A method and device for individual customization of a door and window profile

Abstract

The application provides a door and window profile personalized customization method and device, relates to the technical field of door and window profile personalized customization, and has the technical solution points that unstructured data submitted by a user is received; structured technical parameters are generated; based on the structured technical parameters, a parameterized model library is called to generate a three-dimensional model and production data package of the door and window profile; a scheduling plan for small-batch and multi-variety production is generated according to the production data package and production resource state information; a unique identification code is allocated to a door and window profile component in a production process, and production execution data associated with the unique identification code is collected at a key node; the production execution data is updated to a data management center in real time, and a shared view of an order state is provided to a user, a supplier and a downstream service provider. The door and window profile personalized customization method and device have the advantages that user demand data can be automatically processed, and the data conversion accuracy is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of door and window profile personalized customization, in particular, to a door and window profile personalized customization method and device. BACKGROUND

[0002] As an important component of building structure, door and window profiles have long been produced and manufactured mainly by standardized extrusion process to realize large-scale production of products of the same specification in large quantities. This mode has cost advantages and efficiency in meeting the demand of traditional building projects for relatively uniform product specifications. However, with the continuous development of modern architectural design concepts and the increasing demand of end users for personalized, comfortable and functional living and working environments, the market demand for door and window profiles is no longer limited to standard specifications. The personalized requirements proposed by users show a high degree of diversification and complexity, which may involve non-standard sizes, unique cross-sectional geometries (including complex cavity structures, chamfers or grooves at specific positions), customized surface treatment effects, and even integrated soundproofing, heat insulation, anti-theft, ventilation, intelligent and other specific functional modules.

[0003] The traditional order processing and information flow system centered on standardized production is difficult to effectively support this growing demand for personalized customization. The personalized demand data submitted by users is usually unstructured and may exist in the form of oral description, reference pictures (such as photos, renderings), hand-drawn sketches or non-standard format files (such as PDF, simple CAD sketches). These unstructured raw demand data cannot be directly recognized and processed by existing standardized order systems or design and production management software, and a large amount of analysis, understanding and structured conversion work needs to be done manually. Accurately and losslessly converting these unstructured and highly personalized user intentions into technical parameters required in the design and production stages is a significant technical challenge. For example, accurately extracting the profile cross-sectional contour, key dimensions, wall thickness distribution, installation position of specific functional modules and interface parameters from a picture containing perspective distortion, uneven lighting or blurred details requires extremely high professional experience and meticulous manual operation. In this conversion process, due to subjective understanding, information omission or poor communication, information loss or understanding deviation is extremely easy to occur, which directly leads to a decrease in the accuracy of subsequent design parameters and increases the probability and cost of design rework.

[0004] When applying the preliminary structured technical parameters to the specific design, the designer faces the work of frequent communication confirmation with the user and modification of the scheme. Especially when the customization requirements involve complex profile structures, non-traditional connection node designs, or the need to integrate specific functional modules into the profile internal structure, the design process becomes extremely complex. The number of iterations of the design scheme is often large, and each modification requires repeated communication confirmation with the user, which not only significantly increases the communication cost, but also leads to a long design cycle and is unable to quickly respond to market demand.

[0005] Once the design scheme is finally confirmed by the user and solidified into a set of detailed technical parameters, the next step is to convert it into detailed instructions and data required for production and manufacturing. This includes but is not limited to machining path instructions for numerical control machine tools (such as cutting machines, drilling and milling centers), design parameters for custom molds for special sections or structures, material type, specification, and quantity list (BOM) accurate to each component, and quality inspection standards specific to individual orders. Unlike standardized production, where a set of production data can be repeatedly applied to large quantities of products, each individual customized order may correspond to a unique, non-repeating production data package. Manually or semi-automatically generating, verifying, and managing these unique production data for each customized order is a huge workload and prone to errors, resulting in a significant increase in production preparation period and delaying the delivery of orders.

[0006] The characteristics of individualized customization orders "small batch, multi-variety" or even "single piece customization" bring great complexity to production scheduling. Traditional production line optimization configuration and scheduling strategies mainly serve standardized production with large quantities and few varieties, and the goal is to maximize efficiency by reducing equipment line changes, mold changes, and process adjustments. The characteristics of individualized customization orders are that the requirements of each order are very different, which may involve different profile sections, different processing procedures, different equipment requirements, and different raw materials. Under the constraint of limited production resources, how to efficiently arrange these orders with great differences in the production plan while meeting the unique needs of each order, minimize equipment idle time, optimize the use of raw materials, shorten line change and adjustment time, and ensure that all orders are completed within the promised delivery period, becomes an extremely complex management and optimization problem, and traditional scheduling methods are difficult to cope with.

[0007] In addition, personalized customization puts higher requirements on part traceability and information coordination in the production process. In the small-batch and multi-variety production mode, parts in the same batch or the same order may be split or combined due to process needs, such as cutting long profiles into short materials or assembling multiple small parts into one assembly. The traditional batch or work order-based traceability method cannot be fine-tuned to the level of individual parts, and cannot accurately track the production history, processing status, quality information of each part, and its source and destination in the splitting and combining process. This not only brings difficulties to quality control and problem tracing, but also makes the collection and management of production execution data complex. At the same time, due to the uncertainty in the personalized order production process (such as design modification, production anomaly, part splitting and combining), it is difficult to provide accurate and real-time production progress and expected delivery information to raw material suppliers, profile surface treatment service providers, and downstream door and window assembly manufacturers and installation service providers, affecting the coordination efficiency and response speed of the entire supply chain and increasing the risk.

[0008] Throughout the personalized customization process of door and window profiles, users generally expect to be able to track the progress of their orders in real time, including whether the demand is understood, the modification status of the design scheme, the procurement of raw materials, the specific steps of production and manufacturing (such as extrusion, surface treatment, processing, packaging), and the expected delivery time, etc. However, existing order management systems often lack such fine-grained real-time information tracking and feedback capabilities, with information updates not timely or transparent, and users unable to easily obtain the information they need, resulting in poor user experience, increased user consultation frequency and communication costs for enterprises, and increased uncertainty.

[0009] In view of the above series of challenges encountered in the personalized customization process of door and window profiles, although some enterprises try to accumulate experience through manual means, optimize internal processes, or introduce simple software tools to assist in managing some aspects, this is difficult to cope with the growing, larger-scale, and more complex personalized customization needs. This approach is inefficient, prone to error, and difficult to scale and apply within an enterprise, becoming a bottleneck that restricts the further development of personalized customization of door and window profiles. A systematic solution is urgently needed to integrate and optimize the information flow and business flow of the entire personalized customization process of door and window profiles, especially to efficiently handle unstructured user needs, accurately convert them into customized technical parameters, achieve flexible scheduling and fine-grained traceability for small-batch / single-piece production, and ensure real-time coordination and transparent sharing of information throughout the process among multiple parties (including users, design, production, and supply chain).

[0010] In view of the above problems, the existing technology needs to be improved. SUMMARY

[0011] The purpose of the present application is to provide a door and window profile personalized customization method and device, which has the advantages of automatically processing user demand data, improving data conversion accuracy, automatically generating design and production data, optimizing small-batch multi-variety production scheduling, realizing production process component traceability, improving production information transparency, and promoting supply chain information collaboration.

[0012] In the first aspect, the present application provides a door and window profile personalized customization method, and the technical solution is as follows:

[0013] Receiving unstructured data submitted by a user, which contains personalized requirements of door and window profiles;

[0014] Analyzing the unstructured data to generate structured technical parameters;

[0015] Based on the structured technical parameters, calling a parameterized model library to generate a three-dimensional model of the door and window profile and a production data package;

[0016] Generating a scheduling plan for small-batch multi-variety production according to the production data package and production resource state information;

[0017] Assigning a unique identification code to the door and window profile components in the production process, and collecting production execution data associated with the unique identification code at key production nodes;

[0018] Real-time updating the production execution data to a data management center, and providing a shared view of the order status to users, suppliers, and downstream service providers.

[0019] Further, in the present application, analyzing the unstructured data to generate structured technical parameters includes:

[0020] Receiving unstructured data submitted by a user, which contains reference pictures of door and window profiles;

[0021] Image processing the reference pictures to extract the cross-sectional profile and key feature points of the door and window profiles;

[0022] Matching the cross-sectional profile with the basic models in the preset profile parameterized model library to determine the closest basic model;

[0023] Based on the basic model, combining the cross-sectional profile and key feature points to generate a parameterized profile model of the door and window profile;

[0024] Extracting detailed features in the reference pictures, including cavity structure, wall thickness distribution, connection nodes, and functional areas;

[0025] Converting the detailed features into corresponding technical parameters and associating them with the parameterized profile model;

[0026] A structured set of parameters for a production process is generated, which includes geometric dimensions, structural features and functional requirements.

[0027] Further, in the present application, when the door and window profile components are split or merged during the production process, the unique identification code is assigned to the door and window profile components during the production process, and the production execution data associated with the unique identification code is collected at the key nodes of the production process.

