Search

CN-121973903-A - Marine host cabin entering simulation method and system based on multi-view measurement

CN121973903ACN 121973903 ACN121973903 ACN 121973903ACN-121973903-A

Abstract

The invention discloses a marine main engine cabin entering simulation method and system based on multi-view measurement. The method comprises the steps of collecting cabin space depth image information through multiple visual angles, constructing three-dimensional point cloud data of an in-cabin environment, unifying the multi-view point cloud to the same global coordinate system based on camera calibration, accurately fusing the multi-view point cloud by adopting a point cloud registration algorithm introducing a dynamic weight and a local rejection mechanism, extracting the boundary of a cabin effective space by utilizing an alpha shape algorithm, modeling a motion track of a host cabin entering process, and simultaneously realizing rapid real-time calculation of the minimum gap between a host envelope and the cabin boundary by constructing a gridding distance field of the cabin space, so as to judge whether the host has collision or insufficient space risk under a given cabin entering path. The invention can realize high-precision, high-efficiency and high-real-time host cabin entering space verification in a complex and narrow cabin environment, and can provide reliable technical support for optimization and decision-making of a marine host cabin entering scheme.

Inventors

  • JIN YONGFEI
  • JIN LIAN

Assignees

  • 江苏海通海洋工程装备有限公司

Dates

Publication Date
20260505
Application Date
20260123

Claims (10)

