Industrial drawing searching and comparing system
By using the industrial drawing search and comparison system and cloud computing and image feature extraction technology, the problems of low retrieval efficiency and low resource utilization in drawing management of mechanical processing enterprises have been solved. The system has achieved efficient drawing archiving and resource reuse, thereby improving production efficiency and quality stability.
Patent Information
- Application Number
- CN202411037680.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-31
- Publication Date
- 2024-11-15
AI Technical Summary
Mechanical processing enterprises face problems such as low retrieval efficiency, low resource reuse rate, high cost, long delivery time, and poor quality stability in drawing management. The lack of an efficient and intelligent drawing management and retrieval system restricts the development of the industry.
An industrial drawing search and comparison system was designed. It adopts a cloud computing platform, a drawing archiving framework module, a drawing search engine module, and an interface module. Through image feature extraction and neighbor graph indexing algorithms, it realizes rapid drawing retrieval and resource reuse, and supports efficient drawing archiving, searching, and business information association.
It enables efficient drawing archiving and searching without relying on drawing naming rules, simplifies the drawing management process, improves retrieval efficiency, ensures the association between drawings and business information, and promotes resource reuse and production process optimization.
Smart Images

Figure CN118964290A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the field of mechanical processing drawing retrieval, and in particular to an industrial drawing search and comparison system. Background Art
[0002] In the machining industry, customer drawings, price inquiries and machining services are an integral part of daily operations. With the intensification of market competition and the diversification of customer needs, the number of drawings received by enterprises has increased dramatically, and the types of drawings have become increasingly complex. However, the management of drawings by various machining enterprises has shown significant differences and fragmentation, lacking unified and efficient management standards and tools.
[0003] Traditionally, drawing management relies on manual input, paper archiving or simple electronic document management, which are not only inefficient but also prone to errors. As the number of drawings accumulates, when companies receive drawings from customers, they often find it difficult to quickly determine whether the drawing is a duplicate or has similar processing experience, resulting in the need to start from scratch every time to analyze the drawings, design the process, estimate the cost and quote the quotation. This process is not only time-consuming and labor-intensive, but also easily leads to waste of resources and rising costs.
[0004] What is more serious is that due to the irregular management of drawings, enterprises have accumulated a large number of business resources based on drawings in the long-term operation process, such as process plans, processing parameters, quality feedback, etc. These resources should have become valuable knowledge assets of the enterprise, but they have been shelved due to the lack of effective retrieval and reuse mechanisms. Whenever new drawings arrive, enterprises have to re-evaluate and design, ignoring the potential value of past experience, which not only prolongs the delivery cycle of products, but also increases the uncertainty of quality control, affecting the competitiveness and market reputation of enterprises.
[0005] In addition, with the rapid development of information technology, digitalization and intelligence have become an inevitable trend for the transformation and upgrading of the manufacturing industry. However, the current lagging state of mechanical processing enterprises in drawing management has seriously restricted their pace towards intelligent manufacturing. The lack of advanced drawing management and retrieval systems makes it difficult for enterprises to make substantial breakthroughs in data integration, process optimization, decision support, etc.
[0006] In summary, many problems in drawing management in existing mechanical processing enterprises, such as low retrieval efficiency, low resource reuse rate, high cost, long delivery time, poor quality stability, etc., have become bottlenecks restricting the development of the industry. Therefore, developing an efficient and intelligent drawing management and retrieval system to achieve rapid retrieval of drawings, effective reuse of resources and optimization of production processes is of great significance to improving the competitiveness of mechanical processing enterprises and promoting the transformation and upgrading of the industry. Summary of the invention
[0007] In order to overcome the deficiencies in the above-mentioned prior art, the present invention provides an industrial drawing search and comparison system. Based on the current and future field applications, the industrial drawing search and comparison system is designed and developed to realize searching with drawings and obtain their related business resource information for reuse, thereby improving the efficiency of subsequent quotation processes and other business operations and improving the stability of product quality.
[0008] To achieve the above object, the present invention provides the following technical solutions:
[0009] An industrial drawing search and comparison system, characterized by comprising:
[0010] A cloud computing system platform for completing drawing archiving, indexing and searching tasks;
[0011] Drawing archiving framework module for drawing file reception, drawing file splitting, drawing image standardization, image feature extraction, feature storage, indexing, and file storage;
[0012] A drawing search engine module for searching matching images through incoming files;
[0013] An interface module used to package and publish the above functions.
