Automatic 3D modeling method, system and device based on 2D ship CAD drawings

Abstract

The present invention relates to the field of ship design, and in particular to an automatic three-dimensional modeling method, system and device based on a two-dimensional ship CAD drawing. The method includes obtaining and parsing ship CAD drawing data, extracting metadata from the drawing, and identifying primitives; generating images through rendering technology for unidentified complex primitive areas, and performing preprocessing; inputting the preprocessed images into an object detection model, generating detection frames and category annotations using a deep learning algorithm, and performing coordinate conversion to accurately calibrate the positions of primitives; extracting geometric information, structural attributes and material data based on object detection results and primitive information, performing parametric expression conversion, and generating parametric model data of the ship structure; inputting the generated parametric data into a modeling instruction template, generating a ship CAD modeling instruction sequence, and executing the instructions through a CAD system to automatically construct a three-dimensional model of the ship. The present invention can realize the fully automated conversion from two-dimensional CAD drawings to three-dimensional modeling, thereby improving the efficiency and accuracy of ship design.

Description

Technical Field

[0001] The present invention relates to the field of ship design, and in particular to an automatic three-dimensional modeling method, system and device based on a two-dimensional drawing of a ship CAD. Background Art

[0002] In the traditional ship design process, designers usually rely on 2D CAD drawings to design and layout ship components. 2D drawings contain rich geometric information and structural details, but this information is usually only applicable to plane design and cannot be directly converted into 3D models, which requires designers to do a lot of additional manual modeling work, which is time-consuming and error-prone. In order to improve design efficiency and accuracy, 3D modeling has become an important means of ship design, but current 3D modeling still faces many challenges.

[0003] In the prior art, although there are some methods for automatically generating three-dimensional models based on CAD drawings, most of these methods rely on manual operation or require designers to manually input a large number of geometric parameters and physical properties. Especially in the field of ship design, due to the complex structure of the ship and the close connection between the various components, the prior art often cannot fully and automatically extract all necessary primitive information from the two-dimensional drawings, nor can it effectively realize the automatic generation and execution of three-dimensional modeling instructions. Therefore, how to automatically extract primitive information and generate accurate three-dimensional modeling instructions based on the two-dimensional CAD drawings of ships, and realize the automatic construction of three-dimensional models of ships, is still a technical problem that needs to be solved in the field of ship design. Summary of the invention

[0004] The present invention provides an automatic three-dimensional modeling method, system and device based on two-dimensional drawings of ship CAD, so as to solve the problem of how to automatically extract primitive information and generate three-dimensional modeling instructions based on the two-dimensional drawings of ship CAD, realize automatic construction of three-dimensional model of ship in CAD system, and improve the efficiency and accuracy of ship design.

[0005] In order to solve the above technical problems, the present invention provides an automatic three-dimensional modeling method based on a two-dimensional drawing of a ship CAD, comprising:

[0006] Obtain the input ship standard CAD drawings, perform data analysis, extract metadata from the drawings, and conduct preliminary analysis of the drawings based on ship design specifications to identify graphic elements;

[0007] For unrecognized complex primitive areas, images are generated through rendering technology, and the generated images are preprocessed;

[0008] The rendered image is input into the deployed object detection model, and the detection box and category annotation are generated using the deep learning algorithm. The coordinates are converted according to the drawing coordinate system and the image coordinate system to accurately calibrate the position of the primitives.

[0009] Based on the object detection results and primitive information, relevant geometric information, structural attributes and material data are extracted, parametric expression conversion is performed, and parametric model data of the ship structure is generated;

[0010] Input the generated ship structure parametric data into the modeling instruction template, automatically match and generate the CAD modeling parametric instruction sequence;

[0011] The generated modeling instruction sequence is input into the CAD system to drive the CAD software to execute the modeling instructions, automatically construct the three-dimensional model of the ship and complete the modeling process.

[0012] Furthermore, the automatic three-dimensional modeling method based on the two-dimensional ship CAD drawing, the step of obtaining the input ship standard CAD drawing also includes:

[0013] Parse ship CAD drawing data through open source parsing library or proprietary parsing module to extract geometric information, lines, surfaces, circles, and text element information in the drawings.

[0014] Furthermore, in the automatic three-dimensional modeling method based on the two-dimensional drawings of ship CAD, the data parsing step specifically includes:

[0015] The metadata in the drawings is cleaned and standardized, redundant data is eliminated, and valid graphic element information is extracted.

[0016] Furthermore, in the automatic three-dimensional modeling method based on the two-dimensional drawings of ship CAD, the preprocessing of the rendered image includes:

[0017] Perform image enhancement and smoothing on rendered images.

[0018] Furthermore, in the automatic three-dimensional modeling method based on the two-dimensional drawings of ship CAD, the deep learning algorithm of the object detection model specifically includes:

[0019] Use YOLO and Faster R-CNN neural network models to detect objects in images and generate detection boxes and category labels.

[0020] Furthermore, in the automatic three-dimensional modeling method based on the two-dimensional drawings of ship CAD, the coordinate transformation specifically includes:

[0021] Match the detection box coordinates in the image coordinate system with the CAD drawing coordinate system.

[0022] Furthermore, the automatic three-dimensional modeling method based on the two-dimensional drawings of ship CAD, the step of extracting geometric information, structural attributes and material data specifically includes:

[0023] The geometric data, material properties and structural parameters of the ship structure are extracted from the graphic element information, the size, position, material and other data of the graphic element are calculated, and converted into a parametric form.

[0024] Furthermore, in the automatic three-dimensional modeling method based on the two-dimensional drawings of a ship CAD, the step of generating a modeling instruction sequence specifically includes:

[0025] According to the ship design process and parametric data, a suitable modeling instruction template is selected, and the extracted geometry, structure, and material data are matched with the template to generate a complete modeling instruction sequence.

