A wire frame automatic parsing method, system, device and medium fusing geometric field and target detection

Abstract

This application discloses an automatic wire harness diagram parsing method, system, device, and medium that integrates geometric field and target detection, relating to the field of wire harness diagram parsing. The method includes: segmenting the wire harness diagram to be processed into multiple sub-images and performing multi-target detection; performing differential masking on specific regions of the wire harness diagram to be processed based on the target detection results; inputting the interference-suppressed wire harness diagram into a U-Net-like network to generate distance and orientation geometric fields, and extracting wire harness segments; obtaining the wire harness segment binding method using a binding method classification model based on the wire harness segment set and sub-images; performing character recognition using OCR technology to obtain component location numbers and wire harness segment size information; constructing the wire harness topology based on the target detection results, distance and orientation geometric fields; and integrating the above information to generate a structured output result, completing the automatic wire harness diagram parsing. This application can achieve fully automated processing from drawing input to engineering result output.

Description

Technical Field

[0001] This application relates to the field of wire diagram analysis, and in particular to an automatic wire diagram analysis method, system, device and medium that integrates geometric field and target detection. Background Technology

[0002] Wire harnesses, as a key carrier for power, signal, and data transmission, are indispensable core components in modern automotive, aerospace, and various industrial equipment. Wire harness diagrams, as technical documents for wire harnesses, detail the wiring path, functional connections, dimensions, wrapping materials, and component interface information. They are core technical documents that permeate the entire process of design, production, assembly, and testing. With the increasing complexity of equipment systems and the continuous expansion of the number and structural scale of wire harnesses, the design complexity of wire harness diagrams has significantly increased. However, current wire harness diagram processing still relies on a hybrid model of "fully manual + semi-automated tools," resulting in low efficiency and high error rates at each stage.

[0003] Currently, detection and extraction technologies based on deep learning and traditional image algorithms are gradually being applied in the field of industrial drawing processing, providing ideas for the automated processing of wire harness diagrams. To address the efficiency and accuracy issues in wire harness diagram processing, some related technological explorations have emerged within the industry. However, existing technologies still have significant limitations: the processing stages are relatively fragmented, lacking a unified end-to-end process; the stability of connection relationship inference is insufficient, making it difficult to adapt to diverse drawing styles; the recognition of wire harness wrapping methods is not systematic enough; and the output results lack a structured form that can be directly used in engineering. Furthermore, although deep learning-based line segment detection technologies (such as DeepLSD) have made progress, capable of encoding line segment structures through geometric feature fields (distance maps and angle maps), they have not yet been used for connection relationship inference between components and wire harnesses, nor have they been integrated into a complete industrial process. Summary of the Invention

[0004] The purpose of this application is to provide a method, system, device and medium for automatic analysis of wire harness diagrams that integrates geometric field and target detection. It can realize the fully automated processing from drawing input to engineering result output. Without relying on large-scale annotation data, it can accurately infer the wire harness connection relationship, identify the wrapping method and output structured results that can be directly applied to engineering.

[0005] To achieve the above objectives, this application provides the following solution: In a first aspect, this application provides an automatic beammap parsing method that integrates geometric field and target detection, including: The wire harness image to be processed is divided into multiple sub-images; Multi-object detection is performed on each sub-image based on the object detection model to obtain object detection results; Based on the target detection results, a differential masking process is performed on a specific region of the wire harness image to be processed to obtain a wire harness image after interference suppression; the specific region is the area where the attachments and components are located in the wire harness image to be processed; The interference-suppressed wireframe diagram is input into a U-Net-like network to generate the range geometry and orientation geometry. Based on the distance geometry and the orientation geometry, the improved LSD algorithm is used to extract the wire harness segments to obtain a set of wire harness segments; Based on the set of wire harness segments and the sub-image, the wire harness segment wrapping method is obtained using a wrapping method classification model; Based on the target detection results, OCR technology is used to recognize characters and obtain component location numbers and wire harness segment size information; Based on the target detection results, the distance geometry, and the orientation geometry, a wire harness topology is constructed. By integrating the wiring harness segment wrapping method, the component location number, the wiring harness segment size information, and the wiring harness topology, a structured output result is generated, and the wiring harness diagram is automatically parsed.