[0014] Furthermore, the cloud computing system platform uses several physical servers to build a server cluster as the basic hardware platform for cloud computing. The server uses the Linux system and dynamically allocates computing resources such as CPU, GPU, memory, and storage in the cluster through a resource scheduling management framework module.
[0015] Furthermore, in the storage drawing archiving framework module, new drawing data is received through a function interface or a system interface call. The source files received by the system are saved in the system storage, and the associated structured data are recorded. The newly received files are initially in a pending state. A multi-process data processing service is set up in the system. When obtaining a file in a pending state, it will automatically compete for the file processing authority. Through the docking matching rules, only one process will obtain a unique pending task to prevent duplicate processing or operation errors.
[0016] Furthermore, in the storage drawing archiving framework module, the task is assigned to the message queue of a certain node, and the relevant information of the task is recorded; the file splitting service reads the file in the queue and splits it by page, generates a new image file for each page of the source file, and then standardizes the image file through the image processing pipeline.
[0017] Furthermore, in the processing pipeline, the following steps are performed:
[0018] S1, extract image features from the overall image content,
[0019] S2, identifying the content area in the image and performing different recognition on different areas;
[0020] S3. Rearrange and organize the various extracted information through the feature combination editing system to form drawing image feature data that can comprehensively describe the global and local features of the drawing.
[0021] Furthermore, in step S2, OCR recognition is performed on the text description or digital annotation part to obtain the position and text and digital composite information data; graphic recognition is performed on the product shape image content part in multiple drawings, and after locking the position, relevant content is deducted to extract image features.
[0022] Furthermore, in the storage drawing archiving framework module, the image feature description file format is occupied to speed up the retrieval efficiency when using a search engine; a neighbor graph method is used for data indexing; and a relationship graph is pre-established to speed up the convergence speed during retrieval and reduce the amount of data points that need to be scanned, thereby improving the retrieval efficiency.
[0023] Furthermore, in the drawing search engine module, the incoming files are processed by the drawing archiving method to obtain the features of the file pages. The search engine matches the features in the index file using the neighbor graph indexing algorithm according to the file page features, and ranks them according to the average distance difference of the neighbor graph, and returns relevant information to the application in order of the result distance from small to large.
[0024] Furthermore, in the interface module, the business information associated with the drawing is associated by defining associated information attributes, and this information is reflected in the search results. The called system obtains the relevant business information through the information.
[0025] Furthermore, the interface module uses the WebAPI form.
[0026] Beneficial effects of the present invention:
[0027] Compared with the prior art, the industrial drawing search and comparison system described in the present invention has well solved the problem of not relying on drawing naming rules, achieved the technical effect of searching using drawings, and solved the problem of the difficulty in associating drawings with business information.
[0028] 1. Compared with the traditional drawing management system, the drawing search and comparison system greatly simplifies the archiving process. Drawings can be archived by simply passing in PDF files and the business order number associated with the drawing. There are no cumbersome naming rules or complex labeling systems. Simply transfer files and associated business order numbers.
[0029] 2. Graphic search and comparison function: once the drawings to be searched are input, the system will return search results sorted from high to low by similarity. The search results contain links to the result drawings, which is convenient for users to compare with the drawings to be searched. In addition, the system can identify the differences between the drawings to be searched and the result drawings, helping users to quickly determine whether the results are available.
[0030] Compared with the traditional drawing management system, which needs to use file names, tag information, etc. as keywords for searching, improper use of the rules formed by keywords during the search process will result in failure to obtain search results, resulting in low system availability. The drawing search and comparison system searches by image, which does not have this problem. This feature brings great convenience to users, and the business resources related to drawings in the business system can be reused, meeting the system design expectations. BRIEF DESCRIPTION OF THE DRAWINGS
[0031] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the present invention will be described in detail below in combination with the accompanying drawings and detailed embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:
[0032] Figure 1 This is an example of an industrial drawing of the present invention Figure 1 ;
[0033] Figure 2 This is an example of an industrial drawing of the present invention Figure 2 ;
[0034] Figure 3 This is an example of an industrial drawing of the present invention Figure 3 ;
[0035] Figure 4 It is a diagram illustrating the area of the drawings of the present invention;
[0036] Figure 5 It is a process diagram from PDF file to index file of the present invention;
[0037] Figure 6 A schematic flow chart for filing drawings of the present invention;
[0038] Figure 7 A schematic diagram of the neighbor graph indexing method of the present invention;
[0039] Figure 8 It is a schematic flow chart of the search engine of the present invention;
[0040] Fig. 9 This is a diagram of server configuration parameters in Example 2 of the present invention. DETAILED DESCRIPTION
[0041] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. Figure 1-8 Further explanation of the industrial drawing search and comparison system.