[0028] A main unique identification code containing order information, profile type and serial number is assigned to each initial door and window profile component;

[0029] When a splitting operation of the door and window profile component is detected, a sub-unique identification code containing the main unique identification code and the serial number of the sub-component is assigned to each sub-component after splitting;

[0030] When a merging operation of multiple sub-components is detected, a merged unique identification code containing all sub-unique identification code information is assigned to the merged component;

[0031] A hierarchical association relationship is established among the main unique identification code, the sub-unique identification code and the merged unique identification code;

[0032] Production execution data associated with each level of unique identification code is collected at key nodes of the production process;

[0033] The production execution data is stored in association with the corresponding unique identification code, and a complete traceability chain of component splitting and merging during the production process is established.

[0034] Further, in the present application, the main unique identification code containing order information, profile type and serial number is assigned to each initial door and window profile component, which includes:

[0035] Production environment information containing production line identification and timestamp is obtained;

[0036] The identification code assignment record of the current active order is obtained from the data management center;

[0037] Based on the production environment information and the identification code assignment record, an identification code prefix containing production line identification, timestamp code, order number and component type code is generated;

[0038] A batch number is assigned to each production batch, and the batch number is combined with the identification code prefix;

[0039] According to the sequence position of the component in the batch, a serial number is generated;

[0040] The identification code prefix, the batch number and the serial number are combined to generate the main unique identification code;

[0041] Storing the main unique identification code in association with the technical parameters of the corresponding door and window profile component;

[0042] Submitting an allocation record of the main unique identification code to the data management center and verifying global uniqueness.

[0043] Further, in the present application, the image processing of the reference picture to extract the cross-sectional profile and key feature points of the door and window profile includes:

[0044] Detecting perspective deformation features in the reference picture, including detecting the straight edge bending degree and parallel line convergence degree of the reference object;

[0045] According to the detected perspective deformation features, calculating a perspective transformation matrix;

[0046] Correcting the reference picture using the perspective transformation matrix to generate an orthographic image;

[0047] Applying an edge detection algorithm on the orthographic image to extract the cross-sectional profile line of the door and window profile;

[0048] Noise filtering and breakpoint connection processing on the cross-sectional profile line;

[0049] Identifying corner points, intersection points and curve inflection points on the processed cross-sectional profile line and marking them as key feature points;

[0050] Establishing a proportional relationship between the pixel coordinates in the orthographic image and the actual physical dimensions for subsequent size parameter calculation.

[0051] Further, in the present application, when it is detected that the door and window profile component splitting operation is performed in different stations or different time periods in steps, the allocation of a sub-unique identification code containing a main unique identification code and a sub-component serial number for each sub-component after splitting includes:

[0052] Obtaining process rules and splitting plan information of the splitting operation from the data management center;

[0053] Based on the process rules and splitting plan information, generating a splitting task table containing the total number of splitting steps and the station information of each step;

[0054] Allocating a unique splitting task identification for the splitting task table;

[0055] Storing the splitting task identification in association with the main unique identification code;

[0056] After each splitting station completes the corresponding splitting step, recording the number and type of sub-components produced by the current station;

[0057] According to the number and type of sub-components, combined with the current station position in the split task table, a sub-component serial number prefix containing the station code and split step serial number is generated;

[0058] The main unique identification code is combined with the sub-component serial number prefix and the sequence number of the sub-component at the current station to generate a sub-unique identification code;

[0059] The sub-unique identification code is associated with the technical parameters of the corresponding sub-component and the current split state information for storage;

[0060] When the sub-component is transferred to the next split station, the split task identification and the current split state are obtained by scanning the sub-unique identification code;

[0061] Verify whether the current station is the next expected station in the split task table, and continue to execute the subsequent split step after verification.

[0062] Further, in the present application, the detection of the perspective deformation features in the reference picture includes detecting the straight edge bending degree and parallel line convergence degree of the reference object, comprising:

[0063] Receiving a reference picture containing a plurality of different angle door and window profiles;

[0064] Pretreating the reference picture to improve image contrast and clarity;

[0065] Detecting all possible door and window profile regions in the reference picture and extracting edge features of each region;

[0066] Extracting user marking information or reference scale information from the reference picture;

[0067] According to the user marking information or reference scale information, determine the target door and window profile region;

[0068] If no user marking information or reference scale information is detected, calculate the clarity, integrity and size ratio scores of each door and window profile region according to the preset rules;

[0069] Select the best target door and window profile region based on the scores;

[0070] Detecting straight edges in the target door and window profile region and calculating the bending degree of the straight edges;

[0071] Detecting parallel lines in the target door and window profile region and calculating the convergence degree of the parallel lines;

[0072] Determine the degree of perspective deformation according to the bending degree and the convergence degree.

[0073] Further, in the present application, when it is detected that the split sub-components need to be temporarily assembled for testing before being split again, the assigning of a sub-unique identification code containing the master unique identification code and the sub-component serial number for each split sub-component includes:

[0074] creating a temporary assembly task record containing temporary assembly information and testing requirements;

[0075] assigning a temporary assembly identification code containing information of all sub-unique identification codes participating in the assembly to the temporarily assembled combination;

[0076] establishing a mapping relationship between the sub-unique identification code and the temporary assembly identification code in the temporary assembly task record;

[0077] after the testing is completed, retrieving the temporary assembly task record to obtain the original sub-unique identification code information;

[0078] verifying the consistency of the sub-components after being split again with the original sub-components;

[0079] if the verification passes, keeping the atomic unique identification code unchanged;

[0080] if the verification fails, assigning a new sub-unique identification code containing the atomic unique identification code and a change marker to the changed sub-component;

[0081] updating the sub-component mapping relationship table in the data management center to record the temporary assembly and re-splitting process experienced by the sub-component.

[0082] Further, in the present application, when there is a reflection or a shadow on the surface of the door and window profile in the reference picture, the image processing of the reference picture to extract the cross-sectional profile and key feature points of the door and window profile includes:

[0083] detecting the reflection area and the shadow area in the reference picture to generate a reflection and shadow distribution map;

[0084] segmenting the reference picture into a normal area, a reflection area and a shadow area according to the reflection and shadow distribution map;

[0085] applying an exposure compensation algorithm to the reflection area to reduce the brightness value of the overexposed area;

[0086] applying a brightness enhancement algorithm to the shadow area to improve the visibility of the dark area;

[0087] image fusion of the processed reflection area and shadow area with the normal area to generate an image with balanced lighting;

[0088] applying an edge detection algorithm on the image with balanced lighting to extract an initial contour line;

[0089] The initial profile lines are breakpoint connected and smoothed to generate a complete door and window profile section profile;

[0090] Corner points, intersection points and curve inflection points on the complete door and window profile section profile are identified and marked as key feature points.

[0091] In a second aspect, the application also provides a door and window profile individual customization device for realizing information collaboration in the door and window profile individual customization process, and the system comprises:

[0092] A receiving module is configured to receive unstructured data submitted by a user and containing individual customization requirements of a door and window profile;

[0093] An analysis module is configured to analyze the unstructured data and generate structured technical parameters;

[0094] A model generation module is configured to generate a three-dimensional model of the door and window profile and a production data package based on the structured technical parameters and by calling a parameterized model library;

[0095] A scheduling module is configured to generate a scheduling plan for small-batch and multi-variety production according to the production data package and production resource state information;

[0096] An identification tracking module is configured to assign a unique identification code to a door and window profile component in a production process and collect production execution data associated with the unique identification code at key nodes in the production process;

[0097] A data sharing module is configured to update the production execution data to a data management center in real time and provide a shared view of an order state to a user, a supplier and a downstream service provider.

[0098] As can be seen from the above, the door and window profile individual customization method provided by the application has the advantages that user requirement data is automatically processed, data conversion accuracy is improved, design and production data are automatically generated, small-batch and multi-variety production scheduling is optimized, production process component tracing is realized, production information transparency is improved, and supply chain information collaboration is promoted. BRIEF DESCRIPTION OF DRAWINGS

[0099] Figure 1 A flowchart of the door and window profile individual customization method provided by the application.

[0100] Figure 2 A structural diagram of the door and window profile individual customization device provided by the application.

[0101] In the figure: 210, receiving module; 220, parsing module; 230, model generation module; 240, scheduling module; 250, identification tracking module; 260, data sharing module. DETAILED DESCRIPTION

[0102] The technical solutions in the present application will be described in detail below with reference to the accompanying drawings in the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. The components of the present application described and shown in the accompanying drawings can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0103] It should be noted that: similar reference numbers and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. At the same time, in the description of the present application, the terms "first", "second" and the like are only used to distinguish the description, and cannot be understood as indicating or implying relative importance.

[0104] Reference Figure 1 The present application proposes a door and window profile individual customization method, comprising:

[0105] S110, receiving user-submitted unstructured data containing door and window profile individual customization requirements;

[0106] S120, parsing the unstructured data to generate structured technical parameters;

[0107] S130, based on the structured technical parameters, calling the parameterized model library to generate a three-dimensional model of the door and window profile and a production data package;

[0108] S140, generating a scheduling plan for small-batch multi-variety production according to the production data package and production resource state information;

[0109] S150, assigning a unique identification code to the door and window profile components in the production process, and collecting production execution data associated with the unique identification code at the key nodes of production;

[0110] S160, updating the production execution data to the data management center in real time, and providing a shared view of the order status to the user, the supplier and the downstream service provider.