[0026] Furthermore, an automatic three-dimensional modeling system based on a two-dimensional ship CAD drawing comprises:

[0027] Data acquisition module, used to obtain and analyze ship standard CAD drawing data and extract graphic element information;

[0028] Data processing module, used to clean and standardize image metadata and extract valid information;

[0029] Graphics element recognition and analysis module, used to recognize graphics elements and analyze ship structures according to ship design specifications;

[0030] Object detection and image generation module, used to process complex primitive areas and generate accurate primitive data through rendering and object detection;

[0031] Parametric modeling module, used to convert the extracted geometric information, structural properties and material data into parametric model data;

[0032] A modeling instruction generation module, used for matching parameterized data with a modeling instruction template to generate a modeling instruction sequence;

[0033] CAD modeling execution module, used to input modeling instruction sequences into the CAD system to automatically build a three-dimensional model of the ship;

[0034] Furthermore, an automatic three-dimensional modeling device based on a two-dimensional drawing of a ship CAD comprises:

[0035] A processor is used to execute instructions stored in the memory to complete tasks such as data analysis, primitive recognition, modeling instruction generation and 3D modeling;

[0036] A memory, used to store ship CAD two-dimensional drawing data, parametric modeling data and generated modeling instruction sequences;

[0037] Communication interface, used for data interaction with external devices and drawing data transmission;

[0038] The display is used to display the generated three-dimensional ship model and related data.

[0039] The key innovative features of the present invention include:

[0040] (1) Automated primitive analysis and recognition: The present invention automatically extracts geometric information, structural attributes and material data from drawings through data analysis and primitive recognition modules, eliminating the errors and complexity of manual recognition and improving the efficiency and accuracy of model construction.

[0041] (2) Deep learning object detection and image generation: By using deep learning algorithms (such as YOLO and Faster R-CNN) and object detection models, it is possible to automatically detect and annotate complex primitive areas and accurately calibrate their locations, further enhancing the accuracy of primitive recognition.

[0042] (3) Parametric modeling and modeling instruction generation: The present invention parametrically expresses the extracted geometric information, structural properties and material data, matches them with the modeling instruction template, and automatically generates a modeling instruction sequence, thereby greatly improving the automation level and modeling speed of three-dimensional modeling.

[0043] The automatic three-dimensional modeling method, system and device based on the two-dimensional drawings of ship CAD provided by the present invention can efficiently extract the primitive information from the two-dimensional drawings of ship CAD through an automated process, and generate three-dimensional modeling instructions, so as to realize the automatic construction of the three-dimensional model of the ship in the CAD system. Compared with the traditional manual modeling process, the automated method of the present invention can significantly reduce manual intervention and improve the efficiency and accuracy of ship design.

[0044] Through the method, especially with the support of the primitive parsing and object detection modules, complex primitive areas can be accurately identified, eliminating the errors and inconsistencies in traditional manual primitive identification, thereby ensuring the accuracy of the generated three-dimensional model. At the same time, the use of deep learning algorithms and object detection models can automatically identify key components in the image and accurately calibrate their positions, making the entire modeling process more intelligent and efficient.

[0045] In addition, the present invention converts the geometric information, structural properties and material data of the ship structure into parametric model data, and matches them with the modeling instruction template to generate a modeling instruction sequence, which not only reduces the human design deviation, but also improves the modeling speed and can adapt to complex and changeable ship design requirements. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 A schematic diagram of a process flow of an automatic three-dimensional modeling method based on a two-dimensional ship CAD drawing provided in an embodiment of the present application;

[0047] Figure 2A structural block diagram of an automatic three-dimensional modeling system based on a two-dimensional ship CAD drawing provided in an embodiment of the present application;

[0048] Figure 3 A structural block diagram of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0049] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of the present application; the terms used in the specification of the application herein are only for the purpose of describing specific embodiments and are not intended to limit the present application; the terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of the present application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0050] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0051] Example 1: Reference Figure 1 , is a flow chart of an automatic three-dimensional modeling method based on a two-dimensional ship CAD drawing provided by an embodiment of the present invention. The flow chart may at least include steps S100-S600:

[0052] S100, obtaining an input ship standard CAD drawing, performing data analysis, extracting metadata from the drawing, and performing a preliminary analysis on the drawing based on the ship design specification to identify graphic elements;

[0053] S200, for the unrecognized complex primitive area, generating an image by using a rendering technology, and preprocessing the generated image;

[0054] S300, input the rendered image into the deployed object detection model, use the deep learning algorithm to generate the detection frame and category annotation, and perform coordinate conversion according to the drawing coordinate system and the image coordinate system to accurately calibrate the position of the graphic element;

[0055] S400, based on the object detection results and the primitive information, extract relevant geometric information, structural attributes and material data, perform parametric expression conversion, and generate parametric model data of the ship structure;

[0056] S500, inputting the generated ship structure parametric data into a modeling instruction template, automatically matching and generating a CAD modeling parametric instruction sequence;

[0057] S600, input the generated modeling instruction sequence into the CAD system, drive the CAD software to execute the modeling instructions, automatically construct the three-dimensional model of the ship and complete the modeling process.

[0058] Step S100 at least includes steps S110-S130:

[0059] S110 obtains the input ship standard CAD drawing data and uses open source parsing libraries and / or proprietary parsing modules to parse out the geometric information, lines, surfaces, circles, and text elements in the drawings.

[0060] Specifically, the S110 module first obtains the input ship standard CAD drawing data. The drawing data is generally stored in a file format such as DXF, DWG, etc. The input data contains a variety of information, including the geometric elements of the drawing (such as lines, faces, circles, texts, etc.) and metadata associated with the primitives (such as size, symbol, number, etc.). In order to process this data, we can use open source parsing libraries (such as ezdxf, ODA, Li breDWG, etc.) or our own parsing modules to parse the various elements in the drawing.

[0061] During this process, the CAD parsing module extracts different elements of the input drawing. First, geometric information (including lines, faces, circles) and text elements (such as numbers, annotations, captions, etc.) are extracted independently. Geometric information includes structural information such as shape, size, angle, etc. in the drawing, while text elements contain descriptive information related to these geometric structures, such as material, size, number, etc.