[0042] Currently in the field of mechanical processing, there is no effective means to search for drawings. This invention solves the problem of drawing retrieval.
[0043] In order to reduce the cost of processing products based on drawings, shorten delivery time and ensure quality, in addition to optimizing product processing technology, the most effective method is to use drawings to search in the drawing library. The search results can be associated with business information, so that previously generated business resources can be fully utilized to respond to new business, so as to reduce the cost of products based on drawings, shorten delivery time and ensure quality.
[0044] Example 1
[0045] This system is based on cloud computing architecture, and rationally plans and uses computer CPU, GPU, memory, storage space and other resources. Combined with independently developed industrial drawing feature extraction algorithms, feature storage algorithms, feature search algorithms, etc., it realizes a high-performance, high-reliability, and high-precision drawing archiving and search system. It can support more than 10 million drawings storage and supports up to 1024 clients in parallel. This system provides services in the form of WebAPI, which is easy to integrate with other systems.
[0046] 1.1. Definition of feature extraction algorithm for industrial drawings
[0047] Industrial drawings are the basis of engineering design and manufacturing, and are an important tool for engineers, designers, and manufacturing personnel to communicate and convey information. It includes various drawings and documents used to represent the design and manufacturing process of products. In the field of industrial manufacturing, drawings are very important because they contain all the details and specifications of the product and can guide workers on how to manufacture the product.
[0048] Types of Industrial Drawings
[0049] In industrial manufacturing, common industrial drawings include:
[0050] 1. Design drawings: used to show the design details and specifications of the product. Design drawings include floor plans, elevations, sections, and expansion drawings.
[0051] 2. Manufacturing drawings: used to show the manufacturing process and technology of the product. Manufacturing drawings include process drawings, process drawings, assembly drawings, etc.
[0052] 3. Component drawings: used to represent the components and assemblies of the product. Component drawings include parts drawings, standard parts drawings, visual assembly drawings, etc.
[0053] 4. Standard drawings: used to represent standardized parts and processes. Standard drawings include standard parts drawings, standard process drawings, etc.
[0054] 5. Control drawings: used to indicate product quality control and inspection. Control drawings include inspection drawings, test drawings, reliability drawings, etc.
[0055] How to make industrial drawings
[0056] Industrial drawings can be made by hand, computer and CAD. Hand drawing is the most traditional method, which requires drawing tools and drawing paper. Computer drawing refers to drawing using computer software, which can quickly and efficiently complete the drawing. CAD drawing is computer-aided design and drawing, which is currently the most popular method for making drawings.
[0057] Contents of industrial drawings
[0058] The contents of industrial drawings include the product's geometric shape, size, process requirements, materials and surface treatment requirements, etc. Industrial drawings should include the following:
[0059] 1. The geometric shape of the product: including the plane shape, elevation shape, cross-sectional shape, etc.
[0060] 2. Product dimensions: including product length, width, height, diameter, thickness, etc.
[0061] 3. Process requirements: including product manufacturing process, process parameters, process requirements, etc.
[0062] 4. Material requirements: including product materials, material standards, material specifications, etc.
[0063] 5. Surface treatment requirements: including product surface treatment methods, surface treatment requirements, surface quality, etc.
[0064] Standards for industrial drawings
[0065] The production of industrial drawings needs to comply with certain standards and specifications, mainly national standards, industry standards and enterprise standards. National standards are standards formulated and issued by the National Standardization Administration and are applicable nationwide. Industry standards are standards formulated and issued by relevant industry associations or organizations and are applicable to specific industries and fields. Enterprise standards are standards formulated and issued by enterprises themselves and are applicable within the enterprise.
[0066] Specifications for industrial drawings
[0067] like Figure 1-Figure 3 As shown, the production of industrial drawings needs to comply with certain specifications and requirements, mainly including drawing production requirements, drawing specifications, font specifications, drawing number specifications, etc. Drawing production requirements include the format, drawing method, annotation, annotation symbols, etc. Drawing specifications include the size, frame, border, folding method, etc. of the drawing. Font specifications include the font, font size, font spacing, font position, etc. of the drawing. Drawing number specifications include the code, version, revision, file name, etc. of the drawing.