[0111] Among them, receiving user-submitted unstructured data containing door and window profile individualization needs can be achieved by a data interface module, which can handle multiple input formats such as text description, image file, sketch file, etc. Thus, the system can obtain the user's diversified customization intention.

[0112] Among them, the unstructured data is parsed to generate structured technical parameters, which can be achieved by a parsing engine. The engine extracts key information from unstructured input through natural language processing, image recognition, etc. and converts it into predefined data fields such as profile cross-sectional size, wall thickness, cavity structure, surface treatment requirements, functional module type, etc.

[0113] Based on the structured technical parameters, the parameterized model library is called to generate the three-dimensional model and production data package of the door and window profile, which can be achieved by a model generator. The generator selects the basic model or template from the parameterized model library according to the structured parameters and adjusts the parameters to generate the customized three-dimensional model. At the same time, the generator automatically generates production data package including processing instructions, material list, quality standards, etc. according to the model and parameters.

[0114] According to the production data package and the production resource state information, the scheduling plan for small-batch multi-variety production is generated, which can be achieved by a scheduling optimizer. The optimizer considers the production data package requirements of different orders (such as equipment, process, time) and the real-time state of current production resources (such as equipment availability, personnel configuration, raw material inventory), and uses optimization algorithms to generate production scheduling that meets delivery period and efficiency targets.

[0115] A unique identification code is assigned to the door and window profile components in the production process, and production execution data associated with the unique identification code is collected at key nodes of production, which can be achieved by an identification tracking system. The system assigns a globally unique identification code to the component when it enters the production process, and collects production execution data such as station, time, operator, quality detection results, etc. through scanning the identification code at key stations such as extrusion, surface treatment, processing, assembly, etc.

[0116] Real-time production execution data is updated to the data management center, and a shared view of order status is provided to users, suppliers and downstream service providers, which can be achieved by a data management and sharing platform. The platform receives real-time data from the identification tracking system and stores and manages it. The platform displays order status information related to authorized users (users, suppliers, downstream service providers) through different interfaces or views, such as design progress, production stage, estimated completion time, etc.

[0117] Specifically, the method first receives unstructured data containing customized door and window profile requirements submitted by users through a network interface, such as profile photos, hand-drawn sketches, or text description files uploaded by users.

[0118] The received unstructured data can use image recognition technology to extract the cross-sectional profile and key dimension information from the photos, and use natural language processing technology to identify functional requirements and surface treatment preferences from the text descriptions.

[0119] As a result, the user's vague requirements are converted into clear, standardized technical parameter sets, such as profile cross-section type codes, length and width dimensions, wall thickness values, cavity numbers, surface color codes, and thermal insulation performance levels.

[0120] According to the technical parameter set, the corresponding parameterized model template is retrieved from the pre-set parameterized model library, and the model is adjusted according to the parameter values to generate an accurate three-dimensional model of the customized profile. At the same time, based on the three-dimensional model and technical parameters, a production data package containing numerical control machining instructions, material quantity list, quality inspection specifications, etc. is automatically generated.

[0121] The production data package and the current production line equipment state, personnel scheduling, raw material inventory, and other production resource state information are input to the scheduling, considering the process flow differences and resource requirements of different customized orders, to generate an optimal production scheduling plan to determine the start time, end time, equipment used, and workstation order for each order.

[0122] During production execution, each profile component entering the production line is assigned a unique two-dimensional code or RFID tag as an identification code. At key production nodes such as extrusion molding, cutting, drilling, and surface spraying, the identification code on the component is scanned to collect production execution data at the current workstation, such as operator ID, completion time, equipment parameters, test results, etc. As a result, the production history and status of each component are recorded.

[0123] The collected production execution data is sent to the data management center in real time for storage and management. The data management center updates the production status of each order based on these real-time data. Order status information is obtained from the data management center and shared views of order status are provided to users who submitted the orders, suppliers who provided raw materials, and downstream service providers who are responsible for subsequent assembly and installation through user interfaces, API interfaces, etc.

[0124] Users can view the design progress, production stage, and estimated delivery date of the order; suppliers can understand the demand and delivery time of raw materials; and downstream service providers can arrange assembly and installation plans in advance.

[0125] In some embodiments, a user uploads a JPG file containing a photo of the cross-section of a door and window profile through a webpage. The photo is analyzed to identify the profile line of the profile and measure the key dimensions, such as the profile width of 100 mm, the height of 50 mm, and the wall thickness of 2 mm. At the same time, it is identified that the profile in the photo has three cavity structures. These information is converted into structured technical parameters: {“profile width”: 100, “profile height”: 50, “wall thickness”: 2, “cavity number”: 3, “surface color”: “white”}.

[0126] According to these parameters, a basic three-cavity profile model is retrieved from the parameterized model library, and its size parameters are adjusted to generate a three-dimensional model of the customized profile. At the same time, a production data package containing cutting length, drilling position, and spraying color instructions is generated. The scheduling module receives the production data package and the current available state of the extruder, spraying line, and machining center, calculates the best production time period and equipment allocation for the order, such as scheduling on Wednesday morning using No. 1 extruder and No. 2 spraying line. The identification tracking module assigns a unique two-dimensional code, such as “ORD123-EXT-001”, to each long profile after the profile is extruded.

[0127] At the cutting station, the operator scans the two-dimensional code, and the system records the completion of the cutting operation. If the cut short material needs further processing, a new sub-identification code will be assigned and associated with the main identification code. Production execution data, such as cutting time and cutting length, are uploaded to the data management center in real time. The data management center updates the order status to “processing”. The data sharing module allows users to view the order status through the mobile application, displaying “your order has entered the processing stage and is expected to be completed on Thursday”.

[0128] The application further proposes receiving user-submitted unstructured data containing door and window profile reference pictures; performing image processing on the reference pictures to extract the cross-sectional profile and key feature points of the door and window profile; matching the cross-sectional profile with the basic models in the preset profile parameterized model library to determine the approximate basic model; based on the basic model, combining the cross-sectional profile and key feature points to generate a parameterized profile model of the door and window profile; extracting detailed features from the reference pictures, including cavity structure, wall thickness distribution, connection node, and functional area; converting the detailed features into corresponding technical parameters and associating them with the parameterized profile model; generating a structured technical parameter set containing geometric dimensions, structural features, and functional requirements.

[0129] Among them, receiving user-submitted unstructured data containing door and window profile reference pictures obtains the user's personalized demand information for the door and window profile.

[0130] The reference picture is image-processed to extract the cross-sectional profile and key feature points of the door and window profile, thereby converting the image information into computable geometric feature data, improving the objectivity and efficiency of data extraction.

[0131] The cross-sectional profile is matched with a basic model in a preset profile parameterization model library to determine the approximate basic model, thereby providing a starting point for subsequent parameterization modeling and reducing the complexity of modeling.

[0132] Based on the basic model, the cross-sectional profile and key feature points are combined to generate a parameterized profile model of the door and window profile, thereby constructing a geometric model that can accurately reflect the shape shown in the picture.

[0133] The detailed features in the reference picture are extracted, including the cavity structure, wall thickness distribution, connection nodes, and functional areas, thereby obtaining more comprehensive profile technical information.

[0134] The detailed features are converted into corresponding technical parameters and associated with the parameterized profile model, thereby converting non-geometric features into structured data and establishing a connection with the geometric model.

[0135] A set of structured technical parameters including geometric dimensions, structural features, and functional requirements is generated, thereby forming a standard data format that can be directly used for subsequent design and production.

[0136] Specifically, the technical solution solves the problem of various forms of user demand expression by receiving unstructured data containing door and window profile reference pictures. The received reference pictures are image-processed, such as applying an edge detection algorithm to extract the cross-sectional profile line of the profile, and identifying key feature points such as corner points and intersection points on the profile line. Thereby, visual information is converted into accurate geometric data. The extracted cross-sectional profile is compared and matched with a basic model in a preset profile parameterization model library, such as through shape descriptors or feature point alignment, to determine the basic model most similar to the picture profile. Based on the determined basic model, the accurate profile and key feature point information extracted from the picture are used to adjust the parameters of the basic model to generate a parameterized profile model highly consistent with the picture.

[0137] Further, the picture can be analyzed in more depth, such as based on user annotations on the picture, identifying the cavity structure inside the profile, wall thickness at different positions (user annotations), identifying the geometric features of the connection nodes, and possible functional areas (such as slots for installing specific fittings). These detailed features are converted into specific numerical parameters or descriptive parameters, such as the number of cavities, the size of each cavity, the wall thickness value at different positions, the geometric parameters of the connection slots, etc. These technical parameters are stored in association with the previously generated parameterized profile model.

[0138] Finally, all the extracted and converted geometry dimensions, structural features, and functional requirements, etc. are integrated into a structured technical parameter set, for example, output in JSON or XML format, containing the complete technical description of the profile.