[0006] Secondly, this application provides an automatic beam diagram parsing system that integrates geometric field and target detection, comprising: The segmentation module is used to segment the wire harness image to be processed into multiple sub-images; The multi-object detection module is used to perform multi-object detection on each sub-image based on the object detection model to obtain the object detection results; An interference suppression module is used to perform differential masking on a specific region of the wire harness image to be processed based on the target detection result, so as to obtain an interference-suppressed wire harness image; the specific region is the area where the attachments and components are located in the wire harness image to be processed; The geometry field generation module is used to input the interference-suppressed wireframe diagram into a U-Net-like network to generate the range geometry field and orientation geometry field; The harness segment extraction module is used to extract harness segments based on the distance geometry field and the direction geometry field using an improved LSD algorithm to obtain a set of harness segments; A wire harness segment wrapping method identification module is used to obtain the wire harness segment wrapping method based on the set of wire harness segments and the sub-image, using a wrapping method classification model. The character recognition module is used to perform character recognition using OCR technology based on the target detection results, and obtain the component location number and wire harness segment size information; A wire harness topology construction module is used to construct a wire harness topology based on the target detection results, the distance geometry field, and the orientation geometry field. The structured output result generation module is used to integrate the wiring harness segment wrapping method, the component position number, the wiring harness segment size information, and the wiring harness topology to generate structured output results and complete the automatic parsing of the wiring harness diagram.

[0007] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described automatic parsing method for fusion of geometric field and target detection of wire diagram.

[0008] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method for automatic parsing of wireframe diagrams that fuses geometric fields and target detection.

[0009] According to the specific embodiments provided in this application, this application has the following technical effects: This application achieves fully automated parsing of wire harness diagrams by constructing an end-to-end workflow of "image segmentation → target detection → differential masking → geometric field prediction → line segment extraction → bundle recognition → OCR → topology construction → structured output". It utilizes masking to suppress interference from attachment / component regions, improving line segment extraction accuracy; it integrates geometric field and target detection results to jointly construct topology, enhancing adaptability to diverse drawing styles; and it systematically integrates wire harness bundle recognition, character recognition, and topology construction, ultimately outputting structured engineering-usable data that integrates wire harness bundle recognition, component location numbers, wire harness size information, and wire harness topology. This solves the core pain points of existing technologies, such as fragmented workflows, unstable connection reasoning, unsystematic bundle recognition, and unstructured output. Furthermore, it is the first to extend geometric field technology from line segment detection to connection relationship reasoning and embed it into a complete industrial process. This application is applicable to multiple industries and types of wire harness diagrams, exhibiting good versatility and scalability. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 A flowchart illustrating an automatic parsing method for a wireframe diagram that integrates geometric field and target detection, provided in an embodiment of this application; Figure 2This is a schematic diagram illustrating the process for identifying the wrapping method of wire harness segments. Detailed Implementation

[0012] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0013] This application proposes an intelligent processing and structured output technology solution for wire harness diagrams, aiming to automate the entire process from drawing input to engineering result output. The solution first converts PDF format drawings into PNG images and preprocesses them using slider segmentation. Then, it performs component target detection based on a target detection model, generates a geometric feature field through a U-Net-like network, and uses an improved LSD algorithm to detect the wire harness structure, while simultaneously extracting character information from the drawings using OCR technology. Based on this, it infers the connection relationships between the wire harness and components such as connectors and fixing hooks based on the geometric field information, constructing a complete wire harness topology. This replaces the tedious process of traditional manual drawing interpretation and annotation, accurately inferring wire harness connection relationships and identifying wrapping methods without relying on large-scale annotation data, and outputting structured results that can be directly applied to engineering.

[0014] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0015] In one exemplary embodiment, such as Figure 1 As shown, an automatic parsing method for wireframe diagrams that integrates geometric field and target detection is provided. This method is executed by a computer device, specifically by a terminal or server alone, or by both a terminal and a server. In this embodiment, the method is described using a server as an example, and includes the following steps S1 to S9.

[0016] S1: Divide the wire harness image to be processed into multiple sub-images.

[0017] S2: Perform multi-object detection on each sub-image based on the object detection model to obtain the object detection results.

[0018] S3: Based on the target detection results, perform differential masking on a specific region of the wire harness image to be processed to obtain a wire harness image after interference suppression; the specific region is the area where the attachments and components are located in the wire harness image to be processed.

