ground direction as the positive x-axis, the land-side direction as the positive y-axis, and the bow direction as the positive z-axis. Figure 1 The spatial coordinate system is near the bow of the ship, such as... Figure 2 This is a FULL complete point cloud (a point cloud is represented by a dotted pattern, the same below).

[0056] 2. Point cloud slicing (to obtain the point cloud of the AREA region)

[0057] To facilitate computer processing, the point cloud is segmented into slices of equal thickness. Slices that are too large reduce accuracy, while slices that are too small place a heavy computational load on the workstation. A trade-off was struck, and the step size was set to 0.5 meters. The full point cloud is then divided into several intervals along the ship's length direction (i.e., the -z direction) according to the set step size. Each point in the full point cloud is sequentially placed into different slice intervals, with each point existing only in its corresponding slice interval, forming an AREA region point cloud belonging to that interval.

[0058] 3. Intersection of point clouds (to obtain the CORE core point cloud)

[0059] Based on the point cloud of the AREA region formed by each slice interval, it also belongs to the intersection interval between y = ±5m (that is, five meters from the middle position of the deck towards both sides of the ship). The point cloud in this intersection interval is taken as the CORE core point cloud.

[0060] 4. Point cloud comparison (the EDGE boundary point cloud is obtained from the AREA region point cloud and the CORE core point cloud)

[0061] Calculate the maximum and minimum values ​​of the AREA region point cloud along the x, y, and z directions within the slice interval, and the maximum and minimum values ​​of the CORE core point cloud simultaneously falling within y = ±5m along the x, y, and z directions, and then extract the EDGE boundary point cloud. Specifically:

[0062] (1) In order to deal with the situation where the slice interval is at the edge of the cabin, the difference between the maximum value and the minimum value of the CORE core point cloud in the x-axis direction should be less than the set height difference, because the height from the hatch to the bottom of the cabin is usually greater than 5 meters. In order to ensure that the bottom point cloud can be effectively removed, the height difference is set to 5 meters.

[0063] (2) Ensure that the point cloud data has smaller x-axis values ​​(higher height) on both sides of the ship and larger x-axis values ​​(lower height) in the middle hull. The difference between the minimum value of the CORE core point cloud and the minimum value of the AREA area point cloud in the x-direction should be greater than the set height difference. Since the height from the hatch to the hull is usually greater than 5 meters, the height difference is set to 5 meters.

[0064] (3) Remove point clouds that are below a certain height on the deck, such as point clouds at the bottom of the hold or at the bulkhead. The x-value of the AREA point cloud must be less than the sum of the minimum x-value of the AREA point cloud and the set hold height. The hatch height is 5 meters.

[0065] By traversing all slice intervals and using the three criteria mentioned above, the resulting point cloud set is the EDGE boundary point cloud.

[0066] 5. Point cloud segmentation (using Euclidean clustering principles to segment the EDGE boundary point cloud)

[0067] The KD-Tree-based nearest neighbor query algorithm is an important preprocessing method for accelerating the Euclidean Cluster Extraction algorithm.

[0068] The principle of a KD-Tree is as follows: A KD-Tree is a binary tree where each node is a K-dimensional point. All non-leaf nodes can be viewed as acting on a hyperplane that divides the space into two half-spaces. The subtree to the left of a node represents points to the left of the hyperplane, and the subtree to the right of a node represents points to the right of the hyperplane. The hyperplane is chosen as follows: each node is associated with the K-dimensional dimension perpendicular to the hyperplane. Therefore, if the partition is chosen along the x-axis, all nodes with x-values ​​less than a specified value will appear in the left subtree, and all nodes with x-values ​​greater than a specified value will appear in the right subtree. In this way, the hyperplane can be determined using this x-value, and its normal is the unit vector along the x-axis.

[0069] The specific implementation method of Euclidean clustering is roughly as follows:

[0070] (1) Find a point A1 in space, find the points closest to it using KDTREE, and determine the distance of these points to A1;

[0071] (2) Place points A2, A3... whose distance is less than the set threshold into cluster Q;

[0072] (3) Find a point A in cluster Q. m Repeat step 1 to find A. m+1 A m+2 A m+3 …put them all into cluster Q;

[0073] (4) Find a point A in cluster Q. n Repeat step 1 to find A. n+1 A n+2 A n+3 …put them all into cluster Q;

[0074] (5) When no more new points can be added to cluster Q, the search is complete.

[0075] Because the spacing between ship compartments is typically greater than 3 meters, to ensure that adjacent clusters do not merge into a single cluster, the cluster search radius is set to 3 meters. Input the boundary point cloud using the EDGE function, and output the point cloud clusters at the ship hatches using a specific method. Number the point cloud clusters sequentially, with odd numbers for clusters in the positive y-direction and even numbers for clusters in the negative y-direction, increasing sequentially along the negative z-axis.

[0076] 6. Point cloud computing (calculates cabin number and hatch coordinates)

[0077] Assuming cabin number h and cluster number i, the formula is used.

[0078] h = (i + i % 2) / 2

[0079] This ensures that each point cloud cluster corresponds to a specific hatch number (% represents the remainder). Point cloud cluster 1 and point cloud cluster 2 correspond to hatch number 1, and so on.

[0080] In point cloud clustering, the intersection of the outer and rear olives is hatch point 1; the intersection of the front and outer olives is hatch point 2; the intersection of the inner and front olives is hatch point 3; and the intersection of the rear and inner olives is hatch point 4. Each hatch is marked with the coordinates of four points according to this rule.