  1. 1. The marine main engine cabin entering simulation method based on multi-view measurement is characterized by comprising the following steps of: S1, acquiring a multi-view space point cloud reflecting the space morphology of a cabin entering area through a multi-view depth camera arranged on the cabin entering path and the hatch area of a ship cabin; s2, carrying out coordinate unification and fusion processing on the multi-view space point cloud, and constructing a complete three-dimensional space point cloud of a cabin entrance area of the cabin so as to obtain an integral three-dimensional expression of an actual space structure of the cabin; S3, extracting a cabin space boundary capable of representing the cabin available cabin entering space range based on the complete three-dimensional space point cloud, wherein the cabin space boundary is used for limiting a space region allowed to be occupied in a cabin entering process of a host; s4, obtaining appearance data of the marine host, and constructing a host envelope based on the appearance data, wherein the host envelope is used for representing the overall appearance size and the space occupation range of the host; S5, combining a preset or planned host cabin entering path, and establishing a motion trail model of the host envelope in a cabin entering process, wherein the motion trail model is used for describing space occupation states of the host in different time positions and postures in the cabin entering process; S6, calculating a gap distance between the host envelope and the cabin space boundary aiming at any moment in the motion trail model in the host cabin entering simulation process so as to represent a space margin between the host and the cabin; S7, comprehensively judging the whole cabin entering process of the host according to the clearance distance, and judging that the host can enter the cabin along the cabin entering path when the clearance distance meets the preset safety condition, or judging that the space interference risk exists in the cabin entering process of the host.
  2. 2. The marine host machine entry simulation method according to claim 1, wherein in step S2: Firstly, based on the space calibration relation among the multi-view depth cameras, carrying out coordinate transformation on multi-view space point clouds from different depth cameras to map the multi-view space point clouds to a unified global coordinate system, and then carrying out registration fusion on the mapped multi-view space point clouds by utilizing an ICP algorithm under the global coordinate system, and reducing the space deviation among different view data in an iterative optimization mode to obtain the complete three-dimensional space point clouds of the cabin entering region.
  3. 3. The marine host machine in-cabin simulation method according to claim 2, wherein the ICP algorithm includes the steps of: s201, selecting one of the multi-view space point clouds as a reference point cloud P 0 , and the rest as point clouds to be registered; S202, for each point P 0 in the reference point cloud P 0 , finding a point P i closest to the point cloud to be registered; S203, calculating a rotation matrix by a least square method Translation vector Minimizing the distance between the reference point cloud P 0 and the point cloud P i to be registered, namely optimizing the target ; S204, continuously repeating the step S202 and the step S203 until the error between the point clouds converges to a preset threshold value, so as to meet the precision requirement and output the point cloud registration result.
  4. 4. A marine host computer in-cabin simulation method according to claim 3, wherein the optimization objective is replaced with: Wherein, the Is a dynamic weight function and has Wherein, the Representing the euclidean distance between point P 0 in the reference point cloud P 0 and point P i in the point cloud to be registered, Is a parameter for controlling the weight decay rate; In each ICP algorithm iteration, the distance between each p 0 and p i point pair is calculated, and when the distance of the point pair is greater than the set maximum matching distance threshold, the point pair is considered to be a false match, and is rejected.
  5. 5. The marine host machine cabin-entering simulation method according to claim 1, wherein the step S3 of extracting cabin space boundaries by using an α -shape algorithm comprises the steps of: S301, performing gridding treatment on the complete three-dimensional space point cloud by using Delaunay triangulation, and converting point cloud data into triangular grids; S302, calculating the radius r of the circumscribing circle of each triangle mesh, comparing the radius r with a given parameter alpha, and reserving the boundary of the triangle mesh with the radius not larger than alpha, otherwise, eliminating the boundary of the triangle mesh; S303, splicing and constructing an outer envelope conforming to a parameter alpha according to the screened triangular grid boundary, and taking the outer envelope as a cabin space boundary S hull , wherein the parameter alpha is used for controlling the extraction precision of the cabin space boundary, and the smaller the alpha value is, the finer the boundary is.
  6. 6. The marine host machine cabin-entering simulation method according to claim 5, wherein in the step S4, an axis alignment rectangular frame of the three-dimensional appearance of the host machine is taken as a host machine envelope, and in the step S5, a motion trail model of the host machine envelope in the cabin-entering process is expressed as follows: wherein B (t) represents the motion track of the host envelope, R (t) is a rotation matrix at time t, represents the rotation of the host in the hoisting process, and comprises the following components: Wherein, the 、 And Respectively representing the roll angle, pitch angle and yaw angle at time t; d (t) is a translation vector at time t, describing the position of the host computer over time during hoisting, and has: Wherein, the 、 And The displacement of the host in the x, y, z directions at time T, respectively, T representing the transpose.
  7. 7. The marine host machine in-cabin simulation method of claim 6, wherein in steps S6 and S7, a minimum gap between the host machine and the cabin is defined as a minimum euclidean distance between a host machine envelope boundary and a cabin boundary point, and is denoted as: Where p is the point in the cabin space boundary S hull , b is the surface point of the host envelope, B (t) represents the outer surface of the host envelope at the time t, and when the minimum gap g (t) is smaller than the preset safety gap threshold g safe , the risk of spatial interference at the time is judged.
  8. 8. The marine main engine cabin entering simulation method according to claim 7, wherein in the steps S6 and S7, the nearest distance from any position in cabin space to cabin space boundary is pre-calculated into a table look-up structure in an off-line stage, a conservative distance lower limit is obtained by table look-up for a main engine envelope surface sampling point in an on-line stage, and if the lower limit is not less than a safety threshold, a space interference risk exists; the off-line phase comprises the steps of: the cabin space is subjected to grid division, and if the grid side length is h, the grid quantity capable of covering the cabin space is expressed as: Where n x 、n y 、n z represents the number of meshes in the x, y, z directions, x min 、y min 、z min represents the minimum value of the points on the cabin space boundary S hull in the x, y, z directions, x max 、y max 、z max represents the maximum value of the points on the cabin space boundary S hull in the x, y, z directions, m is the expansion margin, Representing an upward rounding; Any grid is represented by an integer index (i, j, k), i e [0, n x -1],j∈[0, n y -1],k∈[0, n z -1], the geometric center coordinate v i,j,k of which is taken as: For each grid center, its nearest distance to the cabin space boundary S hull is calculated: Indexed by (i, j, k) and corresponding Constructing a lookup table for the value and storing the lookup table; The online phase comprises the steps of: At time t, from the outer surface of the host envelope B (t) selecting a group of surface sampling points: E is the sampling point number, for any sampling point Calculating the grid index of the grid: Then the lookup table is queried to obtain the nearest boundary distance of the corresponding grid center And will As a lower distance limit from the sampling point to the cabin boundary; If the lower limit of the distance of any sampling point at any time is not smaller than a preset safety threshold, the situation that the host can enter the cabin along the cabin entering path is judged, and otherwise, the risk of spatial interference exists in the cabin entering process of the host is judged.
  9. 9. The marine host computer in-cabin simulation method of claim 8, wherein the offline stage further comprises the steps of taking cabin space boundary S hull as input data for constructing a three-dimensional k-d tree, wherein the construction process of the k-d tree follows a standard recursive space division method, and in each layer of recursion, ordering the points contained in S hull , selecting middle points as division nodes so as to divide a point set into two subspaces; After the meshing is completed, the cabin space is discretized into a group of grid center points, for each grid center point v i,j,k , a nearest neighbor search is performed in the k-d tree to obtain a cabin boundary point p nearest to the grid center, and Euclidean distance between the cabin boundary point p and the grid center point is calculated ; The obtained result is organized into a three-dimensional array or equivalent hash mapping structure: The data structure I.e. a gridded distance field, wherein the index (i, j, k) corresponds to the grid position, value Constructing a grid-like distance field D [ after completion of the construction Will act as a look-up table.
  10. 10. A marine host cabin entry simulation system based on the method of any one of claims 1 to 9, comprising the following modules: the space sensing and data acquisition module is used for carrying out multi-view space measurement on the cabin entering area of the ship cabin, and synchronously acquiring space measurement data reflecting the geometric structure of the cabin entering area from different view angles so as to form original space data of the cabin environment; The space modeling and boundary extraction module is connected with the space perception and data acquisition module and is used for carrying out coordinate unification and registration fusion on the multi-view space measurement data, constructing a three-dimensional space model of a cabin entrance area, and extracting a cabin space boundary model representing the usable cabin entrance space range of a cabin on the basis of the three-dimensional space model; the host envelope modeling and motion simulation module is used for acquiring the shape data of the marine host and constructing a host space envelope model, and applying pose transformation which changes with time to the host space envelope model by combining a preset or planned host cabin entering path so as to establish a dynamic motion model of the host in a cabin entering process; The space gap calculation module is connected with the space modeling and boundary extraction module and the host envelope modeling and motion simulation module and is used for calculating space gap information between the host space envelope model and the cabin space boundary model aiming at any moment in the dynamic motion model in the host cabin entering simulation process; and the risk judging and result outputting module is connected with the space clearance calculating module and is used for comprehensively judging the whole process of entering the cabin of the host based on the space clearance information and outputting the feasibility result of entering the cabin of the host and corresponding risk information.

