Ship body segment geometry parameter measurement method and device and storage medium

Abstract

The application relates to a ship body segment geometric parameter measurement method and device and a storage medium, wherein the method comprises the following steps: acquiring multiple ship body segment images by using a camera carried by a drone to perform multi-view image acquisition on a target ship body segment; identifying key feature points of the target ship body segment based on the ship body segment images, and constructing a three-dimensional model of the target ship body segment based on the key feature points; calibrating the camera to obtain internal and external parameters of the camera; selecting a key feature line segment in the three-dimensional model, and calculating a model distance of the key feature line segment by using the calibrated internal and external parameters; performing multi-dimensional error analysis on the model distance and actual measured geometric parameters of the target ship body segment; and adjusting drone shooting parameters and / or parameters of the three-dimensional model according to an error analysis result to perform iterative optimization until the measurement accuracy meets preset requirements. By using the application, efficient, accurate and verifiable measurement of absolute geometric parameters of a ship body segment is realized.

Description

Technical Field

[0001] This invention relates to the field of hull section parameter measurement technology in shipbuilding, and particularly to a method, device and storage medium for measuring the geometric parameters of hull sections. Background Technology

[0002] Hull sections are the basic building blocks in shipbuilding, and the accurate measurement of their geometric parameters has a significant impact on the assembly precision of subsequent section joining. In the field of shipbuilding, accurately acquiring the geometric parameters such as the external dimensions and surface shapes of hull sections is a key step in controlling construction errors and ensuring smooth section docking. Traditionally, the measurement of hull section geometric parameters mainly relies on ground surveying equipment such as total stations, and involves manually setting up target points for point-by-point measurement.

[0003] However, existing technologies typically employ drones for image acquisition and combine them with 3D modeling software to generate models. While this can improve measurement efficiency, in practical applications, it has been found that the geometric accuracy of the generated 3D models in key areas is insufficient to meet the accuracy assessment requirements in the early stages of ship construction, and the reliability of the measurement results lacks effective verification methods.

[0004] Therefore, how to achieve efficient and accurate measurement of the geometric parameters of hull sections and provide verifiable measurement accuracy is a technical problem that urgently needs to be solved in the current shipbuilding field. Summary of the Invention

[0005] In view of this, it is necessary to provide a method, device and storage medium for measuring the geometric parameters of hull sections, so as to solve the technical problems of low efficiency and accuracy of hull section geometric parameter measurement and lack of verifiable accuracy of measurement results in the prior art.

[0006] To address the aforementioned problems, in a first aspect, the present invention provides a method for measuring the geometric parameters of ship hull sections, comprising: Multiple images of the target ship hull sections were obtained by using a camera mounted on a drone to capture images from multiple perspectives. Based on the image of the hull segment, key feature points of the target hull segment are identified, and a three-dimensional model of the target hull segment is constructed based on the key feature points; The camera is calibrated using multiple calibration plates arranged around the target hull section to obtain the camera's intrinsic and extrinsic parameters. Key feature lines are selected in the three-dimensional model, and the model distance of the key feature lines is calculated using the calibrated intrinsic and extrinsic parameters. A multi-dimensional error analysis is performed between the model distance and the actual measured geometric parameters of the target hull segment. Based on the error analysis results, the UAV shooting parameters and / or the parameters of the three-dimensional model are adjusted and iteratively optimized until the measurement accuracy meets the preset requirements.

[0007] In one possible implementation, the key feature points include feature points located at weld seams, butt joint edges, and structural stiffeners; the identification of key feature points of the target hull segment based on the hull segment image includes: Image features are extracted from the multiple images of the ship hull segments using a feature matching algorithm, which includes a scale-invariant feature transform algorithm or an ORB feature matching algorithm. Based on the image features, feature points located at weld seams, butt joint edges, and structural reinforcing ribs on the target hull section are identified as key feature points.

[0008] In one possible implementation, constructing the three-dimensional model of the target hull segment based on the key feature points includes: Encrypted feature point matching is performed on the segmented regions where the key feature points are located in the multiple hull segment images to obtain the feature matching information of the segmented regions. The depth information of the multiple ship hull segment images is extracted based on the multi-view stereo vision algorithm to obtain the corresponding segment depth information, and the point cloud density is increased in the area near the key feature points according to the feature matching information. Dense point cloud sampling is performed on the local image region containing the key feature points to obtain the dense point cloud data of the local image region; Based on the segment depth information, high-density point cloud data covering the surface of the target hull segment is generated; The high-density point cloud data is used to perform three-dimensional reconstruction to generate a three-dimensional model of the target ship hull segment.

[0009] In one possible implementation, after constructing a 3D model of the target hull segment based on the key feature points, the method further includes: Template matching is performed based on a preset hull segment feature library to identify the segment type of the target hull segment; Based on the segment type, an adaptive resolution modeling strategy is adopted to perform high-resolution modeling on the key areas of the docking surfaces of the corresponding hull segments in the 3D model, and standard resolution modeling on other areas.

[0010] In one possible implementation, calibrating the camera includes: The mapping relationship between the pixel coordinates and world coordinates of the feature marker points on the calibration board is established using the Zhang Zhengyou calibration method. The intrinsic and extrinsic parameters of the camera are solved based on the mapping relationship. The intrinsic parameters include focal length, principal point, and distortion coefficients, and the extrinsic parameters include rotation matrix and translation vector.