[0068] The Importance of Industrial Drawings
[0069] Industrial drawings are the basis of engineering design and manufacturing, and are important tools for product development and manufacturing. They can clearly and accurately convey the design requirements, manufacturing requirements, and quality requirements of products. Industrial drawings can guide manufacturing personnel in product manufacturing, help quality inspectors in product inspection, and make product design, manufacturing, and quality control more standardized and efficient.
[0070] In short, industrial drawings are important tools for engineering design and manufacturing, including design drawings, manufacturing drawings, parts drawings, standard drawings, control drawings, etc. The methods of making industrial drawings include manual drawing, computer drawing and CAD drawing. The content of industrial drawings includes the geometric shape, size, process requirements, materials and surface treatment requirements of the product. The production of industrial drawings needs to comply with national standards, industry standards and enterprise standards, and needs to meet the requirements of drawing production, drawing specifications, font specifications, drawing number specifications, etc. Industrial drawings play an important role in product design, manufacturing and quality control, and can promote the progress of product development and manufacturing.
[0071] According to the characteristics of industrial drawings, different content areas are automatically divided, such as Figure 4 As shown, each area uses a different image and graphic recognition algorithm. For example, in the area describing digital and textual content, OCR technology is used to recognize the content and associate it with auxiliary information such as the position. The product image at the center of the drawing uses a shape recognition algorithm to extract a separate product graphic image. The image content features are extracted using an image recognition model pre-trained using deep learning technology. Combined with the different feature information of different areas of the entire drawing, a set of unique feature description data specifically describing the drawing and its content is comprehensively generated. The data format is specially designed by the present invention, which combines multiple data feature description sets and customized data description information and index data encoding formats. Search engine technology is used to customize the same and similar comparison algorithm, and multiple forms of retrieval methods and image content comparison and sorting algorithms are provided to meet retrieval needs in different scenarios.
[0072] The construction of the image recognition model adopts the technical method of AI deep learning, using CNN convolutional neural network, combined with the unique product shape features of the drawings, to create a CNN convolutional neural network structure suitable for the recognition of the unique product shape features. After that, the drawing data is randomly extracted by type in the drawing library, and the product graphic content in the drawing is deducted by the image graphic recognition algorithm to generate a new image. The generated images are used to form a training data set and a verification data set. After 1000 rounds of training and learning, an occupancy data model that can be applied to the recognition of the product content of the drawing is generated after achieving the expected effect. After testing and actual use, the recognition accuracy of the special model for drawing content recognition trained by the present invention is significantly higher than that of other open and public imagenet-based model products. The special model: constructs different recognition strategies according to the sample drawing type, and generally constructs the model according to the points, lines, and surfaces of the drawing elements. Monomer characters, icons, etc. are defined as points, non-closed lines are defined as lines, and closed lines are defined as surfaces.
[0073] The present invention deeply analyzes the data in the drawings, and uses a variety of technical means such as OCR and CNN to extract the content characteristics of different areas in the drawings, and converts the image data in the drawings into multi-dimensional data such as text, numbers, and image features. According to different needs such as application scenarios, the data is combined and arranged in a format specially designed by the present invention, forming a comprehensive global and local comprehensive drawing description data dedicated to the present invention. The data is compatible with standard high-dimensional data search engine technology. Combined with the search engine, all the drawing data of the present invention is processed in a unified pipeline. The specially designed format is used to store the image feature information collection results, and its content includes the classification of image information, location directory, feature content coding, etc. Figure 5 As shown, first, the drawings are extracted from the PDF archive file, and each page of data in the PDF file is generated as a drawing image file. After image area recognition and product shape recognition, different data of each area in the drawings and image recognition feature data are extracted. After a special data combination program, the extracted data and features constitute the special drawing data of the present invention. The search engine technology is used to compile and index the special drawing data of the present invention. The neighbor graph method, such as HNSW, SPTAG, ONNG, etc., is used to accelerate the convergence speed during retrieval by pre-establishing a relationship graph method, reduce the amount of data points that need to be scanned, and improve the retrieval efficiency. The special data combination program: the feature information extracted from the drawings is combined according to the location directory of the file feature collection points, and finally the coded data of the entire file is formed for storage.