[0062] Furthermore, the extracted primitive information is classified and identified through the CAD parsing module. Primitives refer to elements that represent specific structures or design content in CAD drawings, which may include different graphic types (line types, curves, annotations, symbols, etc.). At this stage, the primitive information has been identified, but has not yet been analyzed and matched in detail according to the ship design specifications.

[0063] In this step, the parsed geometric information, lines, faces, circles, and text elements are used as preliminary metadata to provide basic data for subsequent primitive recognition and analysis. This information is the basis for the next step, especially in the data cleaning in step S120 and the rule analysis in step S130.

[0064] S120 cleans the parsed CAD drawings, removes irrelevant and / or redundant graphic element data, and extracts valid graphic element information.

[0065] In step S120, the CAD drawing data parsed by S110 is first cleaned. The main purpose of this process is to identify and remove irrelevant or redundant metadata. For example, some drawings may contain extra decorative lines, blank areas, or annotations that are irrelevant to the ship structure. They will be identified and deleted to ensure that only metadata related to the ship structure is retained.

[0066] Specifically, during the data cleaning process, each primitive will be processed as follows:

[0067] Redundant data elimination: For example, repeated annotations or symbols. These redundant information will not affect the modeling process of ship design and need to be eliminated in this step.

[0068] Elimination of irrelevant entities: These entities are irrelevant to the ship structure, such as elements that are only used to illustrate the layout of the drawing. They will not affect subsequent modeling, so they also need to be deleted.

[0069] Next, the valid primitive information after data cleaning will be extracted and stored in a more concise and efficient format. In this process, the extracted primitive information includes all the meaningful geometric data and text data in the drawing, ensuring that the subsequent steps can be further processed based on this streamlined information.

[0070] In the process of S120, the effective primitive information obtained by cleaning will be used as input data in step S130 for subsequent analysis and recognition. At this point, the primitive information has been removed of redundant and irrelevant parts, ensuring that the subsequent data analysis is more accurate and efficient.

[0071] S130 performs preliminary analysis on the extracted metadata based on the ship design specifications, determines the ship structure and / or parts corresponding to the graphic elements through rules, matches the graphic element information with the ship design specifications, and identifies the specific ship components and / or design parameters corresponding to the graphic elements.

[0072] The S130 module is a key step to perform a preliminary analysis on the extracted graphic element data and match them through the ship design rules. Specifically, in this step, based on the extracted valid graphic element information, the graphic element is classified and identified through the set ship design specifications.

[0073] First, according to the ship design specifications, we can classify the primitives into several categories, such as ship deck contours, structural components, segment lines, etc. For example, a curve representing the hull shape may represent the outer contour of the ship's deck, and a notation symbol may be related to a specific part of the ship (such as the ship's ribs, plates, etc.). Based on these rules, the algorithm will gradually determine the ship component or part represented by each primitive.

[0074] Specifically, when making rule judgments, the following factors can be used:

[0075] Geometric features: such as size, shape, location, etc., help determine whether the element belongs to a specific ship component. For example, a circular element may represent a manhole.

[0076] Text information: Annotations, numbers, text descriptions, etc. in drawings provide more information related to the structure or part and help identify the type of element.

[0077] Spatial relationships: Different graphics elements may be related to each other. For example, two connected lines may represent the outline of a hatch. Spatial relationships are used to further identify components.

[0078] By applying these rules, the S130 module can match each element with the ship design specifications and identify the specific ship components or design parameters corresponding to the element. For example, a certain element may be identified as "hull external structure" while another element may be identified as "manhole structure". This information will provide the necessary basis for the subsequent modeling process.

[0079] In this process, the identified primitive information will be used as input data in the S200 module for further image rendering and object detection. Specifically, the output data generated by the S130 module - the correspondence between primitives and ship components, geometric data and design parameters - will provide important clues for subsequent steps, ensuring that the rendering and object detection stages can handle unrecognized areas more accurately.

[0080] Step S200 at least includes steps S210-S230:

[0081] S210: For a complex graphic element area that cannot be identified through preliminary parsing and analysis, use CAD software or a custom rendering module to generate an image view of the area and perform image rendering processing.

[0082] Specifically, the S210 module first extracts the complex primitive areas in the input CAD drawing that failed to pass the preliminary parsing and analysis. At this point, the previous S110 and S120 steps have completed the preliminary parsing and data cleaning of the drawing, but due to the high complexity of some primitives, or the presence of non-standardized and difficult-to-parse areas, these primitives cannot be fully identified during the parsing process.

[0083] To solve this problem, the S210 module will use CAD software (such as AutoCAD, AvevaMarine, etc.) or a custom rendering module to generate image views of these areas. The generation of image views includes the process of converting the primitive data into a two-dimensional image. Usually, the geometric information of the drawing (such as lines, arcs, text, etc.) is mapped to the rendered image through the mapping relationship between the coordinate system in the drawing and the image coordinate system.

[0084] Specifically, the rendering technology presents the geometric structure and annotation information of these complex areas as visual images, and during the processing, the proportions and geometric relationships of the drawings are maintained, thereby ensuring accuracy in the subsequent image processing stage.

[0085] While generating image views, the rendering module also processes the geometric features of the spatial layout, direction, size, etc. of the primitives to ensure that the rendering results are as close to the actual design requirements as possible. These rendered image views will provide the basis for subsequent image processing and input data for deep learning models.

[0086] S220, pre-processing the generated image, using image enhancement technology to improve image quality.

[0087] In step S220, we further process the rendered image generated in step S210 using image enhancement technology. The goal of image enhancement is to improve the image quality, making it clearer and more detailed, so that the subsequent object detection model (such as a deep learning model) can accurately identify the primitives in the image.

[0088] Specifically, image enhancement technology includes but is not limited to the following processing methods:

[0089] Contrast enhancement: Enhance the contrast of different areas in the image to make the edges of the primitives more obvious, facilitating subsequent contour recognition.

[0090] Sharpening: Sharpening technology is used to enhance the clarity of details in the image, making lines, edges and other geometric shapes more prominent, which is convenient for subsequent model recognition.

[0091] Image smoothing: For irregular edges in the image, smoothing is used to reduce the jagged effects that may occur during the rendering process.