[0011] In one possible implementation, the error analysis results include at least: absolute error, relative error, or spatial position error distribution; the multi-dimensional error analysis of the model distance and the actual measured geometric parameters of the target hull segment includes: The absolute error is obtained by calculating the difference between the model distance and the actual measured geometric parameters. The relative error is obtained by calculating the percentage of the absolute error to the actual measured geometric parameter; For each key feature line segment, its error components in the X, Y, and Z directions are calculated to obtain the spatial position error distribution.

[0012] In one possible implementation, the error analysis results include at least: absolute error, relative error, or spatial position error distribution; the UAV shooting parameters include UAV flight parameters and camera parameters; the 3D modeling parameters include feature point extraction threshold, image matching accuracy level, and dense point cloud reconstruction density; the step of performing multi-dimensional error analysis on the model distance and the actual measured geometric parameters of the target hull segment, and iteratively optimizing the UAV shooting parameters and / or the 3D model parameters based on the error analysis results, includes: When the absolute error exceeds the first preset threshold, the camera's intrinsic parameters are recalibrated, and the drone's flight altitude and the camera's shutter speed are adjusted, while the feature point extraction threshold is reduced. When the relative error exceeds the second preset threshold, adjust the image overlap rate, the UAV flight speed, and the number and distribution of ground control points, and increase the reconstruction density of the dense point cloud; When the spatial position error exceeds the third preset threshold in the horizontal direction, increase the tilt shooting route and adjust the camera sensitivity; When the spatial position error in the vertical direction exceeds the fourth preset threshold, supplement the image by taking images at the elevation angle and adjust the density reconstruction parameters in the vertical direction; After each adjustment, the camera is recalibrated, the distance of key feature line segments is calculated, and multi-dimensional error analysis is performed until the measurement accuracy meets the preset requirements.

[0013] In one possible implementation, before acquiring multi-view images of the target ship hull sections using a camera mounted on a drone, the method further includes: The takeoff point of the UAV is selected based on the hull layout of the target hull sections, environmental safety, wind force, and weather factors; and / or, Set the image overlap rate during drone shooting according to the modeling accuracy requirements; and / or, Establish standardized data naming rules corresponding to the multiple images of ship hull segments collected. The naming rules include the shooting date, segment number, data type, and shooting angle information.

[0014] In one possible implementation, the method of acquiring multi-view images of the target ship hull sections using a camera mounted on a drone includes: The drone's flight altitude is dynamically adjusted based on the position and height of the shooting surface of the target hull segment; and / or, The drone's flight speed is set to a preset speed.

[0015] Secondly, the present invention also provides a device for measuring the geometric parameters of a hull section, comprising: The acquisition unit is used to acquire multi-view images of the target ship hull sections using a camera mounted on a drone, and obtain multiple images of the ship hull sections. The modeling unit is used to identify key feature points of the target hull segment based on the hull segment image, and to construct a three-dimensional model of the target hull segment based on the key feature points; The calibration unit is used to calibrate the camera based on multiple calibration plates arranged around the target hull section to obtain the camera's intrinsic and extrinsic parameters. The calculation unit is used to select key feature line segments in the three-dimensional model and calculate the model distance of the key feature line segments using calibrated intrinsic and extrinsic parameters. The verification and optimization unit is used to perform multi-dimensional error analysis on the model distance and the actual measured geometric parameters of the target hull segment, and to iteratively optimize the UAV shooting parameters and / or the parameters of the three-dimensional model according to the error analysis results until the measurement accuracy meets the preset requirements.

[0016] Thirdly, the present invention also provides a computer-readable storage medium for storing a computer-readable program or instructions, which, when executed by a processor, can implement the steps in the hull section geometric parameter measurement method described in any of the above implementations.

[0017] The beneficial effects of this invention are: The method for measuring the geometric parameters of ship hull sections provided by this invention improves the efficiency of measuring the geometric parameters of ship hull sections by using a camera mounted on an unmanned aerial vehicle to acquire multiple images of the target ship hull section from multiple perspectives. Based on the images of the ship hull sections, key feature points of the target ship hull section are identified, and a three-dimensional model of the target ship hull section is constructed based on the key feature points, which improves the representation accuracy of the three-dimensional model in key areas, thereby ensuring the accuracy of subsequent measurements. Meanwhile, the automated modeling process significantly reduces manual intervention and improves modeling efficiency. Based on multiple calibration plates placed around the target hull section, the camera is calibrated to obtain its intrinsic and extrinsic parameters, thus accurately describing the camera's imaging geometry and providing a basis for coordinate correction in subsequent measurements. Key feature segments are selected in the 3D model, and the model distance of these segments is calculated using the calibrated intrinsic and extrinsic parameters. The endpoint coordinates of these key feature segments are then corrected using the calibrated intrinsic and extrinsic parameters to eliminate measurement errors caused by lens distortion, thereby obtaining accurate model distances. Multi-dimensional error analysis is performed between the model distances and the actual measured geometric parameters of the target hull section. Based on the error analysis results, the UAV shooting parameters and / or 3D model parameters are adjusted iteratively until the measurement accuracy meets the preset requirements, achieving efficient, accurate, and verifiable measurement of the absolute geometric parameters of the hull section. Attached Figure Description

[0018] Figure 1 A schematic flowchart of an embodiment of the method for measuring geometric parameters of hull sections provided by the present invention; Figure 2 This is a schematic diagram of a longitudinal frame double-bottom structure section of a ship pool provided by the present invention; Figure 3 A schematic diagram of a three-dimensional model of the target hull section provided by the present invention; Figure 4 A schematic diagram of the key feature line segments provided for this invention; Figure 5 A schematic diagram of the structure of the hull section geometric parameter measuring device provided by the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and 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 scope of protection of the present invention.

