A powder spreading method, a printing system, an electronic device, and a storage medium

Abstract

The present application belongs to the technical field of additive manufacturing, and particularly relates to a powder spreading method, a printing system, an electronic device and a storage medium, comprising: an acquisition step: after a powder spreading operation is completed, a powder spreading image of a current layer is acquired; the powder spreading image of the current layer is preprocessed to extract a powder spreading abnormal area; area characteristic information of the powder spreading abnormal area is calculated; the powder spreading image of the current layer is input into a pre-trained powder spreading defect recognition model to obtain a recognition result; the recognition result and the area characteristic information of the powder spreading abnormal area are compared; when the area characteristic information of the powder spreading abnormal area is within a predetermined area range and the recognition result indicates that powder spreading is normal, a next layer printing operation is performed; when the area characteristic information of the powder spreading abnormal area is not within the predetermined area range or the recognition result indicates that powder spreading is abnormal, a next powder spreading operation is performed, and the acquisition step is returned to.

Description

Technical Field

[0001] This invention belongs to the field of additive manufacturing technology, specifically relating to a powder spreading method, a printing system, electronic equipment, and a storage medium. Background Technology

[0002] 3D printing, as an advanced manufacturing technology, is based on a three-dimensional data model. It creates three-dimensional objects by layering materials. Powder bed monitoring is a crucial step in additive manufacturing, especially in powder bed fusion (PBF) technology. PBF technology encompasses selective laser melting (SLM) and selective laser sintering (SLS), constructing parts by melting / sintering materials layer by layer on a powder bed. Powder bed monitoring ensures the uniformity and flatness of each powder layer, guaranteeing the final product quality.

[0003] Currently, the powder spreading process often faces several technical challenges. First, uneven powder layer thickness, whether too thick or too thin, can lead to inaccurate energy density distribution of the energy beam, resulting in unfused or over-melted areas during the melting process. This, in turn, creates porosity defects, affecting the density and mechanical properties of the formed part. Second, wear, blockage, or positioning deviations in the scraper or spreading mechanism of the powder spreading device can easily cause localized missing or streaky defects in the powder spreading area, leading to poor interlayer bonding or even forming failure. Furthermore, during printing, warping deformation caused by thermal stress or molten spatter, if not promptly removed, can interfere with the smooth spreading of subsequent powder layers, causing interlayer misalignment or increased surface roughness. Additionally, localized collapse of the powder bed can also disrupt powder spreading uniformity, affecting overall forming accuracy. If these defects are not detected and corrected in real time, the entire printing task will be interrupted, resulting in a serious waste of material and time resources. Summary of the Invention

[0004] This invention provides a powder spreading method, printing system, electronic device, and storage medium based on machine vision and AI. By employing collaborative detection and discrimination of machine vision and AI recognition results, a dual verification mechanism is constructed, effectively overcoming the limitations of single detection methods. It can utilize the high precision and rigor of traditional machine vision methods in quantitative measurement, as well as the adaptability of AI technology in identifying complex defects, thereby improving the reliability and accuracy of powder spreading quality detection results, reducing the false judgment rate, saving time, reducing material waste, and solving the cost problem of manual monitoring.

[0005] A method for applying powder based on machine vision and AI includes:

[0006] Acquisition steps: After the powder spreading operation is completed, acquire the powder spreading image of the current layer;

[0007] The powder-spreading image of the current layer is preprocessed to extract areas of abnormal powder spread;

[0008] Calculate the area characteristics of areas with abnormal powder spreading;

[0009] The powder-laying image of the current layer is input into a pre-trained powder-laying defect recognition model to obtain the recognition result;

[0010] Compare the identification results with the area feature information of the powder spreading abnormal area;

[0011] When the area feature information of the abnormal powder spreading area is within the predetermined area range, and the identification result indicates that the powder spreading is normal, the next layer of printing is executed;

[0012] If the area feature information of the abnormal powder application area is not within the predetermined area range, or if the identification result indicates an abnormal powder application, the next powder application operation is performed, and the process returns to the acquisition step.

[0013] Furthermore, the preprocessing of the powder-spreading image of the current layer to extract powder-spreading abnormal regions includes:

[0014] Multiple images of successfully applied powder are acquired, and a standard background image is generated by calculating the average of image pixels. The images of successfully applied powder are acquired by performing multiple powder application operations in advance.

[0015] The powder-spreading image of the current layer is filtered and denoised, and the filtered and denoised powder-spreading image of the current layer is compared with the standard background image to obtain a difference image;

[0016] The difference image is then subjected to opening operations to remove noise interference, and closing operations to connect neighboring regions to obtain the abnormal image.

[0017] A two-pass scanning algorithm was used to segment the abnormal image into connected components, resulting in multiple independent powder-spreading abnormal regions.