[0092] Through these image enhancement techniques, the final image quality will be significantly improved, the details will be clearer, the image contrast will be improved, and the enhanced image will provide more accurate input data for the deep learning model.

[0093] S230, recording the location information of the unrecognized primitive area, and saving the preprocessed image in a standard format

[0094] In step S230, the location information of those unrecognized primitive areas must first be recorded. The location information of these areas is usually determined by the coordinate system of the CAD drawing, including the geometric position of the primitive (such as center point coordinates, boundary coordinates, etc.) and geometric properties such as size and shape of the primitive. Through the aforementioned rendering and preprocessing operations, the unrecognized primitive areas have been converted into visual images, so the spatial positions of these areas can be associated with the coordinate information in the original CAD drawing to ensure that these areas can be accurately located in subsequent operations.

[0095] Location information can be recorded in the following ways:

[0096] The coordinate information of each area is directly matched to the area in the image through the coordinate system data.

[0097] The actual position of each graphic element in the drawing is further determined through the image annotation information, ensuring that this information can be consistent with subsequent image analysis (such as object detection).

[0098] In addition, one of the key contents of step S230 is to save the preprocessed image in a standard format. The standard format here refers to a format that can widely support image processing and deep learning model input, such as JPEG, PNG, TIFF, etc. The saved image is not just the image itself, but also should include relevant location information and other metadata, so that it can be effectively connected and processed with other data in the subsequent stage.

[0099] After saving the preprocessed image in a standard format, the image will be passed as input to the next deep learning model for object detection and provide support for subsequent primitive recognition and 3D modeling.

[0100] Step S300 at least includes steps S310-S330:

[0101] S310: Input the preprocessed rendered image into the deployed object detection model, use the deep learning algorithm to perform object detection on the image, and generate corresponding detection boxes and category labels.

[0102] Specifically, step S310 first inputs the enhanced rendered image into an already deployed object detection model. The model usually uses a deep learning algorithm, such as a convolutional neural network (CNN) or other object detection algorithm, to identify specific primitives in the image. Since the unrecognized areas were rendered in the previous S200 module, these rendered images contain detailed visual information of complex primitive areas that cannot be directly parsed from the CAD drawings.

[0103] In this process, the object detection model uses deep learning algorithms to analyze the image and identify the various significant geometric elements in the image. Detection boxes and category labels are the main outputs generated by the model, among which:

[0104] Detection box: The location information of each primitive, usually expressed in the form of a rectangular box, which frames the target area;

[0105] Category annotation: The category or type information of the primitive, such as "watertight bulkhead", "segmentation line", "manhole", "reinforcement rib", etc., depends on the annotation information of the training dataset.

[0106] These inspection boxes and category annotations provide spatial distribution and structural information of the elements for subsequent steps, which helps to further analyze the relationship between the elements and map them back to the original CAD drawing coordinate system. Each inspection box represents a specific ship component or design element in the drawing, while the category annotation helps us understand the specific structure or part type represented by the area.

[0107] S320 , performing coordinate conversion according to the difference between the drawing coordinate system and the image coordinate system, and matching the coordinate information of the detection frame with the coordinate information of the original CAD drawing.

[0108] In step S320, the detection box and category annotation information generated in S310 need to be matched and mapped with the original CAD drawing coordinate system. Since the rendered image is generated in a two-dimensional plane, and the coordinates on the CAD drawing are based on the actual size and scale of the drawing, coordinate conversion is required to ensure that the precise positions of the detection box and the primitives in the drawing can correspond.

[0109] Specifically, the difference between the drawing coordinate system and the image coordinate system is usually manifested as the relationship between the coordinate unit on the drawing (such as millimeters or meters) and the pixel unit in the image coordinate system. In this process, based on the previous drawing coordinate system (such as the analysis results mentioned in S110 and S120), we convert the coordinate information in the rendered image into coordinates consistent with the CAD drawing coordinate system through a set of mapping relationships or transformation matrices.

[0110] During this mapping process, we need to transform the coordinates of each vertex of the detection box to ensure that it corresponds to the actual position on the CAD drawing. This coordinate transformation operation is not only a simple conversion of the difference between the image coordinate system and the drawing coordinate system, but also involves the precise position calibration of each element in the drawing to ensure the accuracy of subsequent model construction.

[0111] S330: further analyze and optimize the generated detection frame and category annotation information.

[0112] Finally, in step S330, the detection frame and category label information that have completed the coordinate transformation in S320 will be further analyzed and optimized. Specifically, the task of S330 is to ensure that the primitives represented by each detection frame are consistent with their category labels, and to perform necessary calibration and optimization based on the detection results to improve the accuracy of subsequent modeling and 3D reconstruction.

[0113] During this process, the optimized operations include:

[0114] Filter redundant frames: Sometimes, the object detection model may perform multiple detections on the same primitive, generating multiple overlapping detection frames. In this case, it is necessary to use methods such as the non-maximum suppression algorithm (NMS) to remove redundant frames and retain only the optimal detection frame.

[0115] Category label correction: Some complex areas may be mislabeled, that is, the model labels some elements as the wrong category. The category labels can be further corrected through manual inspection or using the rule system to ensure that the category labeling of each element is accurate.

[0116] Spatial relationship analysis: Further analyze the spatial relationship between the elements (such as alignment, relative position relationship, etc.) to ensure that their geometric layout on the drawing is consistent with the design specifications. If position deviation or unreasonable spatial relationship is found during the calibration process, correction is required.

[0117] Through these optimization steps, the final generated detection box and category annotation information can more accurately represent the design elements in the drawings, providing high-quality input data for subsequent ship structure parametric modeling and three-dimensional modeling.

[0118] Step S400 at least includes steps S410-S430:

[0119] S410: Based on the object detection result and the identified primitive information, the geometric data corresponding to each primitive is extracted, and the specific position relationship of the ship structure represented by these primitives is calculated.

[0120] In step S410, based on the detection box and category labeling information obtained in the previous step (such as S310, object detection), it is first necessary to extract geometric data for each primitive. Specifically, the task of the S410 module is to extract information such as geometric shape, size, boundary, position, etc. from the identified primitives. These geometric data provide key references for the spatial layout of the ship structure.