[0020] In the description of the embodiments of the present invention, unless otherwise stated, "a plurality of" means two or more.

[0021] The terms "first," "second," etc., used in the embodiments of this invention are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a technical feature defined with "first" or "second" may explicitly or implicitly include at least one of that feature.

[0022] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0023] This invention provides a method, apparatus, and storage medium for measuring the geometric parameters of ship hull sections, which will be described below.

[0024] The execution subject of the hull section geometric parameter measurement method in this application embodiment can be the hull section geometric parameter measurement device provided in this application embodiment, or different types of electronic devices such as server equipment, physical host, or user equipment (UE) that integrate the hull section geometric parameter measurement device. The hull section geometric parameter measurement device can be implemented in hardware or software. The UE can be a terminal device such as a smartphone, tablet computer, laptop computer, handheld computer, desktop computer, or personal digital assistant (PDA).

[0025] Figure 1 This is a schematic flowchart of an embodiment of the method for measuring the geometric parameters of ship sections provided by the present invention, as shown below. Figure 1 As shown, the methods for measuring the geometric parameters of hull sections include: S101. Use a camera mounted on a drone to collect multi-view images of the target ship hull sections, and obtain multiple images of the ship hull sections.

[0026] The target hull section refers to the hull section that requires absolute geometric parameter measurement. For example, in this embodiment, a longitudinally framed double-bottom structure section under laboratory conditions is used as the target hull section. Figure 2 The diagram shown is a schematic of the corresponding model of the double-bottom structure section of the longitudinal frame of the ship's pool. The basic dimensions of the model are 1000mm×900mm×600mm (length×width×depth).

[0027] Specifically, during image acquisition, a drone equipped with a high-definition camera can be controlled to photograph the target hull section from multiple angles, including the front, side, and top, covering the section surface and docking area. To avoid blind spots, the shooting angle for each target hull section should cover the entire area as much as possible. Multi-view image acquisition by drones improves the efficiency of measuring the geometric parameters of the hull sections. Drones can quickly cover all areas of large hull sections, acquiring comprehensive and complete image data. Compared to traditional manual point-by-point measurement methods, this shortens on-site operation time, reduces labor costs and operational difficulty, and is suitable for rapid measurement needs in complex shipbuilding environments.

[0028] As an optional implementation, the drone can also be equipped with a LiDAR sensor to improve measurement accuracy and stability in complex environments through multi-source data fusion.

[0029] S102. Identify key feature points of the target hull segment based on the hull segment image, and construct a three-dimensional model of the target hull segment based on the key feature points.

[0030] Among them, key feature points refer to feature locations that have a significant impact on the structural strength and assembly accuracy of the hull sections, such as welds, the edges of mating surfaces, and structural stiffeners.

[0031] A three-dimensional model refers to the three-dimensional spatial coordinate information of each point on the surface of a segment of the target ship's hull.

[0032] Specifically, a feature matching algorithm is used to extract image features from multiple images and identify key feature points. The acquired image data can be imported into 3D modeling software, which automatically performs feature matching and alignment on the hull segment images to generate 3D point cloud data. This data is then used to reconstruct a 3D model of the target hull segment. For example... Figure 3 The image shown is a schematic diagram of a three-dimensional model of the target ship's hull sections.

[0033] This embodiment improves the representation accuracy of the 3D model in key areas by identifying key feature points and constructing a 3D model based on these points, thereby ensuring the accuracy of subsequent measurements. At the same time, the automated modeling process significantly reduces manual intervention and improves modeling efficiency.

[0034] S103. Based on the multiple calibration plates arranged around the target hull section, the camera is calibrated to obtain the camera's intrinsic and extrinsic parameters.

[0035] Among them, the camera's intrinsic parameters refer to the camera's internal geometric parameters, such as focal length, principal point coordinates, and distortion coefficients; the camera's extrinsic parameters refer to its position and attitude parameters relative to the world coordinate system, such as rotation matrix and translation vector.

[0036] The calibration plate specifications can be selected according to the size of the target hull section. For sections with a length greater than 10 meters, a 1-meter × 1-meter calibration plate is used, and for sections with a length less than or equal to 10 meters, a 0.5-meter × 0.5-meter calibration plate is used. At least three calibration plates of the selected specifications should be arranged around the target hull section.

[0037] Specifically, multiple high-precision calibration plates are arranged at different locations around the target hull sections to ensure that the calibration plates can cover different spatial orientations. The surfaces of the calibration plates are engraved with feature marks of known precise dimensions. By capturing images of the calibration plates and processing them, a mapping relationship between pixel coordinates and world coordinates can be established. This allows for the calculation of the camera's intrinsic and extrinsic parameters, resulting in calibrated intrinsic and extrinsic parameters. This accurately describes the camera's imaging geometry and provides a basis for coordinate correction in subsequent measurements.

[0038] Understandably, this embodiment effectively eliminates measurement errors caused by lens distortion through camera calibration, ensuring the accuracy of subsequent model distance calculations. Furthermore, the introduction of the calibration plate provides a traceable benchmark for the measurement results, giving the entire measurement process a verifiable accuracy basis.

[0039] S104. Select key feature lines in the three-dimensional model, and calculate the model distance of the key feature lines using the calibrated intrinsic and extrinsic parameters.

[0040] Key feature segments refer to important line segments that characterize the geometric dimensions and assembly features of hull sections, such as the side length of mating surfaces, rib spacing, and longitudinal bone length. For example, such as... Figure 4 The diagram shown is a schematic representation of the key feature line segment, which is... Figure 3 Select line segments OA, OB, OC, CM, FG, and FL from the 3D model.