[0139] Through this series of steps, this scheme realizes the automatic and accurate extraction of technical parameters of door and window profiles from unstructured data sources such as pictures, and converts them into structured data, effectively solving the problem of low efficiency and easy error of manual processing of picture data, providing accurate data basis for subsequent parameterized modeling and production preparation, and reducing the risk of information loss and understanding bias.

[0140] In some specific embodiments, a JPEG format picture containing a side view of a door and window profile uploaded by a user is received. Image processing algorithms are applied to the picture, such as first grayscale and contrast enhancement, and then the Canny edge detection algorithm is used to extract the edge lines in the image. The extracted edge lines are subjected to Hough transform to identify straight line segments, and disconnected line segments are connected to form a complete cross-section contour. Identify points with large curvature changes on the contour line as key feature points. The extracted cross-section contour is matched with the preset profile base models (such as parameterized templates of standard series profiles) stored in the database, for example, using shape context descriptor to calculate similarity, determining the closest base model ID, for example, "base model A". Based on "base model A", using the extracted cross-section contour and key feature points, adjust the parameters of the base model (such as width, height, cavity position, wall thickness), generate an accurate parameterized contour model, this step can also be performed manually by human, i.e. manual modeling.

[0141] Further, the picture is subjected to region analysis, identifying the closed regions inside the contour as cavities, identifying the size and position of the cavities, and identifying specific geometric shapes as connection nodes or functional areas. These information are converted into technical parameters, such as `total width: 80.5 mm`, `total height: 60.2 mm`, `cavity number: 3`, `cavity 1 size: 20x30 mm`, `minimum wall thickness: 1.8 mm`, `connection notch type: T-shaped notch`. These parameters are associated with the parameterized contour model, and finally a structured technical parameter set containing all these information is generated, for example, a data structure containing key-value pairs.

[0142] The application further proposes that when the door and window profile components are split or combined during production, a unique identification code is assigned to the door and window profile components during production, and production execution data associated with the unique identification code is collected at key nodes of production, including: assigning each initial door and window profile component a main unique identification code containing order information, profile type and serial number; when a split operation of a door and window profile component is detected, assigning each sub-component after splitting a sub-unique identification code containing the main unique identification code and the sub-component serial number; when a combination operation of multiple sub-components is detected, assigning the combined component a combination unique identification code containing all sub-unique identification code information; establishing a hierarchical association relationship among the main unique identification code, the sub-unique identification code and the combination unique identification code; collecting production execution data associated with each level of unique identification code at key nodes of production; storing the production execution data in association with the corresponding unique identification code, and establishing a complete traceability chain of component splitting and combining during production.

[0143] The technical solution aims to solve the problem of how to maintain accurate collection and full traceability of production execution data when door and window profile components are split or combined during production.

[0144] This is achieved by establishing a dynamic and hierarchical unique identification code system. First, a main unique identification code is assigned to each initial door and window profile component entering the production process. This main identification code contains order, profile type and initial sequence information, serving as the original identity of the component throughout the production process. When an initial component is split into multiple sub-components, the system detects the splitting operation and assigns a sub-unique identification code to each newly generated sub-component. The design of this sub-identification code contains the main unique identification code information of its source and the serial number of the sub-component, clearly indicating that the sub-component is split from which main component and distinguishing different sub-components split from the same main component.

[0145] Conversely, when multiple sub-components are combined into a new component, the system detects the combination operation and assigns a combination unique identification code to the combined component. This combination identification code records all the sub-unique identification code information involved in the combination, indicating the composition source of the combined component.

[0146] The core of the scheme is to establish and maintain the hierarchical association relationship among the main unique identification code, the sub-unique identification code and the combination unique identification code. This association relationship clearly records the history and mutual relationship of the component from the initial form to all form changes such as splitting and combining.

[0147] At key nodes of production, the system collects production execution data associated with the unique identification code corresponding to the current component form. These data include process completion status, test results, operator information, etc.

[0148] Finally, the collected production execution data is stored in association with the corresponding unique identification code. Since a hierarchical association relationship is established between the identification codes, through any identification code, the original main component and all the splitting and merging processes experienced thereby can be traced back, thereby constructing a complete traceability chain of the splitting and merging of the components in the production process. Through such hierarchical identification and associated storage, even if the component morphology changes, the production execution data can be accurately associated with the component in a specific morphology, ensuring the transparency and traceability of the production process, and solving the technical problem of the interruption of the traceability chain caused by the splitting and merging of the components.

[0149] Specifically, the working principle of the technical scheme is as follows: when a door and window profile order enters the production process, first, according to the order information, the profile type and the sequence position in the batch, a master unique identification code is allocated for the initial long material of extrusion molding. For example, the long material with order number ORD123, profile type T01 and sequence number 001 in the batch, its master unique identification code can be generated as ORD123-T01-001. The master identification code is recorded in the data management center and is associated with the technical parameters of the long material. When the long material is cut into multiple short materials at the cutting station, the system detects the cutting operation (i.e. splitting). For each short material generated by cutting, the system allocates a sub-unique identification code for it. The sub-identification code contains the original master unique identification code and the sequence number of the short material in this split. For example, ORD123-T01-001 is cut into three short materials, which may obtain sub-unique identification codes ORD123-T01-001-S1 and ORD123-T01-001-S2, ORD123-T01-001-S3 respectively. These sub-identification codes are associated with the technical parameters (such as length) of the respective short materials, and record that they are derived from the master identification code ORD123-T01-001. In subsequent processing stations (such as drilling, slot milling), the collected production execution data (such as drilling completion time, operator) will be stored in association with the sub-unique identification code of the current short material. When multiple short materials are welded into a window frame at the welding station, the system detects the welding operation (i.e. merging). The system allocates a merged unique identification code for the window frame formed by welding. The merged identification code records all the sub-unique identification codes involved in the welding. For example, if ORD123-T01-001-S1, ORD123-T01-001-S2 and the short material ORD123-T01-002-S4 cut from another long material ORD123-T01-002 are spliced together, the merged identification code may be recorded as M-ORD123-T01-001-S1-ORD123-T01-001-S2-ORD123-T01-002-S4. The merged identification code is associated with the overall technical parameters of the window frame. In subsequent surface treatment, assembly and other stations, the collected production execution data will be stored in association with the merged identification code. By establishing and maintaining the hierarchical association relationship between the master, sub and merged identification codes in the data management center, the original long material, the splitting process experienced, the merging process involved and the production execution data collected at each station can be traced back at any time through any identification code. Thus, even if the component morphology changes, the traceability chain of the production process will not be interrupted, ensuring the transparency and management accuracy of the production process.

[0150] The application further proposes that assigning a main unique identification code containing order information, profile type and serial number to each initial door and window profile component includes: obtaining production environment information containing production line identification and time stamp; obtaining identification code assignment record of current active order from data management center; generating identification code prefix containing production line identification, time stamp code, order number and component type code based on production environment information and identification code assignment record; assigning batch number to each production batch and combining batch number with identification code prefix; generating serial number according to the sequence position of the component within the batch; combining identification code prefix, batch number and serial number to generate main unique identification code; storing main unique identification code in association with technical parameters of the corresponding door and window profile component; submitting assignment record of main unique identification code to data management center and verifying global uniqueness.

[0151] Among them, obtaining production environment information containing production line identification and time stamp can automatically read the production line number where the current component is located and the accurate time when the system generates the identification code.

[0152] Obtaining identification code assignment record of current active order from data management center involves querying the list of identification codes that have been assigned to all orders that are being produced or to be produced in the database.

[0153] Generating identification code prefix based on production environment information and identification code assignment record is to combine production line identification, time stamp, order number and component type (for example, converting profile type to code by looking up table) into a string according to preset rules after coding processing.

[0154] Assigning batch number to each production batch is to assign a unique batch identification code to a group of components produced simultaneously or continuously under an order.

[0155] Generating serial number according to the sequence position of the component within the batch is to assign an incremental number according to the production order or arrangement order of the component within the batch.

[0156] Combining identification code prefix, batch number and serial number to generate main unique identification code is to splice the above string parts according to the intended format to form the final identification code.

[0157] Storing main unique identification code in association with technical parameters of the corresponding door and window profile component is to establish a link between the identification code and the detailed technical specifications (such as size, material, structure) of the component in the data management center.

[0158] Submitting assignment record of main unique identification code to data management center and verifying global uniqueness is to register the newly generated identification code to the central system and perform a duplicate checking operation to ensure that the identification code has not been used in the entire system.

[0159] Specifically, in order to ensure that the main unique identification code assigned to each initial door and window profile component is globally unique, contains traceability information and can be associated with the component, the scheme is implemented by constructing an identification code containing multiple layers of information.

[0160] Firstly, the production line information and time information when the component is produced are obtained, and the identification code record assigned for the order is obtained from the data management center. Based on these information, an identification code prefix is generated, which embeds the production background information of the component, such as which production line, when, which order, and what type of component.

[0161] Further, a batch number is assigned for the batch of components being produced, and the batch number is combined with the prefix, thereby distinguishing the products of different batches under the same order.

[0162] Then, according to the specific production order or position of the component in the batch, a serial number is generated for uniquely identifying each individual component in the batch.