[0074] 2.1 Research, develop and deploy cloud computing system platform
[0075] We use multiple physical servers to build a server cluster as the basic hardware platform for cloud computing. The server uses the Linux system to facilitate the deployment and application of self-developed services such as drawing storage, feature extraction, and drawing search. We also develop a self-developed resource scheduling management framework to dynamically allocate computing resources such as CPU, GPU, memory, and storage in the cluster.
[0076] The cloud computing platform uses open source TensorFlow, CUDA, Docker, Kubernetes, Redis and other mature libraries or frameworks, and the independently developed parallel task collaboration system to complete the processing of drawing archiving, index building, search and other tasks.
[0077] The parallel task collaboration system includes modules such as task generation, allocation, queue pool, computing thread pool management, computing data management, data processing pool management, and data persistence. The core algorithm uses Docker for deployment to generate a computing system image, and then uses Kubernetes to manage the synchronization of configurations related to each distributed process, management of each business node, operation monitoring, and routing control. The Redis framework is used to build the task queue pool and data processing pool modules to improve the efficiency of queue usage.
[0078] 2.2 R&D deployment drawing archiving framework
[0079] Drawing archiving includes drawing file receiving, drawing file splitting, drawing image standardization algorithm, image feature extraction algorithm, feature storage and indexing algorithm, file storage and other core functions. Figure 6 shown.
[0080] The system provides drawing archiving management function and a drawing entry interface. It receives new drawing data through the function interface or system interface call. Drawing data is usually in PDF format. The drawing PDF source files received by the system are saved in the system storage and record the associated structured data. The newly received files are initially in a pending state. A multi-process data processing service is set up in the system. When obtaining a file in a pending state, it will automatically compete for the processing authority of the file. Through the docking matching rules, only one process will obtain a unique pending task to prevent duplicate processing or operation errors.
[0081] Assign the task to the message queue of a certain node and record the relevant information of the task. The file splitting service reads the file in the queue and splits it by page. Usually, a new image file is generated for each page of the PDF source file, and then the image file is standardized through the image processing pipeline. The processing pipeline includes multiple processing processes. First, the image feature extraction is performed on the overall image content. Then, the content area in the image is identified, and different areas are identified differently, such as OCR recognition of text descriptions or digital annotations, obtaining location and text, numbers and other composite information data, and graphic recognition of product shape image content parts in multiple drawings. After locking the position, relevant content is deducted, and special image feature extraction is also performed. Finally, the feature combination arrangement system developed by the present invention is used to rearrange and organize the extracted multiple information to form the drawing image feature data that can depict the comprehensive description of the global and local features of the drawing, which is unique to the present invention.
[0082] The present invention designs an occupied image feature description file format. In order to use a search engine to speed up the retrieval efficiency, the present invention sets compatibility between the dedicated image feature description data and the search engine's index data format, and uses a neighbor graph method with high retrieval efficiency and high accuracy to index data. By pre-establishing a relationship graph, the convergence speed during retrieval is accelerated, and the amount of data points that need to be scanned is reduced to improve the retrieval efficiency. Figure 7 shown.
[0083] 2.3 Research and Development of Blueprint Search Engine
[0084] The self-developed drawing search engine searches for matching images by importing files. The search engine schematic process is as follows Figure 8 shown.
[0085] The incoming file is processed (the processing algorithm for the incoming file is the same as the drawing archiving algorithm) to obtain the file page features. The search engine uses the self-developed neighbor graph indexing algorithm to match the features in the index file according to the file page features, and ranks them according to the average distance difference of the neighbor graph, and returns the relevant information to the application in the order of the result distance from small to large. The search algorithm can match the deflection of 90°, 180°, and 270° of the image, and supports tilt within 5°.
[0086] 2.4 Interface
[0087] Use WebAPI to package and publish related functions. The system has good boundaries and can be integrated with any ERP, MES and other systems.
[0088] The drawing search and comparison system has defined associated information attributes, which are used to associate the business information associated with the drawing. This information will be reflected in the search results. The called system can obtain relevant business information through this information to achieve seamless business integration.
[0089] Example 2
[0090] The system deployment uses two servers with minimum configuration, A and B. The configurations of the two servers are as follows: Fig. 9 shown.