[0041] Model distance refers to the three-dimensional spatial length of key feature line segments in a three-dimensional model.

[0042] Specifically, key feature lines are selected in the 3D model, and the endpoint coordinates of the key feature lines are corrected using calibrated intrinsic and extrinsic parameters to eliminate measurement errors caused by lens distortion, thereby obtaining accurate model distances.

[0043] For example, for each selected key feature line segment, the spatial coordinates (x, y, z) of its two endpoints in the 3D model are obtained. 10 , y 10 , z 10 ) and (x 20 , y 20 , z 20When calculating the model distance, the endpoint coordinates are corrected using the camera intrinsic and extrinsic parameters obtained from step three calibration to eliminate errors caused by lens distortion. Then, based on the corrected coordinates (x1, y1, z1) and (x2, y2, z2), the model distance is calculated using the spatial distance formula: Model distance = .

[0044] S105. Perform multi-dimensional error analysis on the model distance and the actual measured geometric parameters of the target hull segment. Adjust the UAV shooting parameters and / or the parameters of the three-dimensional model according to the error analysis results and perform iterative optimization until the measurement accuracy meets the preset requirements.

[0045] Among them, the actual measured geometric parameters refer to the length values ​​obtained by on-site measurement of the same set of key feature line segments using traditional measurement methods, such as a calibrated steel tape measure or a high-precision total station.

[0046] Multidimensional error analysis can include absolute error analysis, relative error analysis, or spatial location error distribution analysis, etc.

[0047] The preset requirements can be set according to the actual application scenario, such as absolute error ≤3mm, relative error ≤0.1%, horizontal spatial position error ≤2mm and vertical spatial position error ≤2mm, etc.

[0048] Drone photography parameters include various adjustable parameters related to image acquisition, such as flight altitude, flight speed, image overlap rate, and camera shutter speed; 3D model parameters include various adjustable parameters related to the modeling process, such as feature point extraction threshold and point cloud density level.

[0049] Specifically, based on the error analysis results, at least one of the parameters of the UAV shooting and the 3D model is adjusted and iteratively optimized. By adjusting the parameters of the UAV shooting and the 3D model, the image acquisition quality and modeling accuracy can be improved in a targeted manner, thereby reducing measurement errors. After each parameter adjustment, camera calibration, key feature line segment model distance calculation, and multi-dimensional error analysis are re-executed, and the results are compared with the measured values ​​again. This process is repeated until the measurement accuracy meets the preset requirements. The absolute geometric parameters of the hull sections are obtained, and their accuracy is verified. This embodiment establishes a complete closed-loop feedback mechanism, making the measurement accuracy verifiable, controllable, and optimizable. It achieves efficient, accurate, and verifiable measurement of the absolute geometric parameters of the hull sections, ensuring the reliability of the final measurement results and meeting the high-precision evaluation requirements of the section geometric parameters in the early stages of ship construction.

[0050] In summary, the method for measuring the geometric parameters of ship hull sections provided in this embodiment of the invention improves the efficiency of measuring the geometric parameters of ship hull sections by using a camera mounted on an unmanned aerial vehicle to acquire multiple images of the target ship hull section from multiple perspectives. Based on the images of the ship hull sections, key feature points of the target ship hull sections are identified, and a three-dimensional model of the target ship hull section is constructed based on the key feature points, which improves the representation accuracy of the three-dimensional model in key areas, thereby ensuring the accuracy of subsequent measurements. Meanwhile, the automated modeling process significantly reduces manual intervention and improves modeling efficiency. Based on multiple calibration plates placed around the target hull section, the camera is calibrated to obtain its intrinsic and extrinsic parameters, thus accurately describing the camera's imaging geometry and providing a basis for coordinate correction in subsequent measurements. Key feature segments are selected in the 3D model, and the model distance of these segments is calculated using the calibrated intrinsic and extrinsic parameters. The endpoint coordinates of these key feature segments are then corrected using the calibrated intrinsic and extrinsic parameters to eliminate measurement errors caused by lens distortion, thereby obtaining accurate model distances. Multi-dimensional error analysis is performed between the model distances and the actual measured geometric parameters of the target hull section. Based on the error analysis results, the UAV shooting parameters and / or 3D model parameters are adjusted iteratively until the measurement accuracy meets the preset requirements, achieving efficient, accurate, and verifiable measurement of the absolute geometric parameters of the hull section.

[0051] In some embodiments of the present invention, the key feature points include feature points located at weld seams, butt joint edges, and structural stiffeners; step S102 includes: extracting image features from the multiple hull segment images using a feature matching algorithm, the feature matching algorithm including a scale-invariant feature transform algorithm or an ORB feature matching algorithm; based on the image features, identifying feature points located at weld seams, butt joint edges, and structural stiffeners on the target hull segment as the key feature points.

[0052] Among them, the Scale-Invariant Feature Transform (SIFT) algorithm locates key points in the image by performing scale space extremum detection and generates feature descriptors with rotation and scale invariance. It can maintain stable matching performance even when the image is rotated, scaled or even changes in illumination.

[0053] The ORB (Oriented FAST and Rotated BRIEF) feature matching algorithm combines an improved FAST corner detection with a rotation-aware BRIEF descriptor. It features fast computation speed and good real-time performance, making it suitable for applications with high processing efficiency requirements.