[0019] S4: Input the interference-suppressed wire diagram into a U-Net-like network to generate the range geometry and orientation geometry.

[0020] S5: Based on the distance geometry field and the direction geometry field, the improved LSD algorithm is used to extract the wire harness segments to obtain a set of wire harness segments.

[0021] S6: Based on the set of wire harness segments and the sub-image, the wire harness segment wrapping method is obtained using a wrapping method classification model.

[0022] S7: Based on the target detection results, OCR technology is used to recognize characters and obtain the component location number and wire harness segment size information.

[0023] S8: Based on the target detection results, the distance geometry field, and the orientation geometry field, construct the wire harness topology.

[0024] S9: Integrate the wiring harness segment wrapping method, the component location number, the wiring harness segment size information, and the wiring harness topology to generate a structured output result and complete the automatic parsing of the wiring harness diagram.

[0025] By implementing steps 201 to 208 above, the entire process of wire harness diagram parsing is automated, significantly reducing labor costs.

[0026] In a specific embodiment, step S1 specifically includes: Obtain the wire harness image to be processed in the initial PDF format, convert the PDF wire harness image to PNG format, scale the PNG wire harness image to be processed, and slide it into sub-images of size 1280×1280 with a 50% overlap ratio.

[0027] In one specific embodiment, the target detection model in step S2 adopts the YOLOv8n model.

[0028] The overall network structure of the YOLOv8n model is divided into four parts: input, backbone, neck, and head. The input layer mainly consists of three parts: Mosaic data augmentation, image size normalization, and adaptive anchor box calculation, providing high-quality input for subsequent feature extraction. The backbone network adopts a lightweight CSPDarknet architecture, extracting multi-scale features through Conv convolutional layers and C2f modules, retaining key features while reducing parameters, and outputting four feature maps at different scales after four downsampling operations. The neck network adopts an FPN+PAN structure, fusing the high semantic features and high detail features output by the backbone network across scales, and transmitting feature information through a lightweight C2f module, ultimately outputting three core feature maps of 80×80, 40×40, and 20×20, corresponding to the detection requirements of small, medium, and large targets, respectively. The head network adopts an anchor-free design, directly convolving the feature maps of the three detection layers, outputting a vector containing coordinate parameters, confidence scores, and class probabilities, and finally generating accurate detection bounding boxes in the original image, completing the recognition and localization of four types of targets in the wire harness image: appendix, connector, fixing hook, and wire harness segment size annotation.

[0029] In addition, object detection models such as YOLOv5, YOLOv7, and Faster-RCNN, or semantic segmentation models, can also be used.

[0030] In a specific embodiment, step S3 specifically includes: The target detection result file output by the YOLOv8n model is parsed to obtain the target category ID and bounding box coordinates. The areas where the appendix and components (plugs and fixing hooks) are located in the wire harness diagram to be processed are selected and filled with white to remove interference from subsequent detection of wire harness segments.

[0031] In one specific embodiment, step S4 specifically includes: Image quality enhancement processing is performed on the interference-suppressed line bundle image by sequentially performing contrast-limited adaptive histogram equalization (CLAHE), Gaussian smoothing, and adaptive thresholding or OTSU binarization to enhance the contrast between the line bundle region and the background and suppress noise interference.

[0032] The enhanced image is input into a U-Net-like network, employing an encoder-decoder structure. Local texture and global structure information are fused through multi-scale downsampling, upsampling, and skip connections, resulting in end-to-end outputs of two types of pixel-level geometric fields: distance geometric field. and directional geometric field .

[0033] The distance geometry can be expressed as: in, Indicates the pixel position in the image. This represents the set of wire harness segments obtained from the detection. This represents the geometric distance from a pixel to a line segment.

[0034] The directional geometric field can be expressed as: in, This represents the directional geometric field value at pixel position x, where x represents the pixel coordinate position in the image. "Represents the line segment that is closest to pixel location x or that dominates the direction in its local neighborhood." "This indicates the direction angle or unit direction vector of the harness segment at that location, used to characterize the local orientation of the harness."

[0035] In addition, the U-Net network can be replaced by the Probabilistic Hough Line Transform.

[0036] In a specific embodiment, step S5 specifically includes: screening candidate pixels in the neighborhood of the line bundle where the distance value in the distance geometry is less than a threshold; clustering the candidate pixels according to similar directions in the directional geometry to form multiple support regions with consistent directions; performing line fitting on each support region to obtain initial line segments, and verifying the initial line segments to obtain an initial line bundle segment set; and optimizing the initial line bundle segment set to obtain the final line bundle segment set.