[0163] Thus, the prefix, batch number and serial number are combined together according to the preset coding rule, and the main unique identification code of the component is generated. The identification code structurally contains the information of production line, time, order, component type, batch and sequence, etc., so that it has traceability.

[0164] Subsequently, the generated main unique identification code is associated with the detailed technical parameters (such as geometric size, material grade, surface treatment requirement, etc.) corresponding to the component in the data management center, and the connection between the identification code and the physical component and its specifications is established.

[0165] Finally, the assignment record of the main unique identification code is submitted to the data management center, and the global uniqueness verification is performed by the system to ensure that the identification code is unique in the entire system, thereby laying a reliable foundation for subsequent data collection, traceability and management of component splitting and merging in the production process.

[0166] The application further proposes to detect perspective deformation features in a reference picture, including detecting straight edge bending degree and parallel line convergence degree of a reference object; calculating a perspective transformation matrix according to the detected perspective deformation features; correcting the reference picture using the perspective transformation matrix to generate an orthographic image; applying an edge detection algorithm on the orthographic image to extract a cross-sectional contour line of the door and window profile; performing noise filtering and breakpoint connection processing on the processed cross-sectional contour line; identifying corner points, intersection points and curve inflection points on the processed cross-sectional contour line and marking them as key feature points; establishing a proportional relationship between pixel coordinates in the orthographic image and actual physical dimensions, which is used for subsequent size parameter calculation.

[0167] The perspective distortion in the reference image is identified by detecting the bending degree of straight edges and the convergence degree of parallel lines. Straight edges appear non-straight under perspective projection, and parallel lines show a converging trend in the image. Detecting these geometric features can determine the type and magnitude of the perspective distortion.

[0168] According to the detected distortion features, a perspective transformation matrix is calculated, which describes the transformation relationship of mapping the image from one perspective plane to another. The reference image is processed using the calculated perspective transformation matrix to generate an orthographic image. The orthographic image eliminates the shape and size distortion caused by perspective projection, and the projection of objects in the image is closer to the actual shape. An edge detection algorithm such as the Canny algorithm is applied on the orthographic image to extract a set of pixel points with sharp changes in brightness or color, forming the cross-sectional contour line of the door and window profile.

[0169] On the processed continuous cross-sectional contour line, points with specific geometric properties are identified, such as points with high curvature change rate (corner points), points where multiple contour lines intersect (intersection points), and points where the direction of the curve changes (curve inflection points). These points are marked as key feature points. In the orthographic image, a reference object with known physical dimensions (such as a ruler or a known size profile part) is used to establish a correspondence between image pixel units and actual physical length units, i.e. a scale factor. This scale relationship is used to convert the measured pixel distance in the image to the actual physical distance.

[0170] Specifically, the method aims to solve the problem of inaccurate extraction of door and window profile cross-sectional contours and key feature points when the reference image has perspective distortion.

[0171] First, a reference image containing a door and window profile is received. The reference image is analyzed to detect the perspective distortion features present in it. Specifically, it is detected whether the straight edges of the reference object (i.e. the door and window profile) are bent and whether the parallel lines are converging. These features are the manifestations of perspective projection on a two-dimensional image.

[0172] According to the detected bending degree of straight edges and the convergence degree of parallel lines, a perspective transformation matrix is calculated. This matrix contains the mathematical parameters required to convert the image from the current perspective view to the orthographic view.

[0173] The reference image is corrected using the calculated perspective transformation matrix to generate an orthographic image. The orthographic image eliminates the perspective distortion, making the cross-sectional shape of the profile accurately reflected in the image. On the corrected orthographic image, an edge detection algorithm is applied to extract the cross-sectional contour line of the door and window profile.

[0174] Since the image has been rectified, the extracted contour line more accurately represents the actual cross-sectional shape of the profile. Noise filtering and breakpoint connection processing are performed on the extracted cross-sectional contour line. The noise is removed, and the broken contour line is connected to form a complete and smooth cross-sectional contour line.

[0175] On the processed cross-sectional contour line, corner points, intersection points, and curve inflection points are identified and labeled. These points are key points that define the geometry of the profile, and their position information is crucial for subsequent generation of a parameterized model.

[0176] Finally, a scaling relationship between pixel coordinates and actual physical dimensions is established in the front view image. Through this scaling relationship, the pixel dimensions extracted from the front view image can be converted into actual physical dimensions. In this way, even if the original reference picture has perspective distortion, the cross-sectional contour and key feature points of the door and window profile can be accurately extracted, and accurate size parameters can be calculated for generating a structured technical parameter set.

[0177] In some specific embodiments, it is assumed that a reference picture containing a side view of a cross-section of a door and window profile is received, and the picture contains a ruler with a known length of 100 mm (provided by the user). The picture is preprocessed, such as adjusting brightness and contrast. A straight line detection algorithm (such as based on Hough transform) is applied to detect straight line segments in the picture. The detected straight line segments are analyzed to identify straight lines belonging to the profile edge and the ruler edge. The degree of curvature of the straight line edge belonging to the profile and the degree of convergence of parallel lines belonging to the profile or background are calculated. According to the detected degrees of curvature and convergence, a perspective transformation matrix is calculated using the least squares method or the RANSAC algorithm.

[0178] The calculated perspective transformation matrix is applied to the original picture to perform perspective correction and generate a front view image. A Canny edge detection algorithm is applied on the front view image to extract the cross-sectional contour line of the profile. Morphological closing operation processing is performed on the extracted contour line to connect broken parts and smooth the contour.

[0179] On the processed contour line, a corner point detection algorithm (such as Harris corner point detection) and curve analysis method are applied to identify and label corner points, intersection points, and curve inflection points. In the front view image, the length of the ruler in the pixel coordinate system is measured, for example, the measurement result is 500 pixels. The actual length of the ruler is known to be 100 mm, and thus a scaling relationship between pixels and physical dimensions is established: 1 pixel corresponds to 0.2 mm. Using this scaling relationship, the position and size of the extracted cross-sectional contour and key feature points in the pixel coordinate system are converted into actual physical dimensions, for example, the wall thickness of the profile is measured to be 100 pixels in the front view image, and the actual wall thickness is 100 * 0.2 = 20 mm. Thus, accurate profile geometric parameters are obtained.

[0180] The application further proposes that when detecting that the door and window profile component splitting operation is carried out step by step at different stations or at different time periods, assigning a sub-unique identification code containing a main unique identification code and a sub-component serial number to each sub-component after splitting includes: obtaining process rules and splitting plan information of the splitting operation from a data management center; based on the process rules and the splitting plan information, generating a splitting task table containing the total number of splitting steps and the station information of each step; assigning a unique splitting task identification to the splitting task table; storing the splitting task identification in association with the main unique identification code; after each splitting station completes the corresponding splitting step, recording the number and type of sub-components generated at the current station; according to the number and type of sub-components, combining the position of the current station in the splitting task table, generating a sub-component serial number prefix containing the station code and the splitting step serial number; combining the main unique identification code, the sub-component serial number prefix and the serial number of the sub-component at the current station to generate a sub-unique identification code; storing the sub-unique identification code in association with the technical parameters of the corresponding sub-component and the current splitting state information; when the sub-component is transferred to the next splitting station, obtaining the splitting task identification and the current splitting state by scanning the sub-unique identification code; verifying whether the current station is the next expected station in the splitting task table, and continuing to execute the subsequent splitting step after verification.

[0181] Wherein, obtaining the process rules and the splitting plan information of the splitting operation from the data management center provides the basis for performing step-by-step splitting. Based on these information, the splitting task table is generated, and a structured process of step-by-step splitting is constructed, which contains the total number of steps in the entire splitting process and which station needs to be completed for each step.

[0182] Assigning a unique splitting task identification to the splitting task table and storing this identification in association with the main unique identification code of the original component ensures that all subsequently generated sub-components can be traced back to the same splitting task and the original component.

[0183] After each splitting station completes the corresponding splitting step, the system records the number and type of sub-components generated at the station, providing output information of the current step. According to these output information and the position of the current station in the splitting task table, a sub-component serial number prefix is generated, which contains the information that the sub-component is generated at which step, which station of which splitting task.

[0184] Combining the main unique identification code of the original component, the sub-component serial number prefix and the serial number of the sub-component at the current station to generate the sub-unique identification code of the sub-component realizes the fine identification of the step-by-step generated sub-component.

[0185] The sub-unique identification code is stored in association with the technical parameters of the sub-component and the current splitting state information, so that the system can master the specific attributes and state of each sub-component in the step-by-step splitting process. When the sub-component needs to be transferred to the next splitting station, by scanning its sub-unique identification code, the system can obtain its associated splitting task identification and current state, and verify whether the current station is the next expected station that the sub-component should enter according to the splitting task table.

[0186] Specifically, when the splitting operation of the door and window profile component needs to be carried out step by step at different stations or in different time periods, first, the preset splitting process rules and the splitting plan information for a specific order are obtained from the data management center. Based on these rules and plans, the system generates a structured splitting task table, which defines in detail the total number of splitting steps that the component needs to go through and at which specific production station each step should be performed. To ensure the traceability of the entire splitting process, a unique splitting task identification is assigned to the generated splitting task table, and the identification is stored in association with the main unique identification code of the original component to be split. Thus, the connection between the original component and the entire step-by-step splitting task is established.