[0121] For example, in step S310, each detection box obtained after object detection represents a primitive area. Next, we need to extract the following geometric information from the area:

[0122] Geometric shapes: such as straight lines, dotted lines, rounded rectangles, curves, etc.;

[0123] Size: the physical size of each primitive, such as length, width, and height;

[0124] Boundary information: edge coordinates, angles, curvature, etc. of the primitive;

[0125] Spatial position: The position coordinates of a graphic element on a two-dimensional drawing, usually expressed through a drawing coordinate system.

[0126] Based on this information, we can further infer the spatial relationship of the ship structure represented by the primitives in the actual ship, including the relative position and angle of the primitives, and then calculate how these primitives are combined into the overall ship structure. For example, if the primitive represents the "hull structure", then the geometric data of the primitive can reflect the specific position of this part in the ship.

[0127] S420, further calculating or deducing material properties, shape and size data of the ship structure according to the ship CAD modeling language rules and the extracted metadata, and converting this information into parametric model data of the ship structure.

[0128] S430, uniformly expressing the extracted geometric information, structural attributes and material data in a parametric manner, and constructing complete ship structure parametric model data.

[0129] S430 is the final stage of step S400. Its main task is to uniformly parametrically express all the geometric information, structural properties, and material data extracted in S410 and S420, and construct the complete parametric model data of the ship structure. Specifically, this step requires organizing all the previous calculation results and data into a complete and standardized model framework to ensure that all components and elements of the ship design can be represented and operated according to unified specifications.

[0130] In this process, all geometric information (such as size, position, etc.), structural properties (such as strength, stiffness, etc.) and material information (such as density, tensile strength, etc.) will be mapped into corresponding parametric data sets. For example, we will generate parametric features adapted to the ship structure based on the size, material and other information in the CAD drawing. These features can be:

[0131] Structure type, geometric coordinates, dimensions, material properties, etc.

[0132] Through these unified parametric model data, a complete digital model of the ship structure can be generated, and data support can be provided for subsequent 3D modeling, structural analysis, simulation testing, etc. The final parametric model can not only represent all the geometric features of the ship design, but also reflect the physical, mechanical and material properties of the design, making the modeling process more accurate and flexible.

[0133] Step S500 at least includes steps S510-S520:

[0134] S510. According to the ship CAD design logic and in combination with the generated ship structure parametric data, a suitable modeling instruction template is created or selected, and the structure parameters are matched with the instruction template.

[0135] In step S510, firstly, according to the logic and design requirements of the ship CAD design, combined with the generated parametric data of the ship structure, a suitable modeling instruction template is created or selected. This template will be responsible for receiving and converting the ship design parameters and generating the corresponding ship CAD modeling instructions.

[0136] Specifically, in the previous step S400, the parametric data of the ship structure is obtained through the primitive information, geometric data and material attributes, etc. These data include but are not limited to:

[0137] Geometric parameters (such as size, position, shape, etc.)

[0138] Structural type (such as plates, bulkheads, manholes, etc.)

[0139] Material data (such as the material number, type, thickness, etc.)

[0140] Next, we need to select a modeling instruction template that meets these parameter requirements. The choice of template is based on the ship CAD design specifications. For example, if the component of the ship structure is a "hull frame", you should select an instruction template related to frame modeling; if it is a "ship deck" or "cabin bulkhead", you should select different templates according to different structure types.

[0141] By docking with the design specifications, the template can handle different types of structural components and adjust according to specific parameters. This process can be understood as the "matching phase". Through the template, we can ensure that the generated modeling instructions meet the required design requirements and technical specifications.

[0142] S520, converting the parameterized data of the ship structure into corresponding ship CAD modeling instructions through template matching to generate a complete set of modeling instruction sequences

[0143] In step S520, after completing the template selection, the next step is to convert the ship structure parametric data matched in step S510 into corresponding ship CAD modeling instructions through template matching, and generate a complete set of modeling instruction sequences. The specific process is:

[0144] Data input: Input the selected template and the parametric data of the relevant ship structure into the template. This template is essentially a structure consisting of a series of predefined rules, which contain the specific steps on how to use these parameters to build a ship.

[0145] Template parsing: The template parses the input ship structure parameters and matches each data point (such as length, width, material, etc.) with the corresponding function in the instruction template one by one. For example, if a certain element corresponds to the rib structure of the hull, the template will generate appropriate CAD commands (such as Create Stiffener,<Pos it i on> ,<Mater i a l> , <s i ze>wait).

[0146] Instruction sequence generation: After template matching is completed, a series of specific ship CAD modeling instructions will be generated based on the input ship structure parameters. These instructions specify how to rebuild the ship structure in the CAD system, including how to draw each element, how to connect different elements, how to set the size and material properties of different elements, etc. The generated instruction sequence is highly automated and accurate, ensuring that the generated model meets the design requirements.

[0147] The generated ship CAD modeling instruction sequence may include the following:

[0148] Create boundaries, segment lines, plates, ribs, holes, etc.

[0149] These instructions rely on the primitive geometry data, material data, and ship design processes and rules extracted in the previous steps to achieve accurate modeling of the ship structure.

[0150] Step S600 at least includes steps S610-S630:

[0151] S610: Input the generated modeling instruction sequence into the CAD system, drive the CAD software to execute the instructions through a predefined interface, and start the automatic modeling process.

[0152] In step S610, the modeling instruction sequence generated in the previous step is first input into the CAD system. Specifically, the modeling instruction sequence generated in step S520 contains the parametric data of the ship structure, and these instructions define how to build the corresponding ship components in the CAD system through geometry, structural attributes and material data.

[0153] Next, the CAD software is driven to execute these instructions through a predefined interface (such as the API interface of the CAD software). The function of the interface is to pass the instruction sequence to the CAD system and ensure that the instructions are executed in the correct order and manner. For example, the CAD system may use a series of ship structure parametric modeling instructions to achieve the creation of the model.