[0037] Unlike traditional LSDs that grow directly on edges based on gradients, this application extracts line segments using... As a local directional prior, with As a neighborhood constraint for the line harness, the growth, aggregation, and verification of line segments are completed in the pixel domain. The specific process is as follows: (1) Candidate pixel selection: In Candidate pixels are selected within the neighborhood of the wire harness; (2) For candidate pixels, press Clustering or region growing is performed on similar directions to form several support regions with consistent directions; (3) Line segment fitting and verification: Perform linear (or local linear) fitting on each support region to obtain the initial line segment. The initial set of line segments is obtained by verifying the data based on factors such as the number of supported pixels, length, and orientation consistency. .in and Let represent the coordinates of the start and end points of the initial line segment, respectively. This represents the i-th initial line segment.

[0038] Using the above method, a more continuous and complete line segment structure can be obtained from complex wire harness diagrams. Furthermore, the direction of the line segments is consistent with... Keep it consistent, the position of the line segment and The low-distance areas are all at a consistent height.

[0039] Obtain the initial set of line segments Then, the following optimization processing is performed on the line segment detection results: (1) Short line segment removal: Remove line segments whose length is less than the threshold or whose support area has insufficient pixels; (2) Collinear merging: Collinear merging of line segments with similar directions and close endpoints to form longer continuous line segments; (3) Break completion: Bridge and complete line segments that are located in the same direction and have small intervals to reduce the situation where the line bundle is split into multiple segments; (4) Consistency of direction and position: utilizing and Fine-tune the endpoints of the line segments to make them closer to the wire harness skeleton and maintain a consistent direction.

[0040] After the above optimization process, the final set of line segments is obtained. .

[0041] In a specific embodiment, step S6 specifically includes: determining the spatial position and orientation of each wire harness segment in the set of wire harness segments; cropping the sub-image at multiple preset positions along the orientation of the wire harness segment to obtain multiple local images; and inputting the multiple local images into the bandaging method classification model for recognition to obtain the wire harness segment bandaging method. The specific process is as follows: Figure 2 As shown.

[0042] Based on the wire harness segment detection results in step S5, the spatial position and orientation of each wire harness segment are determined, including the starting position, ending position, and length of the wire harness segment.

[0043] Along the direction of the wire harness segment, at multiple preset proportional positions, the sub-image is cropped with the corresponding position as the center to obtain multiple local images (i.e., Figure 2 The local images (1-n) are defined as follows: Each local image is square, with its size determined by the wireframe resolution and wireframe width, used to fully cover the wireframe and its encapsulating texture features. Multiple sampling points reduce the impact of local occlusion or noise on the recognition results.

[0044] Since the bundle segments may have arbitrary orientations in the image, to reduce the impact of orientation changes on bandaging pattern recognition, the cropped local images are oriented. Multiple oriented-aligned local images are then input into the bandaging pattern classification module for recognition. The bandaging pattern classification module is implemented using a lightweight convolutional neural network structure, including several convolutional layers, pooling layers, and fully connected layers, to extract texture and structural features from local regions and output the corresponding bandaging category prediction results.

[0045] Each local image independently generates a bandaging category judgment result, thereby obtaining multiple bandaging method classification results for the same bundle segment at different locations (i.e., Figure 2 The classification results are 1-n. The classification results of each local image are fused by voting or weighted fusion, and the bandaging category with the highest frequency or the highest overall confidence is selected as the final bandaging method judgment result for the bundle segment.

[0046] In one specific embodiment, step S7 specifically includes: This application utilizes OCR (Optical Character Recognition) technology and employs the PaddleOCR deep learning model to automatically extract and process character information from drawings. It uses the PP-OCRv5 mobile lightweight model as the basic recognition engine and customizes it using the UnifiedOCR class: disabling unnecessary functions like document orientation classification and document correction, loading pre-trained text detection and recognition models, and optimizing parameter settings to balance recognition accuracy and processing speed. This allows the engine to focus on engineering drawing scenarios, reducing unnecessary calculations and improving processing efficiency.