Description

Marine host cabin entering simulation method and system based on multi-view measurement Technical Field The invention belongs to the technical field of ship construction, and particularly relates to a marine host cabin entering simulation method and system based on multi-view measurement. Background In the cabin entering process of the marine main engine, ensuring whether the reserved space meets the requirement of the main engine for smoothly entering the cabin is an important link. Most of the existing cabin entering verification methods rely on manual ranging or single-point laser scanning, and whether the space is enough or not is judged by comparing the existing cabin entering verification methods with a design drawing. However, these conventional methods have significant drawbacks in practical applications, particularly in terms of efficiency and accuracy in the verification process. Firstly, as one of the main ways of traditional cabin entering verification, the efficiency and the precision of the manual ranging have larger problems. Manual measurement generally relies on manual force and hand tools, particularly in narrow and complex cabin spaces, and is time-consuming and susceptible to mishandling by personnel and environmental factors, resulting in large errors in measurement results. This approach fails to meet the need for rapid and accurate performance in some special situations, especially when a compact schedule is required during the construction of a ship, the time and labor costs consumed by manual measurements are obviously unacceptable. Secondly, the application of the single-point laser scanning technology in cabin entering verification has certain limitation. Although laser scanning is capable of generating highly accurate point cloud data, data acquisition is typically only possible from a single view angle due to the inherent limitations of scanning equipment. As such, the scanning device often cannot cover every corner and detail of the cabin, resulting in a dead zone problem with the data. Even by scanning and stitching a plurality of times, there is inevitably a data stitching error, so that the finally generated three-dimensional model may not be in line with the actual situation. Especially, under the condition that a hoisting path is tortuous and the structure in the cabin is complex, the splicing error of the point cloud data is more obvious, so that the accuracy of a final judgment result is affected. In addition, existing in-cabin verification methods lack rapid feedback of spatial decision results. In practice, especially during heavy ship construction, the time taken to hoist is very valuable. The existing manual ranging and laser scanning methods generally require a long time to complete data acquisition and subsequent comparison and analysis, so that a conclusion on whether the space is enough cannot be given in real time. This time delay not only increases the construction period, but also tends to cause unnecessary waiting and adjustment, thereby affecting the efficiency and cost of the overall construction. Moreover, the traditional method relies on more manual links of drawing comparison and data processing, so that the influence of human factors is very easy to occur, and the error in the judging process is increased. With the increasing complexity of ship construction and hoisting processes, the operability and expandability of the prior art in space judgment are limited, and the prior art cannot meet the more complex and flexible cabin entering verification requirements, especially in the aspects of efficient and accurate cabin entering judgment and quick decision, and still cannot meet the requirements of modern ship construction. Therefore, the existing manual ranging and single-point laser scanning technologies, although capable of providing basic data of space verification in some cases, still have significant shortcomings in efficiency, precision, instantaneity and the like, which make it difficult for the traditional method to provide accurate and efficient cabin entering space verification in a complex actual construction environment. Disclosure of Invention In order to solve the technical problems, the invention provides a marine main engine cabin entering simulation method and system based on multi-view measurement. Specifically, the technical scheme provided by the invention is as follows: a marine main engine cabin entering simulation method based on multi-view measurement comprises the following steps: S1, acquiring a multi-view space point cloud reflecting the space morphology of a cabin entering area through a multi-view depth camera arranged on the cabin entering path and the hatch area of a ship cabin; s2, carrying out coordinate unification and fusion processing on the multi-view space point cloud, and constructing a complete three-dimensional space point cloud of a cabin entrance area of the cabin so as to obtain an integral three-dimensional ex