[0187] In the production process, when the component arrives at the first splitting station and completes the corresponding splitting step, the system records the number and type of sub-components actually produced at the station. According to these recorded information and the position of the current station in the preset splitting task table (represented by the station code and the splitting step sequence number), the system generates a sub-component sequence number prefix.

[0188] Further, the main unique identification code of the original component, the generated sub-component sequence number prefix, and the sequence number of the sub-component produced at the current station are combined to generate the sub-unique identification code of the sub-component. This sub-unique identification code contains information about the original component, the splitting task, the splitting step and the station, as well as the sequence in that step, achieving fine and hierarchical identification of the step-by-step produced sub-components.

[0189] The generated sub-unique identification code is stored together with the current technical parameters of the sub-component (which may change due to splitting) and the current splitting state information, so that the system can master the attributes and state of each sub-component in real time.

[0190] When the sub-component completes the operation at the current station and is ready to be transferred to the next splitting station, by scanning the sub-unique identification code on the sub-component, the system can quickly retrieve its associated splitting task identification and current state information. The system will verify whether the station that the sub-component is about to enter is the next expected station defined in the splitting task table. Only when the verification is passed, the sub-component is allowed to enter and continue to perform the subsequent splitting step.

[0191] The verification mechanism ensures that the components are correctly transferred according to the predetermined process flow, avoiding out-of-sequence or missing sequences, thereby effectively solving the problems of tracking, management and process control of sub-components in a multi-station, multi-step split scene, and maintaining the accuracy and traceability of the production process.

[0192] The application further proposes to detect perspective distortion features in a reference picture, including detecting the straight edge bending degree and parallel line convergence degree of a reference object, comprising: receiving a reference picture containing multiple door and window profiles of different angles; preprocessing the reference picture to improve image contrast and clarity; detecting all possible door and window profile regions in the reference picture and extracting edge features of each region; extracting user marking information or reference scale information from the reference picture; determining the target door and window profile region according to the user marking information or reference scale information; if no user marking information or reference scale information is detected, calculating the clarity, integrity and size ratio scores of each door and window profile region according to preset rules; selecting the best target door and window profile region based on the scores; detecting straight edges in the target door and window profile region and calculating the bending degree of the straight edges; detecting parallel lines in the target door and window profile region and calculating the convergence degree of the parallel lines; determining the degree of perspective distortion according to the bending degree and the convergence degree

[0193] Among them, receiving a reference picture containing multiple door and window profiles of different angles can increase the data dimension for analysis and improve the robustness of subsequent processing.

[0194] Preprocessing the reference picture to improve image contrast and clarity can be achieved by applying image enhancement algorithms such as histogram equalization or adaptive contrast enhancement, which helps to more accurately detect edges and features.

[0195] Detecting all possible door and window profile regions in the reference picture and extracting edge features of each region can be achieved by target detection algorithms or methods based on edge detection and region segmentation, thereby locking the analysis range.

[0196] Extracting user marking information or reference scale information from the reference picture can recognize text markings through optical character recognition (OCR) technology, or recognize preset reference scale patterns through image matching, feature point detection, etc. These information provides direct basis for determining the target region and subsequent size calculation.

[0197] According to the user marking information or the reference ruler information, the target door and window profile region is determined, which is to preferentially use the explicit information provided by the user to locate the region to be analyzed. If no user marking information or reference ruler information is detected, the clarity, completeness and size ratio scores of each door and window profile region are calculated according to preset rules, which is a strategy for automatically selecting the best analysis region. The preset rules can be based on image quality evaluation indexes. For example, the clarity can be measured by using the average value or variance of image gradient, the completeness can be evaluated based on the continuity or closeness of the edge contour, and the size ratio can check whether the aspect ratio of the region meets the typical proportion range of the profile.

[0198] The best target door and window profile region is selected based on the score, which is to select the region with the highest score as the object of subsequent analysis. Straight line edges are detected in the target door and window profile region, and the curvature of the straight line edges is calculated. The edge pixels can be extracted by applying an edge detection algorithm such as the Canny algorithm, and then a straight line is fitted using the Hough transform or the least squares method. The maximum distance or average distance of the edge pixels to the fitted straight line is calculated as a quantitative indicator of the curvature.

[0199] Parallel lines are detected in the target door and window profile region, and the convergence of the parallel lines is calculated. Multiple straight line edges can be detected, and straight line pairs that should theoretically be parallel lines are identified (for example, by comparing the angles of the straight lines). Then, the intersection positions or angles of these straight line pairs in the image are calculated. The closer the intersection points or the larger the angles, the higher the convergence.

[0200] The degree of perspective distortion is determined according to the curvature and the convergence, which is to combine the quantitative curvature and convergence values to calculate one or a set of parameters to describe the degree and direction of the perspective transformation of the reference picture.

[0201] Specifically, the technical scheme receives multiple angle pictures and performs preprocessing, enhancing the analyzability of the images. Then, potential door and window profile regions are identified in the preprocessed pictures, narrowing down the processing range. The scheme preferentially uses user-provided markings or ruler information to accurately lock the target region, which improves the accuracy of positioning. When there is a lack of user markings, the system can automatically evaluate the image quality and features of each region and select the most suitable region for analysis. In the determined target region, the scheme directly measures the performance of perspective distortion in the image by detecting straight line edges and quantifying their curvature, and detecting parallel lines and quantifying their convergence. Curvature and convergence are typical distortion characteristics of perspective projection on a two-dimensional image. By accurately calculating the quantitative values of these characteristics, the scheme can accurately determine the degree of perspective distortion contained in the reference picture.

[0202] Thus, the necessary data basis is provided for subsequent perspective correction of the picture, so that the corrected image is closer to the real orthographic view of the door and window profile, thereby enabling more accurate extraction of the cross-sectional profile and key feature points, ensuring the accuracy of the generated structured technical parameters, and solving the technical problem of inaccurate parameter extraction caused by picture perspective distortion.

[0203] In some embodiments, three reference pictures containing the same door and window profile but with slightly different shooting angles are received. The pictures are preprocessed, for example, the local contrast is enhanced by applying the CLAHE (Contrast Limited Adaptive Histogram Equalization) algorithm. All possible door and window profile regions in the pictures are detected, and the image regions within each detection box are extracted. It is checked whether there is a preset two-dimensional code or bar code as a user mark in each region, or whether there is a specific pattern reference scale. If the user mark or scale is detected, the corresponding region is determined as the target region. If not, the clarity (for example, the score is calculated using the Tenengrad gradient function), edge integrity (for example, the proportion of edge pixels in the total pixels in the Canny edge detection result is calculated), and the matching score of the aspect ratio with the typical profile ratio of each detection region are calculated. The region with the highest comprehensive score is selected as the target region. On the image of the selected target region, the LSD (Line Segment Detector) algorithm is applied to detect straight line segments. For each detected straight line segment, the maximum perpendicular distance between its endpoints and all edge pixel points is calculated as the bending degree of the straight line edge. The pairs of straight line segments with an angle difference less than 5 degrees are identified, the intersection position of their extension lines is calculated, and the convergence degree is calculated according to the distance from the intersection point to the center of the image or the cosine value of the included angle between the straight line segments. For example, the included angle θ is calculated, and the convergence degree can be represented as cos(θ). The average value of the bending degrees of all detected straight line edges and the average value of the convergence degrees of all identified parallel line pairs are taken as the quantitative indicators of the perspective distortion degree of the reference picture.

[0204] The application further proposes to create a temporary assembly task record containing temporary assembly information and test requirements; assign a temporary combination identification code containing the information of all sub-unique identification codes participating in the assembly to the combination after temporary assembly; establish a mapping relationship between the sub-unique identification code and the temporary combination identification code in the temporary assembly task record; after the test is completed, retrieve the original sub-unique identification code information from the temporary assembly task record; verify the consistency of the sub-components after re-splitting with the original sub-components; if the verification is passed, the original atomic unique identification code is kept unchanged; if the verification fails, a new sub-unique identification code containing the original atomic unique identification code and a change marker is assigned to the changed sub-component; update the sub-component mapping relationship table in the data management center to record the temporary assembly and re-splitting process experienced by the sub-component.

[0205] In the data record established in advance, the association between the unique identification code of each sub-component and the temporary combination identification code is recorded, and the ownership of the components in the temporary state is clarified. After the end of the test process, the original unique identification code information of each sub-component before the formation of the temporary combination is obtained by querying the data record, providing a basis for subsequent processing.

[0206] In the data record established in advance, the association between the unique identification code of each sub-component and the temporary combination identification code is recorded, and the ownership of the components in the temporary state is clarified. After the end of the test process, the original unique identification code information of each sub-component before the formation of the temporary combination is obtained by querying the data record, providing a basis for subsequent processing.

[0207] After the temporary combination is disassembled again, the components obtained are checked to confirm whether they are consistent with the original sub-components in terms of physics or state, which is the key to determining whether the components have been changed. If the check result confirms the consistency, the original sub-component unique identification code is continued to be used, maintaining the continuity of the component tracking chain. If the check result shows inconsistency, a new unique identification code is generated for the changed sub-component, which contains the unique identification code of the original sub-component and a mark indicating the change, ensuring the traceability of the changed components.