[0154] Specifically, it is converted into a natural language description, for example: "Create a segment line at the rib position FR30+300."

[0155] After these instructions are passed to the CAD software through the interface, the CAD software will start processing according to the instructions and gradually build the three-dimensional structure of the ship. The core of this process is that the CAD system can parse and execute these pre-defined instructions.

[0156] S620, executing modeling instructions in the CAD system to gradually create a three-dimensional model of the ship.

[0157] In step S620, the CAD system executes the input modeling instructions step by step and starts to build a three-dimensional ship model. Specifically, the CAD system generates each component of the ship model one by one according to the steps of geometry creation, size setting, material assignment, etc. in the modeling instructions. This process can be divided into several key links:

[0158] Geometry creation: Based on the structural information determined in step S510, such as length, width, shape, etc., the CAD system starts to create the corresponding geometry. For example, for a transverse frame of a ship, the CAD system may generate a rectangle or other geometric shape according to the instruction to represent the basic part of the frame.

[0159] Attribute setting and adjustment: According to the material and structural parameters passed in from step S520, the CAD system sets the material for each ship structure. For example, the metal material number and thickness of the deck surface are set. For each created geometry, the CAD system adjusts its size, relative position and connection method according to the template instructions.

[0160] Step-by-step combination and connection: During the ship modeling process, multiple geometric bodies may need to be combined or connected in a specific way. After receiving the instructions, the CAD system will connect these different parts in sequence according to the ship design specifications to form a complete ship structure.

[0161] Adjust details: After all basic geometries are completed, the system will continue to make detailed adjustments according to instructions, including fine-scale adjustment, position correction, material update, etc., to ensure that the final generated 3D ship model meets the design requirements.

[0162] S630: After the three-dimensional model is constructed, the finally generated ship model is manually inspected.

[0163] The key innovative features of the present invention include:

[0164] (1) Automated primitive analysis and recognition: The present invention automatically extracts geometric information, structural attributes and material data from drawings through data analysis and primitive recognition modules, eliminating the errors and complexity of manual recognition and improving the efficiency and accuracy of model construction.

[0165] (2) Deep learning object detection and image generation: By using deep learning algorithms (such as YOLO and Faster R-CNN) and object detection models, it is possible to automatically detect and annotate complex primitive areas and accurately calibrate their locations, further enhancing the accuracy of primitive recognition.

[0166] (3) Parametric modeling and modeling instruction generation: The present invention parametrically expresses the extracted geometric information, structural properties and material data, matches them with the modeling instruction template, and automatically generates a modeling instruction sequence, thereby greatly improving the automation level and modeling speed of three-dimensional modeling.

[0167] The automatic three-dimensional modeling method, system and device based on the two-dimensional drawings of ship CAD provided by the present invention can efficiently extract the primitive information from the two-dimensional drawings of ship CAD through an automated process, and generate three-dimensional modeling instructions, so as to realize the automatic construction of the three-dimensional model of the ship in the CAD system. Compared with the traditional manual modeling process, the automated method of the present invention can significantly reduce manual intervention and improve the efficiency and accuracy of ship design.

[0168] Through the method, especially with the support of the primitive parsing and object detection modules, complex primitive areas can be accurately identified, eliminating the errors and inconsistencies in traditional manual primitive identification, thereby ensuring the accuracy of the generated three-dimensional model. At the same time, the use of deep learning algorithms and object detection models can automatically identify key components in the image and accurately calibrate their positions, making the entire modeling process more intelligent and efficient.

[0169] In addition, the present invention converts the geometric information, structural properties and material data of the ship structure into parametric model data, and matches them with the modeling instruction template to generate a modeling instruction sequence, which not only reduces the human design deviation, but also improves the modeling speed and can adapt to complex and changeable ship design requirements.

[0170] Embodiment 2: Figure 2 FIG. 2 shows a block diagram of an automatic three-dimensional modeling system based on a two-dimensional drawing of a ship CAD according to an embodiment of the present invention. Figure 2 As shown, the system may include:

[0171] The data acquisition module 10 is responsible for acquiring the original design data from the two-dimensional CAD drawings of the ship. The data includes geometric information (such as lines, surfaces, circles, annotations, etc.), material information, structural data, and other related graphic metadata in the drawings. By parsing CAD files (such as DXF, DWG formats), the module can extract various parameters of the ship design and provide basic data for subsequent processing.

[0172] Extract and analyze 2D CAD drawing data, including geometric data, material information, structural data, etc.;

[0173] Ensure the accuracy and completeness of data to avoid data loss or misunderstanding due to drawing format or quality issues;

[0174] Convert the collected data into a unified format to facilitate processing by subsequent modules.

[0175] The data processing module 20 pre-processes the raw data obtained from the data acquisition module 10, including operations such as filling missing values ​​and removing abnormal data. This module ensures that the data is consistent and accurate when entering subsequent analysis, thereby avoiding unnecessary modeling errors.

[0176] Identify and repair missing values ​​or abnormal data to ensure data validity;

[0177] Standardized data formats make data from different sources compatible and processable.

[0178] The graphic element recognition and analysis module 30 performs graphic element analysis on the processed data based on the ship design rules and specifications, identifies the structural components in the drawings, and matches these graphic elements with the actual ship components according to the design specifications. This module is the core of the entire modeling process and is responsible for correctly classifying each graphic element (such as cabin, hull frame, deck, etc.) in the CAD drawings and matching them with the design standards.

[0179] Identify all the elements in the drawing, including line types, annotations, symbols, etc.;

[0180] Match the elements with the ship design specifications to ensure that each element correctly corresponds to the components of the ship structure;

[0181] According to ship design rules, the size, position relationship and interconnection method of each component are analyzed to provide structured data for subsequent modeling.

[0182] The object detection and image generation module 40 is mainly used to process complex primitive areas that cannot be directly identified through rule parsing. This module uses off-screen rendering technology to generate images and uses object detection models (such as YOLO, Faster R-CNN, etc.) to perform object recognition on the images. After recognition, the system can extract information from the image and further supplement and improve the primitive data.