[0054] Specifically, a feature matching algorithm is used to extract image features from multiple images of the hull segments. Then, based on these extracted features, feature points located at specific locations on the target hull segment are identified, namely, feature points at weld seams, butt joint edges, and structural stiffeners, thus obtaining key feature points. Identifying key feature points using feature transformation algorithms or ORB feature matching algorithms avoids the subjective errors and inefficiencies of traditional manual marking methods, making the identification process more objective, accurate, and efficient. This improves the accuracy and efficiency of key feature point identification, thereby enhancing the focus on critical parts of the hull segments during 3D modeling and ensuring that key geometric parameters on the hull segments can be accurately obtained and verified.

[0055] In some embodiments of the present invention, step S102 includes: performing encrypted feature point matching on the segment regions where the key feature points are located in the multiple hull segment images to obtain feature matching information of the segment regions; extracting depth information from the multiple hull segment images based on a multi-view stereo vision algorithm to obtain corresponding segment depth information, and increasing the point cloud density in the region near the key feature points according to the feature matching information; performing dense point cloud sampling on the local image region containing the key feature points to obtain dense point cloud data of the local image region; generating high-density point cloud data covering the surface of the target hull segment based on the segment depth information; and performing three-dimensional reconstruction on the high-density point cloud data to generate a three-dimensional model of the target hull segment.

[0056] Multi-View Stereo (MVS) is a technique for recovering the 3D structure of a scene from multiple viewpoints. Its basic principle is to calculate the depth value of the spatial point corresponding to each pixel by matching pixels with the same name in different images and using triangulation. In this embodiment, the algorithm can be implemented using relevant functions from open-source computer vision libraries (such as OpenCV) or built-in algorithm modules in commercial modeling software (such as Context Capture).

[0057] 3D reconstruction refers to the process of converting discrete point cloud data into a continuous triangular mesh surface model. By connecting adjacent points to form triangular facets, the geometry of the object's surface is constructed. In this embodiment, the 3D reconstruction process can be implemented using common algorithms such as Poisson surface reconstruction, rolling sphere algorithm, or Delaunay triangulation.

[0058] Encrypted feature point matching can target the local region where key feature points are located, i.e., segmented region. By adjusting the parameters of the matching algorithm or adopting a denser sampling strategy, the number of matching feature points in the segmented region can be increased. The encrypted feature matching information provides more reliable geometric constraints for subsequent point cloud generation, enabling the spatial location of key regions to be determined more accurately.

[0059] Specifically, encrypted feature point matching is performed on segmented regions to obtain richer feature matching information for each segment. Depth information is extracted from multiple hull segment images using a multi-view stereo vision algorithm to obtain corresponding segment depth information. During depth information extraction, the point cloud sampling density is increased in areas near key feature points based on the feature matching information obtained from encrypted feature matching. This means that more points will be generated at key locations such as welds and mating surface edges, resulting in a denser point cloud distribution. Dense point cloud sampling is performed on local image regions containing key feature points to obtain dense point cloud data for these local image regions, enhancing the point cloud density in key areas and enabling a more accurate reflection of the actual geometric shape of the hull segment. Finally, 3D reconstruction is performed on the high-density point cloud data to generate a 3D model of the target hull segment. Because the point cloud data has a higher density in key areas, the reconstructed 3D model has higher resolution and geometric accuracy in these areas, accurately restoring the location of welds, the contour of mating surface edges, and the shape of structural stiffeners. This improves modeling accuracy and helps ensure that the measurement results truly reflect the actual geometric state of the hull segment.

[0060] In some embodiments of the present invention, after step S102, the method further includes: performing template matching according to a preset hull segment feature library to identify the segment type of the target hull segment; and, according to the segment type, adopting an adaptive resolution modeling strategy to perform high-resolution modeling on the key areas of the docking surface of the corresponding hull segment in the three-dimensional model, and standard resolution modeling on other areas.

[0061] The hull section feature library records geometric feature templates for common section types used in shipbuilding. Hull section types mainly include, but are not limited to: double-bottom structures, side structures, deck structures, and bulkhead structures. Each section type corresponds to a specific geometric configuration, structural features, and key area distribution pattern. For example, a double-bottom structure has typical features such as a bottom plate, inner bottom, longitudinal ribs, and ribs, and its key areas typically include the mating edges between the bottom plate and the side, and the area around the longitudinal rib through-holes.

[0062] The adaptive resolution modeling strategy adopts differentiated modeling resolutions based on the importance of different areas of the hull section in the subsequent assembly process.

[0063] Specifically, template matching is performed based on a pre-defined hull segment feature library to identify the segment type of the target hull segment. Then, an adaptive resolution modeling strategy is adopted to perform high-resolution modeling (e.g., resolution ≤ 5mm) on key areas of the corresponding hull segment docking surfaces in the 3D model, while standard resolution modeling is used for other non-critical areas. 3D reconstruction and texture mapping are then performed to complete the construction of the 3D model of the target hull segment. This approach ensures the modeling accuracy of key areas while maintaining overall modeling efficiency.

[0064] In some embodiments of the present invention, step S103 includes: establishing a mapping relationship between the pixel coordinates of the feature marker points on the calibration board and the world coordinates using the Zhang Zhengyou calibration method; and solving the intrinsic and extrinsic parameters of the camera based on the mapping relationship. The intrinsic parameters include focal length, principal point, and distortion coefficient, and the extrinsic parameters include rotation matrix and translation vector.

[0065] Among them, Zhang Zhengyou's calibration method is a classic camera calibration method. It solves for camera parameters by taking images of the calibration board from different angles and using the spatial constraint relationship of the checkerboard corner points.