[0047] The system processes the results of target detection and line segment extraction, locating the character detection area in connectors, fixing hooks, wire harness segment size annotations, and the area surrounding the wire harness segments. The character area is rotated in four directions (0°, 90°, 180°, 270°) and then recognized using the aforementioned recognition engine to obtain confidence scores. The optimal result is automatically selected based on the confidence scores. Filename parsing is used to classify three types of character information: connector position codes, fixing hook position codes, and wire harness segment sizes. Regularization techniques are combined to adapt to dedicated filtering logic and format standardization rules, ensuring the accuracy of information recognition. Simultaneously, dynamic threshold clustering based on text box coordinate distance is used, selecting the topmost text in single-cluster scenarios and the center text in multi-cluster scenarios to remove interference between strings in complex backgrounds. Finally, the data is sorted by coordinates and saved in a structured format.

[0048] In a specific embodiment, step S8 specifically includes: generating terminal candidate points based on the target detection results, and filtering the terminal candidate points using the distance geometry field to obtain valid terminal candidate points; performing geometric consistency evaluation on the valid terminal candidate points based on the direction geometry field and the distance geometry field, and calculating the connection score; determining the connection relationship of the valid terminal candidate points according to the connection score, and constructing the harness topology.

[0049] (1) Generation of terminal candidate points.

[0050] Several candidate terminal points are generated by sampling at certain intervals at the boundaries of detected connectors, fixing hooks, and other component areas or within a preset neighborhood. These candidate terminal points represent potential locations where connections to the wiring harness may occur.

[0051] (2) Harness neighborhood filtering.

[0052] By utilizing distance geometry information, candidate terminal points are filtered, retaining only those with distance values ​​less than a preset threshold as valid candidate terminal points within the wire harness neighborhood, thus reducing invalid matching and calculation.

[0053] (3) Geometric consistency assessment.

[0054] Based on the directional and distance geometric fields, a geometric consistency analysis is performed on each valid terminal candidate point along its pointing direction to evaluate whether a continuous and stable geometric association exists between the valid terminal candidate point and a certain harness segment. The specific process is as follows: First, based on the location and pointing direction of the terminal candidate point, the sampling direction from the terminal candidate point to the harness area is determined. Then, multiple sampling points are acquired along the sampling direction at a preset step size, and the corresponding distance geometric field value and orientation geometric field value are read at each sampling point. Next, the change of distance geometric field value in the sampling point sequence is analyzed to determine whether it shows a continuous decrease and eventually stabilizes in a low value region, so as to evaluate the spatial continuity between the terminal candidate point and the harness segment. At the same time, the consistency analysis of orientation geometric field value in the sampling point sequence is performed to evaluate whether its orientation change is smooth and consistent with the main direction of the candidate harness segment. If the above-mentioned distance continuity and orientation consistency simultaneously meet the preset conditions, it is considered that there is a continuous and stable geometric association between the valid terminal candidate point and the corresponding harness segment.

[0055] In the geometric consistency assessment process, corresponding scoring indicators are calculated from three dimensions: distance, direction, and geometric field stability. Distance Consistency Index This is used to measure the spatial proximity of a candidate terminal point to a wire harness segment. Specifically, the distance value corresponding to the candidate terminal point is obtained through distance geometry, or the distance geometry values ​​of multiple points along the terminal pointing direction are sampled and normalized. The smaller the distance value, the closer the candidate terminal point is to the wire harness skeleton, and the higher the corresponding distance consistency. The larger the value, the better.

[0056] Directional Consistency Index This is used to measure the consistency between the connection direction of a candidate terminal and the orientation of the harness segment. Specifically, it utilizes directional geometry to obtain the main direction information of the harness segment within the terminal's neighborhood and calculates the angle between this angle and the connection direction pointing from the candidate terminal to the harness. A smaller angle indicates higher directional consistency. The larger the value.

[0057] Geometric field stability index This is used to reflect the continuity and stability of the geometric field information along the path from the terminal candidate point to the harness segment. Specifically, along the connection path from the terminal candidate point to the harness segment, the changes in distance and orientation geometric fields are acquired at multiple sampling locations. The stability of the geometric field is evaluated by statistically analyzing its variance or consistency. The smoother and more continuous the geometric field changes, the more reliable the connection relationship. The larger the value.

[0058] (4) Connection score calculation.