[0091] The two servers are installed with Ubuntu 22.04.1LTS operating system, TensorFlow, CUDA, Docker, Kubernetes, Redis and other related frameworks and libraries, and the self-developed graphics feature extraction service, search engine and task scheduling management are deployed on the two servers respectively. The load balancing of the two services is achieved by deploying a reverse proxy. Through the above deployment, a drawing search and comparison system is built.
[0092] Develop a Windows service to read PDF drawings and related information in the business system, extract their features and write them into index files according to index rules. The original drawings are written into the corresponding storage structure for storage by the file management system. At this point, the drawings are archived.
[0093] On the other end, the business system application passes the drawings to be processed to the graphic search engine through the interface. The search engine extracts the features of the drawings and uses the drawing features to search and match in the index. The search results within the threshold range are evaluated according to the evaluation algorithm to obtain the best results. Finally, the search engine returns the results to the business system. The business system displays the search results and uses the associated information to extract the business resources associated with the drawings in the search results, thereby realizing resource reuse and reducing production costs.
[0094] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent substitutions or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention. Without departing from the concept of the present invention, simple deductions and substitutions can be made, which should be covered by the protection scope of the present invention.
Claims
1. An industrial drawing search and comparison system, characterized in that: include: A cloud computing system platform for completing drawing archiving, indexing and searching tasks; Drawing archiving framework module for drawing file reception, drawing file splitting, drawing image standardization, image feature extraction, feature storage, indexing, and file storage; A drawing search engine module for searching matching images through incoming files; An interface module used to package and publish the above functions.
2. The industrial drawing search and comparison system according to claim 1, characterized in that: The cloud computing system platform uses several physical servers to build a server cluster as the basic hardware platform for cloud computing. The server uses the Linux system and dynamically allocates computing resources such as CPU, GPU, memory, and storage in the cluster through the resource scheduling management framework module.
3. The industrial drawing search and comparison system according to claim 1, characterized in that: In the storage drawing archiving framework module, new drawing data is received through a function interface or a system interface call. The source files received by the system are saved in the system storage, and the associated structured data are recorded. The newly received files are initially in a pending state. A multi-process data processing service is set up in the system. When obtaining a file in a pending state, it will automatically compete for the file processing authority. Through the docking matching rules, only one process will obtain a unique pending task to prevent duplicate processing or operation errors.
4. The industrial drawing search and comparison system according to claim 1, characterized in that: In the storage drawing archiving framework module, the task is assigned to the message queue of a certain node and the relevant information of the task is recorded; the file splitting service reads the file in the queue and splits it by page, generates a new image file for each page of the source file, and then the image file is standardized through the image processing pipeline.
5. The industrial drawing search and comparison system according to claim 4, characterized in that: In the processing pipeline, the following steps are performed: S1, extract image features from the overall image content, S2, identifying the content area in the image and performing different recognition on different areas; S3. Rearrange and organize the various extracted information through the feature combination editing system to form drawing image feature data that can comprehensively describe the global and local features of the drawing.
6. The industrial drawing search and comparison system according to claim 5, characterized in that: In step S2, OCR recognition is performed on the text description or digital annotation part to obtain the position and text and digital composite information data; graphic recognition is performed on the product shape image content part in multiple drawings, and the relevant content is deducted after the position is locked to extract image features.
7. The industrial drawing search and comparison system according to claim 1, characterized in that: In the storage drawing archiving framework module, the image feature description file format is occupied to speed up the retrieval efficiency when using the search engine; the neighbor graph method is used for data indexing; and the method of pre-establishing a relationship graph is used to speed up the convergence speed during retrieval and reduce the amount of data points that need to be scanned, so as to improve the retrieval efficiency.
8. The industrial drawing search and comparison system according to claim 1, characterized in that: In the drawing search engine module, the incoming files are processed by the drawing archiving method to obtain the features of the file page. The search engine uses the neighbor graph indexing algorithm to match the features in the index file according to the file page features, and ranks them according to the average distance difference of the neighbor graph. The relevant information is returned to the application in the order of the result distance from small to large.
9. The industrial drawing search and comparison system according to claim 1, characterized in that: In the interface module, the business information associated with the drawing is associated by defining associated information attributes, and this information is reflected in the search results. The called system obtains the relevant business information through the information.
10. The industrial drawing search and comparison system according to claim 1, characterized in that: The interface module uses the WebAPI form.