[0183] The unrecognized complex areas are converted into images through off-screen rendering technology;

[0184] Use deep learning object detection models to identify rendered images and generate corresponding detection boxes and category annotations;

[0185] Connect the identified graphic element information with the original data in the drawing to supplement the missing parts.

[0186] The parametric modeling module 50 performs parametric modeling of the ship structure based on the data generated by the object detection and primitive analysis modules. By combining the CAD design specifications, the module converts the extracted geometric data, material data, etc. into a structured parametric model. This model not only reflects the geometric shape of the ship components, but also covers design elements such as material, size, and position relationship.

[0187] Convert the geometric data, structural attributes and material data of the graphic elements into parametric expressions to build a parametric model of the ship structure;

[0188] Adjust and optimize the size, shape, material and other parameters of each component according to the ship design requirements;

[0189] Provide complete and standardized parametric model data for subsequent CAD modeling.

[0190] The modeling instruction generation module 60 automatically generates the instructions required for CAD modeling based on the parametric data of the ship structure. The module selects a suitable modeling instruction template according to the design requirements of each ship component, matches the parametric data with the template, and finally generates a complete set of modeling instruction sequences. These instructions will be directly used to control the CAD software for three-dimensional modeling.

[0191] Select appropriate modeling instruction template according to ship design process and rules;

[0192] Match the parametric data of the ship structure with the template to generate specific modeling instructions;

[0193] Output a series of standardized modeling instructions for execution by the CAD system.

[0194] The CAD modeling execution module 70 is responsible for receiving the instruction sequence transmitted from the modeling instruction generation module 60, and executing these instructions through the interface of the CAD system to gradually build a three-dimensional model of the ship. By executing these instructions, the system can automatically convert each component of the ship from a two-dimensional drawing into a three-dimensional solid model.

[0195] Receive modeling instruction sequences and execute the instructions through the interface of CAD software;

[0196] Follow the modeling instructions to create the three-dimensional structural components of the ship step by step;

[0197] Supports dynamic model adjustment to ensure that the final generated 3D model meets the design requirements.

[0198] Through the system structure module of the embodiment of the present invention, the two-dimensional CAD drawings of ships can be automatically converted into accurate three-dimensional models. The system has the following beneficial effects:

[0199] Efficiency: By automating the steps of data collection, analysis, modeling, etc., the ship design cycle is greatly shortened and the design efficiency is improved.

[0200] Accuracy: Parametric modeling and strict design specifications are used to ensure that the generated 3D model is highly accurate and meets the actual ship construction requirements.

[0201] Intelligence: Through object detection and deep learning algorithms, the system can automatically identify and process complex graphics, further improving the intelligence of model generation.

[0202] Modularity: The various modules of the system work closely together, each independent but able to collaborate efficiently, making ship design more flexible and scalable.

[0203] Strong applicability: The system is not only suitable for ship design, but can also be extended to other engineering design fields, such as construction, machinery manufacturing, etc., to achieve cross-industry application.

[0204] In general, the system of the present invention significantly improves the efficiency and accuracy of design by automating the traditional ship design process, and provides a powerful digital tool for the ship design industry.

[0205] Embodiment three:

[0206] Figure 3 FIG. 2 shows a block diagram of a computer device according to an embodiment of the present application. Figure 3 As shown, the computer device includes:

[0207] The memory 310 is used to store program instructions and data required during the operation of the computer device. In the embodiment of the present invention, the memory 310 mainly stores the following contents:

[0208] Operating system: supports the basic operation of computer devices;

[0209] Modeling algorithm instructions: used to realize the automatic 3D modeling method of ship CAD 2D drawings, including data collection, analysis, modeling instruction generation, 3D model construction and other steps;

[0210] Ship design data: including geometric data, material information, design specifications, etc. of 2D CAD drawings;

[0211] Model parameterized data: The ship structure parameterized data generated by the parameterized modeling module 50 is used for subsequent modeling instruction generation and three-dimensional model construction.

[0212] The memory 310 may include, but is not limited to, disk storage, a solid state drive (SSD), or other storage devices. Depending on the specific implementation, the memory 310 may be one or more storage media.

[0213] The processor 320 is responsible for loading and executing the instructions in the memory 310, thereby realizing various functions of the ship three-dimensional modeling method. The functions of the processor 320 include but are not limited to:

[0214] Data analysis and processing: Extract geometric information, graphic metadata, material information, etc. from ship CAD 2D drawings;

[0215] Modeling instruction generation: Generate ship CAD modeling instructions based on the extracted parametric data;

[0216] Model generation and visualization: Use ship CAD software to execute modeling instructions, complete 3D modeling and visualization, and support manual inspection and modification.

[0217] The performance of the processor 320 has a direct impact on the efficiency of the overall modeling system, and it needs to be able to handle a large amount of drawing data and complex modeling calculations. Therefore, the processor 320 can select a multi-core or parallel processor to improve computing power and response speed.

[0218] The communication interface 330 is used for the computer device to communicate, exchange data and transmit data with external devices (such as sensors, input and output devices, other computer systems, etc.). In the implementation of the present invention, the communication interface 330 has the following responsibilities:

[0219] Connect with external data sources: used to receive ship CAD drawing data (such as DXF, DWG files, etc.) and exchange data with other systems (such as engineering analysis systems);

[0220] Data transmission: Output the generated ship 3D model data, modeling instructions, etc. to other devices such as displays, printers or cloud storage through the network or local bus;

[0221] External device interaction: interact with the operator's input devices (such as mouse, keyboard, etc.) or other systems to receive user instructions or external data.

[0222] The communication interface 330 can support a variety of standard interfaces, such as a USB interface, an Ethernet interface, a wireless communication module, etc., and can connect and transmit data with different types of devices as needed.

[0223] The various components in the computer device (memory 310, processor 320 and communication interface 330) are interconnected through a bus system for data transmission and interaction. The bus system can be a standard industrial bus architecture (such as ISA bus, PCI bus or EISA bus) for connecting various modules and devices in the computer system.