[0208] In the data management system, the data table recording the association of components is updated, and the temporary assembly and disassembly process experienced by the sub-components is recorded in detail, constructing a complete traceability chain of components in complex production links.

[0209] Specifically, the technical scheme solves the problem of how to maintain the integrity and accuracy of the component tracking chain when the disassembled sub-components need to be temporarily assembled for testing and then disassembled in the production of personalized door and window profiles. First, a data record is established, which contains detailed information about the purpose of temporary assembly and test requirements, providing a basis for subsequent tracking. Then, an identification code is generated for the component set formed by temporary assembly, which contains the unique identification code information of each sub-component that constitutes the set, so that in the temporary assembly state, multiple sub-components can be managed and tracked as a whole, while retaining the original sub-component information that constitutes the whole.

[0210] Subsequently, in the data record established in advance, the association between the unique identification code of each sub-component and the temporary combination identification code is recorded, and it is recorded which sub-components constitute which temporary combination, establishing the association of components in the temporary state.

[0211] After the test is completed, the original unique identification code information of each sub-component before the temporary assembly is obtained by querying the data record, so that the original sub-component identity can be traced after the temporary assembly is disassembled again.

[0212] Then, the components obtained after the temporary assembly is disassembled again are checked to confirm whether they are consistent with the original sub-components in physical or state, for judging whether the sub-components have changed physically or replaced during the temporary assembly and test. If the checking result confirms the consistency, the original sub-component unique identification code is continued to be used, indicating that the sub-component has not substantially changed after the temporary assembly and test, and the original identification code can be continued to be used for tracking, maintaining the continuity of the tracking chain. If the checking result shows inconsistency, a new unique identification code is generated for the changed sub-component, which contains the unique identification code of the original sub-component and a change indicating mark, indicating that the sub-component has changed and needs to be assigned a new identification code to reflect its current state, while the original unique identification code is contained to retain the historical information before the change, ensuring the tracking accuracy after the change.

[0213] Finally, in the data management system, the data table recording the association relationship of the components is updated, and the temporary assembly and disassembly process experienced by the sub-components is recorded in detail, so that the entire temporary assembly, test, disassembly and possible component change process are recorded in the data management center, and a complete tracking chain of the components in the complex production link is constructed, ensuring the accurate collection and management of production execution data.

[0214] The application further proposes image processing on the reference picture, extracting the cross-sectional profile and key feature points of the door and window profile, including: detecting the reflection area and shadow area in the reference picture, and generating a reflection and shadow distribution map; according to the reflection and shadow distribution map, the reference picture is divided into a normal area, a reflection area and a shadow area; an exposure compensation algorithm is applied to the reflection area to reduce the brightness value of the overexposed area; a brightness enhancement algorithm is applied to the shadow area to improve the visibility of the dark area; the processed reflection area and shadow area are image fused with the normal area to generate an image with balanced lighting; an edge detection algorithm is applied on the image with balanced lighting to extract the initial contour line; the initial contour line is subjected to breakpoint connection and smoothing processing to generate a complete door and window profile cross-sectional profile; and corner points, intersection points and curve inflection points on the complete door and window profile cross-sectional profile are identified and marked as key feature points.

[0215] In the application, the reflection area and shadow area in the reference picture are detected to generate a reflection and shadow distribution map, which can be realized by analyzing the brightness distribution or gradient information of the image pixels. The reflection area is usually represented as an area with a brightness value close to the maximum value and a gentle gradient change, and the shadow area is represented as an area with a brightness value close to the minimum value and a gentle gradient change.

[0216] According to the reflection shadow distribution map, the picture can be segmented into different regions, and a threshold-based segmentation method or region growing algorithm can be used. The exposure compensation algorithm can be applied to the reflection region, and gamma correction or histogram equalization technology can be used. The brightness enhancement algorithm can be applied to the shadow region, and logarithmic transformation or local histogram stretching technology can be used.

[0217] Image fusion can be weighted average or Poisson fusion between the processed region and the normal region. The edge detection algorithm can be applied to the image of balanced light, and Canny operator or Sobel operator can be used. The breakpoint connection can be applied to the initial contour line, and morphological closing operation or distance-based connection algorithm can be used.

[0218] The contour line can be smoothed by spline interpolation or Gaussian filtering. The key feature points can be identified on the processed contour line by using Harris corner detection algorithm or curvature analysis-based method.

[0219] Specifically, the technical scheme solves the problem that the presence of reflection or shadow in the reference picture leads to inaccurate extraction of the cross-section contour and key feature points of the door and window profile. First, the reflection shadow distribution map is detected and generated to determine the area of abnormal light in the image. Thus, the picture is segmented into normal, reflection and shadow regions, providing a basis for subsequent targeted processing.

[0220] Further, the exposure compensation is performed on the reflection region to reduce the excessively high brightness and restore the details; the brightness enhancement is performed on the shadow region to improve the excessively low brightness and enhance the visibility. The fusion of these processed regions and the normal region generates an image with uniform overall light.

[0221] Edge detection on the image with uniform light can more accurately extract the initial contour line of the door and window profile, reducing the interference caused by uneven light. The extracted initial contour line is connected and smoothed to obtain a continuous and smooth cross-section contour.

[0222] Finally, the corner points, intersection points and curve inflection points are identified and marked on the complete contour, and these points are used as key feature points, and the position accuracy is improved to provide accurate data for subsequent technical parameter generation.

[0223] In some embodiments, a reference picture containing the section of the door and window profile is received. The picture is analyzed to identify areas with brightness higher than threshold A as potential reflection areas, and areas with brightness lower than threshold B as potential shadow areas, and a reflection and shadow distribution map is generated. According to the distribution map, the picture pixels are divided into three sets: normal pixels, reflection pixels, and shadow pixels. The brightness value of the reflection pixels is applied with a function f(L) = L^gamma, where gamma < 1, to reduce the brightness. The brightness value of the shadow pixels is applied with a function g(L) = 255 * (L / 255)^alpha, where alpha > 1, to increase the brightness. The processed reflection and shadow pixels are merged with the normal pixels to generate a new image. A Canny edge detection algorithm is applied on the new image to extract an initial edge pixel set. A morphological closing operation is performed on the edge pixel set to connect broken edges. Curve fitting is performed on the connected edges to generate a smooth contour curve. On the contour curve, the curvature of each point is calculated, and points with curvature greater than threshold C are marked as corner points or curve inflection points. The intersection points of the contour line segments are detected and marked as intersection points. These marked points are the key feature points.

[0224] In a second aspect, referring to Figure 2 The present application further proposes a door and window profile individual customization device for realizing information collaboration in the door and window profile individual customization process. The system comprises:

[0225] A receiving module 210 is configured to receive unstructured data submitted by a user and containing individual customization requirements of a door and window profile.

[0226] An analysis module 220 is configured to analyze the unstructured data and generate structured technical parameters.

[0227] A model generation module 230 is configured to call a parameterized model library to generate a three-dimensional model of the door and window profile and a production data package based on the structured technical parameters.

[0228] A scheduling module 240 is configured to generate a scheduling plan for small-batch and multi-variety production according to the production data package and production resource state information.

[0229] An identification tracking module 250 is configured to assign a unique identification code to a door and window profile component in a production process and collect production execution data associated with the unique identification code at key nodes in the production process. A data sharing module is configured to update the production execution data to a data management center in real time and provide a shared view of an order status to a user, a supplier, and a downstream service provider.

[0230] By receiving and analyzing unstructured user requirements, structured technical parameters are generated, and then models and production data are generated, small batch multi-variety scheduling is carried out, parts are identified and tracked, and production data is collected, and finally real-time updating and sharing of production information are realized, which has the advantages of automatically processing user demand data, improving data conversion accuracy, automatically generating design and production data, optimizing small batch multi-variety production scheduling, realizing part traceability in the production process, improving production information transparency, and promoting supply chain information collaboration.

[0231] In addition, in some preferred embodiments, the door and window profile individual customization device provided by the application can perform any one of the steps of the above method.

[0232] The above only describes the embodiments of the application and is not used to limit the protection scope of the application. For those skilled in the art, the application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the application shall be included in the protection scope of the application.