[0018] Furthermore, the calculation of the area feature information of the abnormal powder spreading region includes:

[0019] The number of pixels contained in each of the powder-spreading abnormal regions is counted, and the number of pixels is used as the area feature information of the corresponding powder-spreading abnormal region and stored in memory.

[0020] Furthermore, the two-pass scanning algorithm is used to segment the abnormal image into connected components, resulting in multiple independent powder-spreading abnormal regions, including:

[0021] The abnormal image is binarized to obtain a binary image;

[0022] The binary image is scanned twice. In the first scan, each pixel is traversed row by row and column by column. If the current pixel is a foreground pixel and there are no labeled foreground neighbor pixels in the connected component of the current pixel, a first preset label is assigned to the current pixel. If the current pixel is a foreground pixel and there is only one labeled foreground neighbor pixel in the connected component of the current pixel, a second preset label is assigned to the current pixel. If the current pixel is a foreground pixel and there are two or more foreground neighbor pixels in the connected component of the current pixel, the labels of all pixels in the connected component are recorded as equal.

[0023] During the second scan, pixels marked by labels with equal relationships are grouped into the same connected component, and a third preset label is assigned to pixels marked by labels with equal relationships.

[0024] Based on the first preset label, the second preset label, and the third preset label, the binary image is segmented into multiple connected components to obtain multiple independent powder spreading abnormal regions, ensuring the accuracy of the powder spreading abnormal regions and thus providing an accurate data foundation for subsequent statistics.

[0025] Furthermore, acquiring multiple successfully applied powder images includes:

[0026] Multiple powder spreading experiments were conducted using the same type of powder;

[0027] Collect powder-spreading images that meet the preset requirements during the powder-spreading experiment to obtain multiple images of successful powder-spreading.

[0028] Furthermore, the step of inputting the powder-laying image of the current layer into a pre-trained powder-laying defect recognition model to obtain the recognition result includes:

[0029] Multiple images of successful powder application and images of abnormal powder application are acquired, and the abnormal areas in each of the abnormal powder application images are marked using tagging software. The normal powder application images and the abnormal powder application images are acquired by performing multiple powder application operations in advance.

[0030] The AI ​​model was trained using the normal powder spreading images and the labeled abnormal powder spreading images to obtain a powder spreading defect recognition model;

[0031] The powder spreading defect recognition model is used to perform AI recognition on the powder spreading image of the current layer to obtain the recognition result.

[0032] Furthermore, if the current layer fails to spread toner twice consecutively, the printing process will stop.

[0033] A printing system implementing a toner spreading method based on machine vision and AI, comprising:

[0034] The acquisition module is used to acquire the powder-spreading image of the current layer after the powder-spreading operation is completed;

[0035] The calculation module is used to preprocess the powder-laying image of the current layer, extract the powder-laying abnormal region and calculate the area feature information of the powder-laying abnormal region; and input the powder-laying image of the current layer into the pre-trained powder-laying defect recognition model to obtain the recognition result;

[0036] The detection and discrimination module is used to compare the identification result with the area feature information of the powder spreading abnormal area; when the area feature information of the powder spreading abnormal area is within a predetermined area range and the identification result indicates that the powder spreading is normal, the next layer printing operation is performed; when the area feature information of the powder spreading abnormal area is not within the predetermined area range, or the identification result indicates that the powder spreading is abnormal, the next powder spreading operation is performed, and the process returns to the acquisition module.

[0037] An electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the powder spreading method as described above.

[0038] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the powder spreading method as described above.

[0039] The beneficial effects of this invention are as follows:

[0040] This invention constructs a dual verification mechanism by employing collaborative detection and discrimination of machine vision and AI recognition results. This effectively overcomes the limitations of a single detection method. It can utilize the high precision and rigor of traditional machine vision methods in quantitative measurement, as well as the adaptability of AI technology in identifying complex defects. This improves the reliability and accuracy of powder spreading quality detection results, reduces the false judgment rate, saves time, reduces material waste, and solves the cost problem of manual monitoring. Attached Figure Description

[0041] Figure 1 This is a flowchart of the present invention;

[0042] Figure 2 This is a flowchart for extracting abnormal powder spreading areas in this invention;

[0043] Figure 3 for Figure 2 The detailed extraction flowchart for step S24 is shown below.

[0044] Figure 4 This is a flowchart of the powder spreading method based on machine vision and AI in this invention;

[0045] Figure 5 This is a schematic diagram of the printing system of the present invention;

[0046] Figure 6 This is a schematic diagram of the structure of a computer device. Detailed Implementation

[0047] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0048] It should be noted that various aspects of embodiments within the scope of the appended claims are described below. It will be apparent that the aspects described herein can be embodied in a wide variety of forms, and any particular structure and / or function described herein is merely illustrative. Based on this disclosure, those skilled in the art will understand that one aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number of aspects set forth herein can be used to implement the device and / or practice the method. Additionally, this device and / or method can be implemented using other structures and / or functionalities besides one or more of the aspects set forth herein.