[0059] The connection score between the candidate valid terminal points and the harness segment is calculated by considering the consistency of distance, orientation, and geometric field stability between the candidate valid terminal points and the harness segment. Connection rating A weighted combination of multiple indicators: in, This indicates a distance consistency index. Indicators representing directional consistency. Indicates the geometric field stability index. These are the weighting coefficients.

[0060] (5) The connection relationship is determined.

[0061] Based on the connection scoring results, and under the premise of satisfying the connection constraints, the connection relationship between the valid terminal candidate points and the harness segments is determined, thereby constructing the harness topology.

[0062] In one specific embodiment, step S9 specifically includes: After completing the component and harness structure detection, connection relationship inference, wrapping method recognition and text information extraction, all information is integrated to generate structured output results.

[0063] The structured output results include: (1) harness segment size information: the connection relationship between the harness segment and the connector, fixing hook and other components; (2) the harness segment wrapping method; (3) text engineering information such as the location number of various components.

[0064] Structured output results can be output in tables, data files, or other engineering-readable formats for subsequent testing, statistics, analysis, or production guidance.

[0065] Based on the same inventive concept, this application also provides a system for implementing the above-mentioned automatic parsing method for fusion of geometric field and target detection wire diagrams. The solution provided by this system is similar to the implementation described in the above method. Therefore, the specific limitations of one or more embodiments of the automatic parsing system for fusion of geometric field and target detection wire diagrams provided below can be found in the limitations of the automatic parsing method for fusion of geometric field and target detection wire diagrams described above, and will not be repeated here.

[0066] In one exemplary embodiment, an automatic wireframe parsing system that integrates geometric field and target detection is provided, comprising the following modules.

[0067] The segmentation module is used to segment the wire harness image to be processed into multiple sub-images.

[0068] The multi-object detection module is used to perform multi-object detection on each sub-image based on the object detection model to obtain the object detection results.

[0069] The interference suppression module is used to perform differential masking on a specific region of the wire harness image to be processed based on the target detection result, so as to obtain the interference-suppressed wire harness image; the specific region is the area where the attachments and components are located in the wire harness image to be processed.

[0070] The geometry field generation module is used to input the interference-suppressed wire harness diagram into a U-Net-like network to generate the range geometry field and orientation geometry field.

[0071] The harness segment extraction module is used to extract harness segments based on the distance geometry field and the direction geometry field using an improved LSD algorithm to obtain a set of harness segments.

[0072] The wire harness segment wrapping method identification module is used to obtain the wire harness segment wrapping method based on the set of wire harness segments and the sub-image using a wrapping method classification model.

[0073] The character recognition module is used to perform character recognition using OCR technology based on the target detection results, and obtain the component location number and wire harness segment size information.

[0074] A wire harness topology construction module is used to construct a wire harness topology based on the target detection results, the distance geometry field, and the orientation geometry field.

[0075] The structured output result generation module is used to integrate the wiring harness segment wrapping method, the component position number, the wiring harness segment size information, and the wiring harness topology to generate structured output results and complete the automatic parsing of the wiring harness diagram.

[0076] In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments. The computer device may be a server or a terminal. The computer device includes a processor, a memory, an input / output interface (I / O), and a communication interface. The processor, memory, and I / O are connected via a system bus, and the communication interface is connected to the system bus via the I / O interface. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device stores data to be processed. The I / O interface of the computer device is used for exchanging information between the processor and external devices. The communication interface of the computer device is used for communicating with an external terminal via a network connection. When the computer program is executed by the processor, it implements the steps in the above-described method embodiments.

[0077] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0078] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0079] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0080] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0081] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0082] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0083] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for automatic parsing of wireframe diagrams that integrates geometric field analysis and target detection, characterized in that, include: The wire harness image to be processed is divided into multiple sub-images; Multi-object detection is performed on each sub-image based on the object detection model to obtain object detection results; Based on the target detection results, a differential masking process is performed on a specific region of the wire harness image to be processed to obtain a wire harness image after interference suppression; the specific region is the area where the attachments and components are located in the wire harness image to be processed; The interference-suppressed wireframe diagram is input into a U-Net-like network to generate the range geometry and orientation geometry. Based on the distance geometry and the orientation geometry, the improved LSD algorithm is used to extract the wire harness segments to obtain a set of wire harness segments; Based on the set of wire harness segments and the sub-image, the wire harness segment wrapping method is obtained using a wrapping method classification model; Based on the target detection results, OCR technology is used to recognize characters and obtain component location numbers and wire harness segment size information; Based on the target detection results, the distance geometry, and the orientation geometry, a wire harness topology is constructed. By integrating the wiring harness segment wrapping method, the component location number, the wiring harness segment size information, and the wiring harness topology, a structured output result is generated, and the wiring harness diagram is automatically parsed.