Claims

1. A method of individual customisation of a door and window profile, characterised by, The method comprises the following steps: receiving user-submitted unstructured data containing door and window profile personalization requirements; parsing the unstructured data to generate structured technical parameters; based on the structured technical parameters, calling a parameterized model library to generate a three-dimensional model of the door and window profile and a production data package; generating a scheduling plan for small-batch and multi-variety production according to the production data package and production resource state information; allocating a unique identification code to each door and window profile component during production, and collecting production execution data associated with the unique identification code at key production nodes; updating the production execution data in real time to a data management center, and providing a shared view of order status to users, suppliers, and downstream service providers; parsing the unstructured data to generate structured technical parameters comprises the following steps: receiving user-submitted unstructured data containing door and window profile reference pictures; performing image processing on the reference pictures to extract the cross-sectional profile and key feature points of the door and window profile; matching the cross-sectional profile with the basic models in the preset profile parameterized model library to determine the closest basic model; based on the basic model, combining the cross-sectional profile and key feature points to generate a parameterized profile model of the door and window profile; extracting detailed features from the reference pictures, including cavity structure, wall thickness distribution, connection nodes, and functional areas; converting the detailed features into corresponding technical parameters and associating them with the parameterized profile model; generating a set of structured technical parameters containing geometric dimensions, structural features, and functional requirements; allocating a unique identification code to each door and window profile component during production, and collecting production execution data associated with the unique identification code at key production nodes comprises the following steps: allocating a main unique identification code containing order information, profile type, and serial number to each initial door and window profile component; when a splitting operation is detected for a door and window profile component, allocating a sub-unique identification code containing the main unique identification code and the serial number of the sub-component to each sub-component after splitting; when a merging operation is detected for multiple sub-components, allocating a merged unique identification code containing the information of all sub-unique identification codes to the merged component; establishing a hierarchical association relationship among the main unique identification code, sub-unique identification code, and merged unique identification code; collecting production execution data associated with each level of unique identification code at key production nodes; storing the production execution data in association with the corresponding unique identification code, and establishing a complete traceability chain for component splitting and merging during production.

2. A method of personalized customization of a door and window profile according to claim 1, characterized in that, allocating a main unique identification code containing order information, profile type, and serial number to each initial door and window profile component comprises the following steps: obtaining production environment information containing production line identification and timestamp; obtaining identification code allocation records of the current active order from the data management center; based on the production environment information and the identification code allocation records, generating an identification code prefix containing production line identification, timestamp encoding, order number, and component type code; allocating a batch number to each production batch, and combining the batch number with the identification code prefix; generating a serial number according to the sequence position of the component within the batch; combining the identification code prefix, the batch number, and the serial number to generate a main unique identification code; The main unique identifier is associated with and stored in relation to the technical parameters of the corresponding door and window profile components; Submit the allocation record of the primary unique identifier to the data management center and verify its global uniqueness.

3. A method of personalized customization of a door and window profile according to claim 1, characterized in that, The reference Image processing was performed on the image to extract the cross-sectional contours and key feature points of the door and window profiles, including: Detecting perspective distortion features in the reference image includes detecting the curvature of straight edges and the convergence of parallel lines of the reference object; Calculate the perspective transformation matrix based on the detected perspective distortion features; The reference image is corrected using the perspective transformation matrix to generate a front view image; An edge detection algorithm is applied to the front view image to extract the cross-sectional contour lines of the door and window profiles; The cross-sectional contour line is subjected to noise filtering and breakpoint connection processing; Identify corner points, intersection points, and curve inflection points on the processed cross-sectional contour line and mark them as key feature points; Establish the proportional relationship between the pixel coordinates in the front view image and the actual physical size for subsequent size parameter calculations.

4. A method of personalized customization of a door and window profile according to claim 1, characterized in that, When door and window type is detected When the material component disassembly operation is carried out step by step at different workstations or at different time periods, the assignment of a sub-unique identifier containing a primary unique identifier and a sub-component sequence number to each disassembled sub-component includes: Obtain the process rules and splitting plan information for the splitting operation from the data management center; Based on the process rules and splitting plan information, a splitting task table containing the total number of splitting steps and the workstation information for each step is generated. Assign a unique split task identifier to the split task table; The split task identifier is associated with and stored with the primary unique identifier; After completing the corresponding disassembly steps at each disassembly station, record the number and type of sub-components generated at the current station; Based on the number and type of the sub-components, and combined with the current workstation's position in the splitting task table, a sub-component sequence number prefix containing the workstation code and the splitting step sequence number is generated; The primary unique identifier is combined with the sub-component serial number prefix and the sub-component's serial number at the current workstation to generate a sub-unique identifier. The sub-unique identifier is associated with and stored in conjunction with the technical parameters and current splitting status information of the corresponding sub-component; When a sub-component is transferred to the next disassembly station, the disassembly task identifier and the current disassembly status are obtained by scanning the sub-unique identifier. Verify whether the current workstation is the next expected workstation in the split task table, and continue to execute subsequent splitting steps after the verification is successful.

5. A method of personalized customization of a door and window profile according to claim 3, characterized in that, The detection of perspective distortion features in the reference image, including detecting the curvature of straight edges and the convergence of parallel lines of the reference object, includes: Receive reference images containing multiple door and window profiles at different angles; The reference image is preprocessed to improve image contrast and clarity; Detect all possible door and window profile areas in the reference image and extract the edge features of each area; Extract user-marked information or reference ruler information from the reference image; The target door and window profile area is determined based on the user marking information or reference ruler information; If no user-marked information or reference ruler information is detected, the clarity, integrity, and size ratio scores for each door and window profile area are calculated according to preset rules. Select the optimal target door and window profile area based on the score; Detecting straight edges in the target door and window profile region, and calculating the curvature of the straight edges; Detecting parallel lines in the target door and window profile region, and calculating the convergence of the parallel lines; Determining the degree of perspective distortion according to the curvature and the convergence.

6. A method of personalized customization of a door and window profile according to claim 1, characterized in that, When it is detected that the split sub-components need to be temporarily assembled for testing before being split again, the assigning of a sub-unique identification code for each split sub-component, which contains a main unique identification code and a sub-component serial number, includes: Creating a temporary assembly task record containing temporary assembly information and testing requirements; Assigning a temporary combination identification code to the temporarily assembled combination, which contains information of all sub-unique identification codes involved in the assembly; Establishing a mapping relationship between the sub-unique identification code and the temporary combination identification code in the temporary assembly task record; After the testing is completed, retrieving the temporary assembly task record to obtain the original sub-unique identification code information; Verifying the consistency of the sub-components after being split again with the original sub-components; If the verification is passed, the original atomic unique identification code remains unchanged; If the verification is not passed, assigning a new sub-unique identification code to the changed sub-component, which contains the original atomic unique identification code and a change marker; Updating the sub-component mapping relationship table in the data management center to record the temporary assembly and re-splitting process experienced by the sub-component.

7. A method of personalized customization of a door and window profile according to claim 1, characterized in that, When there is reflection or shadow on the surface of the door and window profile in the reference picture, the image processing of the reference picture to extract the cross-sectional profile and key feature points of the door and window profile includes: Detecting the reflection and shadow areas in the reference picture to generate a reflection and shadow distribution map; According to the reflection and shadow distribution map, the reference picture is divided into normal areas, reflection areas, and shadow areas; Applying an exposure compensation algorithm to the reflection areas to reduce the brightness value of the overexposed areas; Applying a brightness enhancement algorithm to the shadow areas to improve the visibility of the dark areas; Image fusion of the processed reflection and shadow areas with the normal areas to generate an image with balanced lighting; Applying an edge detection algorithm on the image with balanced lighting to extract the initial profile line; Performing breakpoint connection and smoothing processing on the initial profile line to generate a complete cross-sectional profile of the door and window profile; Identifying and marking the corner points, intersection points, and curve inflection points on the complete cross-sectional profile of the door and window profile as key feature points.

8. A device for door and window profile individual customization for information coordination in the process of door and window profile individual customization, characterized in that, The device comprises: a receiving module for receiving user-submitted unstructured data containing door and window profile individualization requirements; a parsing module for parsing the unstructured data to generate structured technical parameters; further for receiving user-submitted unstructured data containing reference pictures of door and window profiles; performing image processing on the reference pictures to extract the cross-sectional profile and key feature points of the door and window profile; matching the cross-sectional profile with the base models in the preset profile parameterization model library to determine the closest base model; based on the base model, combining the cross-sectional profile and key feature points to generate a parameterized profile model of the door and window profile; extracting detailed features from the reference pictures, including cavity structure, wall thickness distribution, connection nodes, and functional areas; converting the detailed features into corresponding technical parameters and associating them with the parameterized profile model; Generating a set of structured technical parameters including geometric dimensions, structural features and functional requirements; A model generation module for generating a three-dimensional model of the door and window profile and a production data package based on the structured technical parameters by calling a parameterized model library; A scheduling module for generating a scheduling plan for small-batch and multi-variety production according to the production data package and production resource state information; An identification tracking module for assigning a unique identification code to the door and window profile component during the production process and collecting production execution data associated with the unique identification code at key production nodes; Also for assigning each initial door and window profile component a main unique identification code containing order information, profile type and serial number; When detecting a splitting operation of the door and window profile component, assigning each sub-component after splitting a sub-unique identification code containing the main unique identification code and the sub-component serial number; When detecting a merging operation of multiple sub-components, assigning the merged component a merged unique identification code containing all sub-unique identification code information; Establishing a hierarchical association relationship among the main unique identification code, the sub-unique identification code and the merged unique identification code; Collecting production execution data associated with each level of unique identification code at key production nodes; Storing the production execution data in association with the corresponding unique identification code and establishing a complete traceability chain of component splitting and merging in the production process; A data sharing module for updating the production execution data to the data management center in real time and providing a shared view of the order status to users, suppliers and downstream service providers.

Images