[0066] Specifically, the Zhang Zhengyou calibration method is used to establish the mapping relationship between the pixel coordinates of the feature marker points on the calibration board and the world coordinates. Based on this mapping relationship, the intrinsic and extrinsic parameters of the camera are solved, effectively eliminating the influence of lens distortion on measurement accuracy.

[0067] In some embodiments of the present invention, the error analysis results include at least: absolute error, relative error, or spatial position error distribution; step S104 includes: calculating the difference between the model distance and the actual measured geometric parameters to obtain the absolute error; calculating the percentage of the absolute error to the actual measured geometric parameters to obtain the relative error; and for each key feature line segment, calculating its error components in the X, Y, and Z directions to obtain the spatial position error distribution.

[0068] Specifically, for each key feature line segment, the absolute error is calculated using the following formula: Absolute error = Model distance - Actual measured geometric parameters; For each key feature line segment, the relative error is calculated using the following formula: Relative error = (Absolute error / Actual measured geometric parameters) × 100%; For each key feature line segment, obtain the spatial coordinates of its two endpoints in the 3D model, as well as the spatial coordinates of the corresponding points in the actual measurement, and then calculate the error components in the X, Y and Z directions respectively.

[0069] In one specific implementation, for Figure 3Field measurements were conducted on line segments OA, OB, OC, CM, FG, and FL. The error analysis results are shown in Table 1. The absolute error of each line segment is between -1.7 mm and 0 mm. This demonstrates that the geometric parameter measurement method of this application has high accuracy.

[0070]

[0071] In some embodiments of the present invention, the error analysis results include at least: absolute error, relative error, or spatial position error distribution; the UAV shooting parameters include UAV flight parameters and camera parameters; the 3D modeling parameters include feature point extraction threshold, image matching accuracy level, and dense point cloud reconstruction density; the step of performing multi-dimensional error analysis on the model distance and the actual measured geometric parameters of the target hull segment, and iteratively optimizing the UAV shooting parameters and / or the 3D model parameters based on the error analysis results, includes: when the absolute error exceeds a first preset threshold, recalibrating the intrinsic parameters of the camera, and simultaneously adjusting the UAV... The system adjusts the flight altitude and camera shutter speed, and lowers the feature point extraction threshold. When the relative error exceeds a second preset threshold, it adjusts the image overlap rate, UAV flight speed, and the number and distribution of ground control points, and increases the density of the dense point cloud reconstruction. When the spatial position error exceeds a third preset threshold in the horizontal direction, it increases the tilt shooting route and adjusts the camera sensitivity. When the spatial position error exceeds a fourth preset threshold in the vertical direction, it supplements the image with elevation and depression angles and adjusts the dense reconstruction parameters in the vertical direction. After each adjustment, the camera is recalibrated, the distance of the key feature line segment model is calculated, and multi-dimensional error analysis is performed until the measurement accuracy meets the preset requirements.

[0072] Specifically, when the absolute error exceeds a first preset threshold (e.g., 3mm), it indicates a systematic deviation in the overall measurement accuracy. In this case, the camera's intrinsic parameters are recalibrated first to eliminate any possible calibration errors; simultaneously, the UAV's flight altitude is adjusted (e.g., reduced by 10% to 20%) to improve the ground resolution of the image; the camera shutter speed is adjusted to no less than 1 / 1000s to reduce motion blur caused by flight vibration; and the feature point extraction threshold in the modeling parameters is reduced (e.g., reduced by 20%) to increase the number of effective matching points and improve the accuracy of point cloud reconstruction.

[0073] When the relative error exceeds the second preset threshold (e.g., 0.1%), it indicates a problem with the scale consistency between the model and the actual object. At this point, adjust the image acquisition parameters: increase the forward image overlap rate to no less than 80% and the lateral overlap rate to no less than 70% to enhance the geometric constraints between images; reduce the UAV flight speed by 10% to 15% to ensure image clarity; increase the number of ground control points to no less than 6 and optimize their spatial distribution so that the control points evenly cover each area of ​​the hull segment; and increase the dense point cloud reconstruction density level in the modeling parameters (e.g., adopt a high-density mode) to improve the overall scale consistency.

[0074] When the spatial position error in the horizontal direction exceeds the third preset threshold (e.g., 2mm), it indicates insufficient measurement accuracy in the X or Y direction. In this case, a 45° oblique shooting path is added to the original path to enrich the geometric constraints in the horizontal direction; simultaneously, the camera's ISO sensitivity is adjusted to ensure the imaging quality of the obliquely shot images. When the spatial position error in the vertical direction exceeds the fourth preset threshold (e.g., 2mm), it indicates insufficient height measurement accuracy in the Z direction. In this case, close-range elevation and depression angle images are supplemented, and the camera's elevation angle is adjusted to within the range of ±15° to ±30°; the iteration number and density level of the dense reconstruction in the Z direction are increased separately in the modeling parameters to improve the measurement accuracy in the height direction.

[0075] The parameter adjustments in the above three aspects need to be performed simultaneously. After each adjustment, camera calibration, key feature line segment model distance calculation, and multi-dimensional error analysis should be re-executed until the absolute error, relative error, and spatial position errors in the X, Y, and Z directions simultaneously meet the preset measurement accuracy requirements. In this embodiment, the preset requirements are set as absolute error ≤ 3mm, relative error ≤ 0.1%, and spatial position errors in each direction ≤ 2mm. This accuracy index can meet the geometric parameter evaluation needs in the early stage of hull section construction.