[0049] In addition, specific details are provided in the following description to facilitate a thorough understanding of the examples, and those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0050] Example 1

[0051] Figure 1 This paper presents a powder-spreading method based on machine vision and AI. By employing collaborative detection and discrimination of machine vision and AI recognition results, a dual verification mechanism is constructed, effectively overcoming the limitations of single detection methods. It leverages both the high precision and rigor of traditional machine vision methods in quantitative measurement and the adaptability of AI technology in identifying complex defects, thereby improving the reliability and accuracy of powder-spreading quality inspection results, reducing the false judgment rate, saving time, reducing material waste, and simultaneously solving the cost problem of manual monitoring. The specific steps include the following:

[0052] S1: Acquisition Steps: After the powder spreading operation is completed, acquire the powder spreading image of the current layer;

[0053] Powder bed image refers to a two-dimensional image of the surface of a single layer of powder bed that can be captured in real time during the powder bed fusion 3D printing process by a vision sensor (usually a high-speed linear or area array camera) installed above or inside the powder bed equipment, after the powder has been spread and before it has been melted by laser / electron beam scanning.

[0054] S11: The powder spreading equipment performs a powder spreading operation on the current layer, and after the powder spreading operation is completed, the controller of the powder spreading equipment sends a powder spreading completion signal;

[0055] The powder spreading device is the core actuator in the printing system. Its core function is to accurately, uniformly and reliably spread a layer of metal or non-metal powder of a specific thickness on the building platform, creating conditions for subsequent energy source (laser / electron beam) scanning and melting to manufacture a single-layer entity.

[0056] In the powder spreading process, the current layer refers to the layer of powder that is being spread or is about to be melted by the laser / energy source.

[0057] The printing system may also include monitoring equipment that can receive a powder spreading completion signal and trigger a camera set above the forming chamber to capture images of the powder spreading area and obtain powder spreading images of the current layer.

[0058] By responding to the powder spreading completion signal sent by the controller and triggering powder spreading image acquisition, valid powder spreading images can be obtained, avoiding invalid shooting and eliminating detection errors caused by shooting timing deviations. Moreover, no manual operation is required, avoiding human operation errors, thus providing an accurate data foundation for subsequent analysis.

[0059] S2: Preprocess the powder-spreading image of the current layer and extract powder-spreading abnormal areas;

[0060] Please refer to Figure 2 , Figure 2 This is a flowchart for extracting areas with abnormal powder spreading. The extraction of areas with abnormal powder spreading includes the following steps:

[0061] S21: Acquire multiple images of successfully applied powder and generate a standard background image using the image pixel averaging method. The images of successfully applied powder are acquired by performing multiple powder application operations in advance.

[0062] Obtain multiple images of successfully applied toner, including:

[0063] Multiple powder spreading experiments were conducted using the same type of powder;

[0064] Collect powder-spreading images that meet the preset requirements during the powder-spreading experiment to obtain multiple images of successful and normal powder-spreading.

[0065] Optionally, a standard background image is generated using the image pixel averaging method, and the standard pixel values ​​of the standard background image are expressed as follows:

[0066] ;

[0067] In the formula, Standard background image The Middle Line number The standard pixel values ​​after summing the columns; Indicates the first In Zhang Zheng's normal powdering image, the first Line number Column pixel values, , This indicates the total number of normally powder-coated images;

[0068] Similar powders refer to powder materials that have the same key characteristics such as chemical composition, particle size distribution, and morphology.

[0069] Multiple powder-spreading experiments were conducted using the same type of powder, specifically:

[0070] Multiple powder-spreading experiments were conducted on the same type of powder used in the powder-spreading method. In each powder-spreading experiment, the powder-spreading equipment was used to spread multiple layers of the same type of powder to obtain different powder-spreading images.

[0071] S22: Perform filtering and noise reduction processing on the powder-spreading image of the current layer, and perform image subtraction between the filtered and noise-reduced powder-spreading image of the current layer and the standard background image to obtain the difference image;

[0072] The expression for filtering and noise reduction is as follows:

[0073] ;

[0074] In the formula, Indicates the first Line number The output image pixel matrix of the column; This represents the median value after all pixel values ​​have been sorted. Represents row coordinates to Column coordinates arrive The input image pixel matrix; Indicates the kernel size.

[0075] S23: Perform opening operations on the difference images sequentially to remove noise interference, and closing operations to connect neighboring regions to obtain the abnormal image;

[0076] Opening is a fundamental operation in mathematical morphology, consisting of two basic operations, erosion and dilation, combined in a specific order (erosion first, then dilation), and is used to eliminate small objects.