[0224] Bus systems are divided into different bus types, including:

[0225] Address bus: used to transmit address information and determine the location where data is stored or read;

[0226] Data bus: used to transmit data, connect memory, processor, external devices, etc.;

[0227] Control bus: used to coordinate the operations of each module and ensure the correctness of data transmission.

[0228] In a specific embodiment of the present invention, the bus system supports efficient transmission of data streams and ensures close cooperation among the memory 310 , the processor 320 and the communication interface 330 .

[0229] The computer device structure block diagram provided by the present invention can effectively support the automatic three-dimensional modeling process based on the two-dimensional drawings of ship CAD through the cooperation of various modules. Its beneficial effects are mainly reflected in the following aspects:

[0230] Efficient processing capability: Through the collaborative work of the high-performance processor 320 and the memory 310, a large amount of ship CAD drawing data can be quickly processed and three-dimensional modeling tasks can be performed efficiently.

[0231] Strong data interaction and compatibility: Through the communication interface 330, the system can flexibly interact and transmit data with external devices, and can quickly receive drawing data and user instructions to meet diverse needs.

[0232] Modular design: The modular design of the computer device allows the various parts of the system to work independently and collaborate efficiently, supporting future expansion and upgrades, such as adding more external interfaces or improving algorithms.

[0233] Integrated solution: The optional integrated solution can integrate the memory 310, the processor 320 and the communication interface 330 on a single chip, thereby improving system performance and saving space, and is suitable for implementation in embedded or mobile devices.

[0234] Flexibility and scalability: The system supports extensive interactions with other devices and platforms, including functional modules such as display, printing, and cloud platform, and has good scalability.

[0235] Through this structured computer device architecture, the present invention can efficiently execute the ship three-dimensional modeling method, improve the automation level of ship design, and provide support for automated modeling applications in other fields.

[0236] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.< / s>

Claims

1. An automatic three-dimensional modeling method based on a two-dimensional drawing of a ship CAD, characterized in that: The method comprises the following steps: Obtain the input ship standard CAD drawings, perform data analysis, extract metadata from the drawings, and conduct preliminary analysis of the drawings based on ship design specifications to identify graphic elements; For unrecognized complex primitive areas, images are generated through rendering technology, and the generated images are preprocessed; The rendered image is input into the deployed object detection model, and the detection box and category annotation are generated using the deep learning algorithm. The coordinates are converted according to the drawing coordinate system and the image coordinate system to accurately calibrate the position of the primitives. Based on the object detection results and primitive information, relevant geometric information, structural attributes and material data are extracted, parametric expression conversion is performed, and parametric model data of the ship structure is generated; Input the generated ship structure parametric data into the modeling instruction template, automatically match and generate the ship CAD modeling parametric instruction sequence; The generated modeling instruction sequence is input into the CAD system to drive the CAD software to execute the modeling instructions, automatically construct the three-dimensional model of the ship and complete the modeling process.

2. The automatic three-dimensional modeling method based on the two-dimensional drawing of ship CAD according to claim 1 is characterized in that: The step of obtaining the input ship standard CAD drawing also includes: Parse ship CAD drawing data through open source parsing library or proprietary parsing module to extract geometric information, lines, surfaces, circles, and text element information in the drawings.

3. The automatic three-dimensional modeling method based on the two-dimensional drawing of ship CAD according to claim 1 is characterized in that: The data parsing step specifically includes: The metadata in the drawings is cleaned and standardized, redundant data is eliminated, and valid graphic element information is extracted.

4. The automatic three-dimensional modeling method based on the two-dimensional drawing of ship CAD according to claim 1 is characterized in that: The preprocessing of the rendered image includes: Perform image enhancement and smoothing on rendered images.

5. The automatic three-dimensional modeling method based on the two-dimensional drawing of ship CAD according to claim 1 is characterized in that: The deep learning algorithm of the object detection model specifically includes: Use YOLO and Faster R-CNN neural network models to detect objects in images and generate detection boxes and category labels.

6. The automatic three-dimensional modeling method based on the two-dimensional drawing of ship CAD according to claim 1 is characterized in that: The coordinate transformation specifically includes: Match the detection box coordinates in the image coordinate system with the CAD drawing coordinate system.

7. The automatic three-dimensional modeling method based on the two-dimensional drawing of ship CAD according to claim 1 is characterized in that: The step of extracting geometric information, structural attributes and material data specifically includes: The geometric data, material properties and structural parameters of the ship structure are extracted from the graphic element information, the size, position, material and other data of the graphic element are calculated, and converted into a parametric form.

8. The automatic three-dimensional modeling method based on the two-dimensional drawing of ship CAD according to claim 1 is characterized in that: The step of generating a modeling instruction sequence specifically includes: According to the ship design process and parametric data, a suitable modeling instruction template is selected, and the extracted geometry, structure, and material data are matched with the template to generate a complete modeling instruction sequence.

9. An automatic three-dimensional modeling system based on ship CAD two-dimensional drawings, characterized in that: The system comprises: Data acquisition module, used to obtain and analyze ship standard CAD drawing data and extract graphic element information; Data processing module, used to clean and standardize image metadata and extract valid information; Graphics recognition and analysis module, used to recognize graphics and analyze ship structures according to ship design specifications; Object detection and image generation module, used to process complex primitive areas and generate accurate primitive data through rendering and object detection; Parametric modeling module, used to convert the extracted geometric information, structural properties and material data into parametric model data; A modeling instruction generation module, used for matching parameterized data with a modeling instruction template to generate a modeling instruction sequence; The CAD modeling execution module is used to input the modeling instruction sequence into the CAD system to automatically build a three-dimensional model of the ship.

10. An automatic three-dimensional modeling device based on a two-dimensional ship CAD drawing, characterized in that: The device comprises: A processor is used to execute instructions stored in the memory to complete tasks such as data analysis, primitive recognition, modeling instruction generation and 3D modeling; A memory, used to store ship CAD two-dimensional drawing data, parametric modeling data and generated modeling instruction sequences; Communication interface, used for data interaction with external devices and drawing data transmission; The display is used to display the generated three-dimensional ship model and related data.

Images