2. The automatic parsing method for fusion of geometric field and target detection of the wireframe diagram according to claim 1, characterized in that, The target detection model uses the YOLOv8n model.

3. The automatic parsing method for fusion of geometric field and target detection of the wireframe diagram according to claim 1, characterized in that, Based on the distance geometry and the orientation geometry, an improved LSD algorithm is used to extract the wire harness segments, resulting in a set of wire harness segments, specifically including: Filter candidate pixels within the neighborhood of the beam whose distance value is less than a threshold in the distance geometry field; Candidate pixels are clustered according to similar directions in the directional geometry field to form multiple support regions with consistent orientations; Linear fitting is performed on each support region to obtain initial line segments, and the initial line segments are verified to obtain an initial set of line bundle segments; The initial set of wire harness segments is optimized to obtain the final set of wire harness segments.

4. The automatic parsing method for fusion of geometric field and target detection of the wireframe diagram according to claim 1, characterized in that, Based on the set of wire harness segments and the sub-image, a bandaging method classification model is used to obtain the bandaging method of the wire harness segments, specifically including: Determine the spatial position and orientation of each wire harness segment in the set of wire harness segments; Cropping of the sub-image at multiple preset positions along the direction of the wire harness segment yields multiple local images; Multiple local images are input into the bandaging method classification model for identification to obtain the bandaging method of the wire harness segment.

5. The automatic parsing method for fusion of geometric field and target detection of the wireframe diagram according to claim 1, characterized in that, The bandaging method classification model uses a lightweight convolutional neural network.

6. The automatic parsing method for fusion of geometric field and target detection of the wireframe diagram according to claim 1, characterized in that, Based on the target detection results, the distance geometry, and the orientation geometry, a wire harness topology is constructed, specifically including: Based on the target detection results, terminal candidate points are generated, and the distance geometry field is used to filter the terminal candidate points to obtain valid terminal candidate points; The geometric consistency of the effective terminal candidate points is evaluated based on the directional geometry field and the distance geometry field, and the connection score is calculated. The connection relationship of the valid terminal candidate points is determined based on the connection score, and the wire harness topology is constructed.

7. The automatic parsing method for fusion of geometric field and target detection of the wireframe diagram according to claim 6, characterized in that, The formula for calculating the connectivity score is: in, Rate the connection. As a distance consistency indicator, As an indicator of directional consistency, As a geometric field stability index, These are the weighting coefficients.

8. An automatic beam diagram parsing system integrating geometric field and target detection, characterized in that, include: The segmentation module is used to segment the wire harness image to be processed into multiple sub-images; The multi-object detection module is used to perform multi-object detection on each sub-image based on the object detection model to obtain the object detection results; An interference suppression module is used to perform differential masking on a specific region of the wire harness image to be processed based on the target detection result, so as to obtain an interference-suppressed wire harness image; the specific region is the area where the attachments and components are located in the wire harness image to be processed; The geometry field generation module is used to input the interference-suppressed wireframe diagram into a U-Net-like network to generate the range geometry field and orientation geometry field; The harness segment extraction module is used to extract harness segments based on the distance geometry field and the direction geometry field using an improved LSD algorithm to obtain a set of harness segments; A wire harness segment wrapping method identification module is used to obtain the wire harness segment wrapping method based on the set of wire harness segments and the sub-image, using a wrapping method classification model. The character recognition module is used to perform character recognition using OCR technology based on the target detection results, and obtain the component location number and wire harness segment size information; A wire harness topology construction module is used to construct a wire harness topology based on the target detection results, the distance geometry field, and the orientation geometry field. The structured output result generation module is used to integrate the wiring harness segment wrapping method, the component position number, the wiring harness segment size information, and the wiring harness topology to generate structured output results and complete the automatic parsing of the wiring harness diagram.

9. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the automatic parsing method for fusion geometry field and target detection according to any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the automatic parsing method for fusion geometry and target detection of any one of claims 1-7.

Images