[0081] Assuming cabin number h, the formula for the four-point coordinates is as follows:

[0082]

[0083] P h1.x Here is the x-coordinate of point 1 at hatch h; Let x be the minimum value of the x-coordinate of the cloud cluster at point i.

[0084]

[0085] P h1.y Here is the y-coordinate of point 1 at hatch h; Let y be the maximum value of the cloud clustering for point i.

[0086]

[0087] P h1.z Here is the z-coordinate of point 1 at hatch h; Let z be the maximum value of the z-coordinate of the cloud cluster at point i.

[0088]

[0089] P h2.x Here is the x-coordinate of point 2 at hatch h; Let x be the minimum value of the x-coordinate of the cloud cluster at point i.

[0090]

[0091] P h2.y Here is the y-coordinate of point 2 at hatch h; Let y be the minimum value of the cloud clustering for point i.

[0092]

[0093] P h2.zHere is the z-coordinate of point 2 at hatch h; Let z be the maximum value of the z-coordinate of the cloud cluster at point i.

[0094]

[0095] P h3.x Here is the x-coordinate of point 3 at hatch h; Let x be the maximum value of the x-coordinate of the cloud cluster at point i.

[0096]

[0097] P h3.y Here is the y-coordinate of point 3 at hatch h; Let y be the minimum value of the cloud clustering for point i.

[0098]

[0099] P h3.z Here is the z-coordinate of point 3 at hatch h; Let z be the minimum value of the z-coordinate of the cloud cluster at point i.

[0100]

[0101] P h4.x Here is the x-coordinate of point 4 at hatch h; Let x be the maximum value of the x-coordinate of the cloud cluster at point i.

[0102]

[0103] P h4.y Here is the y-coordinate of point 4 at hatch h; Let y be the maximum value of the cloud clustering for point i.

[0104]

[0105] P h4.z Here is the z-coordinate of point 4 at hatch h; Let z be the minimum value of the z-coordinate of the cloud cluster at point i.

[0106] The coordinates of the four points in the remaining cabins can be deduced in the same way.

Claims

1. A method for identifying hatch edges of large port machinery, characterized in that... Includes the following steps: (1) Establish a spatial rectangular coordinate system. The initial position of the ship loader is the bow of the ship. The radar's own coordinate system is used as the standard. The ground direction is the positive x-axis direction, the land side direction is the positive y-axis direction, and the bow direction is the positive z-axis direction. (2) Extract point clouds in a spatial rectangular coordinate system and take all the collected point clouds as FULL complete point clouds; (3) By slicing the point cloud, the deck part in the FULL complete point cloud is extracted as the AREA region point cloud; the FULL complete point cloud is divided into slice intervals along the positive z-axis according to the set step value, and each point in the FULL complete point cloud falls into different slice intervals in sequence. Each point is only in the corresponding slice interval, forming the AREA region point cloud belonging to that slice interval. (4) Take the intersection of the point cloud. Extend five meters from the middle of the deck towards both sides of the ship to form an intersection with the AREA area. Take points in the intersection area to form the CORE core point cloud. (5) Point cloud comparison: The CORE core point cloud is used as the EDGE boundary point cloud through calculation and filtering; the maximum and minimum values ​​of the AREA region point cloud along the x, y, and z directions are calculated, and the maximum and minimum values ​​of the CORE core point cloud along the x, y, and z directions are calculated to extract the EDGE boundary point cloud; specifically, the following three judgments are included: (a) To address the situation where the sliced ​​area is at the edge of the cabin, the difference between the maximum value and the minimum value of the CORE core point cloud in the x-axis direction is less than the set height difference, ensuring that the bottom point cloud can be effectively removed; (b) Ensure that the point cloud data of the ship has small x-axis values ​​on both sides, i.e., high height, and large x-axis values ​​in the middle hull, i.e., low height; and that the difference between the minimum value of the CORE core point cloud and the minimum value of the AREA area point cloud in the x direction is greater than the set height difference. (c) Remove point clouds below a certain height of the deck. The value of the point cloud in the AREA region on the x direction is less than the sum of the minimum value of the point cloud in the AREA region on the x direction and the set cabin height. By traversing all intervals and using the three criteria mentioned above, the resulting point cloud set is the EDGE boundary point cloud. (6) By segmenting the point cloud, the point cloud at the EDGE boundary is divided to obtain the point cloud cluster at the ship hatch; (7) Calculate the boundary point information of each point cloud cluster, and then obtain the coordinates of the hatch edge intersection point corresponding to each cabin number.

2. The method for hatch edge recognition of large port machinery according to claim 1, characterized in that: In step (6), the EDGE boundary point cloud is divided, and the point cloud cluster at the ship hatch is output through the Euclidean clustering principle; the search radius of the point cloud cluster is set to be smaller than the interval between the hatches; the point cloud clusters are numbered sequentially, with the point cloud clusters in the positive y direction numbered as odd numbers and the point cloud clusters in the negative y direction numbered as even numbers, and the numbering increases sequentially along the negative z-axis direction.

3. The method for hatch edge recognition of large port machinery according to claim 1, characterized in that: In step (7), the formula is used. The hatch number corresponding to each point cloud cluster is calculated, where For the hatch number, Here, % represents the point cloud cluster number, and % represents the remainder. In point cloud clustering, the intersection of the outer and rear olives is hatch point 1, the intersection of the front and outer olives is hatch point 2, the intersection of the inner and front olives is hatch point 3, and the intersection of the rear and inner olives is hatch point 4. Each hatch is marked with the coordinates of four points according to this rule to obtain the hatch coordinate value.

Images