[0076] Through the multi-dimensional error analysis and iterative parameter optimization described above, this embodiment achieves closed-loop control of measurement accuracy. Each iteration is based on the error analysis results of the previous stage, and relevant parameters are adjusted in a targeted manner to gradually improve the measurement accuracy until the requirements are met. This dynamic adjustment mechanism ensures that this method can obtain reliable measurement results under different measurement environments and conditions, effectively solving the technical problems of uncontrollable and unverifiable accuracy in traditional UAV measurement methods.

[0077] In some embodiments of the present invention, before step S101, the method further includes: selecting the take-off point of the UAV based on the hull layout, environmental safety, wind force and climate factors of the target hull segment; and / or setting the image overlap rate when the UAV takes pictures according to the modeling accuracy requirements; and / or formulating standardized data naming rules corresponding to the multiple hull segment images collected, wherein the naming rules include the shooting date, segment number, data type and shooting angle information.

[0078] Specifically, in actual operation, it is first necessary to formulate UAV operation specifications based on the site environment. In this embodiment, considering the environmental layout of the laboratory ship pool and the size of the model, the UAV takeoff point is selected in a stable, unobstructed area more than one meter away from the model to ensure safe takeoff and landing. Based on the modeling accuracy requirements, the image overlap rate is set to 80%, meaning that the overlap area between adjacent images reaches 80% in both the heading and lateral directions. This helps with feature matching and model integrity during subsequent 3D modeling. Simultaneously, to facilitate subsequent data management and traceability, standardized data naming rules are established, with the naming format being "Shooting Date_Segment Number_Data Type_Shooting Angle", for example, "20231015_Double Bottom_Image_Front View". This allows for the acquisition of multiple comprehensive images of the ship's hull segments.

[0079] In some embodiments of the present invention, step S101 includes: dynamically adjusting the flight altitude of the UAV according to the shooting surface position height of the target hull segment; and / or setting the flight speed of the UAV to a preset speed.

[0080] The preset speed can be 10cm / s.

[0081] Specifically, the drone's flight altitude is dynamically adjusted according to the position of the shooting surface, such as setting the flight speed to 10cm / s to ensure image clarity.

[0082] As an optional implementation, the drone can also be equipped with a LiDAR sensor to improve measurement accuracy and stability in complex environments through multi-source data fusion.

[0083] To better implement the hull section geometric parameter measurement method in this embodiment of the invention, based on the hull section geometric parameter measurement method, correspondingly, as follows: Figure 5 As shown, this embodiment of the invention also provides a hull section geometric parameter measuring device, the hull section geometric parameter measuring device 500 comprising: The acquisition unit 501 is used to acquire multi-view images of the target ship hull section using a camera mounted on a drone, and obtain multiple images of the ship hull section. Modeling unit 502 is used to identify key feature points of the target hull segment based on the hull segment image, and to construct a three-dimensional model of the target hull segment based on the key feature points; The calibration unit 503 is used to calibrate the camera based on multiple calibration plates arranged around the target hull section to obtain the camera's intrinsic and extrinsic parameters. The calculation unit 504 is used to select key feature line segments in the three-dimensional model and calculate the model distance of the key feature line segments using calibrated intrinsic and extrinsic parameters. The verification and optimization unit 505 is used to perform multi-dimensional error analysis on the model distance and the actual measured geometric parameters of the target hull segment, and to iteratively optimize the UAV shooting parameters and / or the parameters of the three-dimensional model according to the error analysis results until the measurement accuracy meets the preset requirements.

[0084] The hull section geometric parameter measuring device 500 provided in the above embodiments can realize the technical solutions described in the above embodiments of the hull section geometric parameter measuring method. The specific implementation principles of each module or unit can be found in the corresponding content in the above embodiments of the hull section geometric parameter measuring method, which will not be repeated here.

[0085] Accordingly, this application also provides a computer-readable storage medium for storing computer-readable programs or instructions. When the programs or instructions are executed by a processor, they can implement the steps or functions in the hull section geometric parameter measurement methods provided in the above-described method embodiments.

[0086] Those skilled in the art will understand that all or part of the processes of the methods described in the above embodiments can be implemented by a computer program instructing related hardware (such as a processor, controller, etc.), and the computer program can be stored in a computer-readable storage medium. The computer-readable storage medium may be a disk, optical disk, read-only memory, or random access memory, etc.

[0087] The foregoing has provided a detailed description of the method, apparatus, and storage medium for measuring the geometric parameters of ship sections provided by the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, those skilled in the art will recognize that there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for measuring the geometric parameters of ship hull sections, characterized in that, include: Multiple images of the target ship hull sections were obtained by using a camera mounted on a drone to capture images from multiple perspectives. Based on the image of the hull segment, key feature points of the target hull segment are identified, and a three-dimensional model of the target hull segment is constructed based on the key feature points; The camera is calibrated using multiple calibration plates arranged around the target hull section to obtain the camera's intrinsic and extrinsic parameters. Key feature lines are selected in the three-dimensional model, and the model distance of the key feature lines is calculated using the calibrated intrinsic and extrinsic parameters. A multi-dimensional error analysis is performed between the model distance and the actual measured geometric parameters of the target hull segment. Based on the error analysis results, the UAV shooting parameters and / or the parameters of the three-dimensional model are adjusted and iteratively optimized until the measurement accuracy meets the preset requirements.

2. The method for measuring the geometric parameters of hull sections according to claim 1, characterized in that, The key feature points include those located at weld seams, butt joint edges, and structural stiffeners; the identification of key feature points of the target hull segment based on the hull segment image includes: Image features are extracted from the multiple images of the ship hull segments using a feature matching algorithm, which includes a scale-invariant feature transform algorithm or an ORB feature matching algorithm. Based on the image features, feature points located at weld seams, butt joint edges, and structural reinforcing ribs on the target hull section are identified as key feature points.

3. The method for measuring the geometric parameters of hull sections according to claim 1, characterized in that, The construction of the three-dimensional model of the target hull segment based on the key feature points includes: Encrypted feature point matching is performed on the segmented regions where the key feature points are located in the multiple hull segment images to obtain the feature matching information of the segmented regions. The depth information of the multiple ship hull segment images is extracted based on the multi-view stereo vision algorithm to obtain the corresponding segment depth information, and the point cloud density is increased in the area near the key feature points according to the feature matching information. Dense point cloud sampling is performed on the local image region containing the key feature points to obtain the dense point cloud data of the local image region; Based on the segment depth information, high-density point cloud data covering the surface of the target hull segment is generated; The high-density point cloud data is used to perform three-dimensional reconstruction to generate a three-dimensional model of the target ship hull segment.

4. The method for measuring the geometric parameters of hull sections according to claim 3, characterized in that, After constructing a 3D model of the target hull segment based on the key feature points, the process also includes: Template matching is performed based on a preset hull segment feature library to identify the segment type of the target hull segment; Based on the segment type, an adaptive resolution modeling strategy is adopted to perform high-resolution modeling on the key areas of the docking surfaces of the corresponding hull segments in the 3D model, and standard resolution modeling on other areas.

5. The method for measuring the geometric parameters of hull sections according to claim 1, characterized in that, The calibration of the camera includes: The mapping relationship between the pixel coordinates and world coordinates of the feature marker points on the calibration board is established using the Zhang Zhengyou calibration method. The intrinsic and extrinsic parameters of the camera are solved based on the mapping relationship. The intrinsic parameters include focal length, principal point, and distortion coefficients, and the extrinsic parameters include rotation matrix and translation vector.

6. The method for measuring the geometric parameters of hull sections according to claim 1, characterized in that, The error analysis results include at least: absolute error, relative error, or spatial position error distribution; the multi-dimensional error analysis of the model distance and the actual measured geometric parameters of the target hull section includes: The absolute error is obtained by calculating the difference between the model distance and the actual measured geometric parameters. The relative error is obtained by calculating the percentage of the absolute error to the actual measured geometric parameter; For each key feature line segment, its error components in the X, Y, and Z directions are calculated to obtain the spatial position error distribution.

7. The method for measuring the geometric parameters of hull sections according to claim 1, characterized in that, The error analysis results include at least: absolute error, relative error, or spatial position error distribution; the UAV shooting parameters include UAV flight parameters and camera parameters; the 3D modeling parameters include feature point extraction threshold, image matching accuracy level, and dense point cloud reconstruction density; the step of performing multi-dimensional error analysis on the model distance and the actual measured geometric parameters of the target hull segment, and iteratively optimizing the UAV shooting parameters and / or the 3D model parameters based on the error analysis results, includes: When the absolute error exceeds the first preset threshold, the camera's intrinsic parameters are recalibrated, and the drone's flight altitude and the camera's shutter speed are adjusted, while the feature point extraction threshold is reduced. When the relative error exceeds the second preset threshold, adjust the image overlap rate, the UAV flight speed, and the number and distribution of ground control points, and increase the reconstruction density of the dense point cloud; When the spatial position error exceeds the third preset threshold in the horizontal direction, increase the tilt shooting route and adjust the camera sensitivity; When the spatial position error in the vertical direction exceeds the fourth preset threshold, supplement the image by taking images at the elevation angle and adjust the density reconstruction parameters in the vertical direction; After each adjustment, the camera is recalibrated, the distance of key feature line segments is calculated, and multi-dimensional error analysis is performed until the measurement accuracy meets the preset requirements.

8. The method for measuring the geometric parameters of hull sections according to claim 1, characterized in that, Before using a camera mounted on a drone to acquire multi-view images of the target ship hull sections, the process also includes: The takeoff point of the UAV is selected based on the hull layout of the target hull sections, environmental safety, wind force, and weather factors; and / or, Set the image overlap rate during drone shooting according to the modeling accuracy requirements; and / or, Establish standardized data naming rules corresponding to the multiple images of ship hull segments collected. The naming rules include the shooting date, segment number, data type, and shooting angle information.

9. The method for measuring the geometric parameters of hull sections according to claim 1, characterized in that, The method of acquiring multi-view images of target ship sections using cameras mounted on drones includes: The drone's flight altitude is dynamically adjusted based on the position and height of the shooting surface of the target hull segment; and / or, The drone's flight speed is set to a preset speed.

10. A device for measuring the geometric parameters of a ship hull section, characterized in that, include: The acquisition unit is used to acquire multi-view images of the target ship hull sections using a camera mounted on a drone, and obtain multiple images of the ship hull sections. The modeling unit is used to identify key feature points of the target hull segment based on the hull segment image, and to construct a three-dimensional model of the target hull segment based on the key feature points; The calibration unit is used to calibrate the camera based on multiple calibration plates arranged around the target hull section to obtain the camera's intrinsic and extrinsic parameters. The calculation unit is used to select key feature line segments in the three-dimensional model and calculate the model distance of the key feature line segments using calibrated intrinsic and extrinsic parameters. The verification and optimization unit is used to perform multi-dimensional error analysis on the model distance and the actual measured geometric parameters of the target hull segment, and to iteratively optimize the UAV shooting parameters and / or the parameters of the three-dimensional model according to the error analysis results until the measurement accuracy meets the preset requirements.

Images