[0077] Closing is a fundamental operation in mathematical morphology, consisting of two basic operations, erosion and dilation, combined in a specific order (dilation first, then erosion), and is used to fill small holes.

[0078] S24: A two-pass scanning algorithm is used to segment the abnormal image into connected components, resulting in multiple independent powder-spreading abnormal regions;

[0079] Please refer to Figure 3 , Figure 3 This is the detailed extraction flowchart for step S24. A two-pass scanning algorithm is used to segment the abnormal image into connected components, resulting in multiple independent powder-spreading abnormal regions. The specific steps include:

[0080] S241: Binarize the abnormal image to obtain a binary image;

[0081] S242: Perform two scans on the binary image; in the first scan, traverse each pixel row by row and column by column. If the current pixel is a foreground pixel and there are no labeled foreground neighbor pixels in the connected component of the current pixel, assign a first preset label to the current pixel; if the current pixel is a foreground pixel and there is only one labeled foreground neighbor pixel in the connected component of the current pixel, assign a second preset label to the current pixel; if the current pixel is a foreground pixel and there are two or more foreground neighbor pixels in the connected component of the current pixel, record that the labels of all pixels in the connected component are equal.

[0082] Foreground pixels are pixels in a binary image whose difference from the standard pixel value of the standard background image is greater than a preset threshold. The preset threshold can be set according to actual needs. Example: If the area corresponding to the current pixel has abnormal toner application (e.g., unsuccessful toner application), the difference between the current pixel value and the standard pixel value will be greater than the preset threshold, and the current pixel will be determined as a foreground pixel.

[0083] Foreground neighbor pixels refer to other pixels that are spatially directly adjacent to a given foreground pixel.

[0084] The connected component of a foreground pixel refers to the set of all foreground neighbor pixels that can be reached from that foreground pixel through specific connectivity rules.

[0085] Among them, the foreground neighbor pixels are determined according to connectivity rules, namely:

[0086] (1) When the connected component is 4-connectivity, check the pixels to its left and above;

[0087] (2) When the connected component is 8-connectivity, check the pixels to its left, upper left, upper right and right.

[0088] An example of S242 execution is as follows:

[0089] Starting from pixel (0,0) in the binary image, if the current pixel (0,0) is a foreground pixel and there are no labeled foreground neighbor pixels in its connected component, it is determined that the current pixel belongs to a new connected component, and a first preset label is assigned to the current pixel (0,0); the first preset label is a new and unique label.

[0090] When the current pixel (0,0) is a foreground pixel and there is only one labeled foreground neighbor pixel in its connected domain, a second preset label is assigned to the current pixel (0,0); the second preset label is the label of the only labeled foreground neighbor pixel in the connected domain of the current pixel;

[0091] When the current pixel (0,0) is a foreground pixel and there are two or more labeled foreground neighbor pixels in its connected domain, assign a label to the current pixel (0,0) of any foreground neighbor pixel (in specific implementation, it is usually assigned the smallest label among the labels of all foreground neighbor pixels), and record that the labels of all pixels in the connected domain of the current pixel are equal to indicate that they belong to the same connected domain.

[0092] The first scan is completed by iterating through each pixel in the binary image row by row and column by column until all pixels in the binary image have been traversed.

[0093] S243: During the second scan, pixels marked by labels with equal relationships are grouped into the same connected component, and a third preset label is assigned to pixels marked by labels with equal relationships.

[0094] S244: Based on the first preset label, the second preset label and the third preset label, the binary image is segmented into multiple connected components to obtain multiple independent powder spreading abnormal regions, ensuring the accuracy of the powder spreading abnormal regions, and thus providing an accurate data foundation for subsequent statistics.

[0095] By using the image pixel averaging method to generate a standard background image, the shooting error and interference factors of a single image are effectively offset, providing a reliable basis for subsequent image subtraction. By comparing and contrasting images to highlight anomalies and identify abnormal images, noise interference is removed through opening and closing operations, and the missegmentation of a single continuous defect into multiple fragmented regions is avoided, ensuring the accuracy of subsequent statistics.

[0096] S3: Calculate the area characteristics of the powder spreading anomaly region;

[0097] S31: Count the number of pixels contained in each powder spreading abnormal area, use the number of pixels as the area feature information of the corresponding powder spreading abnormal area, and store it in memory.

[0098] S4: Input the powder spreading image of the current layer into the pre-trained powder spreading defect recognition model to obtain the recognition result;

[0099] S41: Acquire multiple images of successful powder spreading and images of abnormal powder spreading, and use labeling software to mark the abnormal areas in each abnormal powder spreading image. The images of successful powder spreading and abnormal powder spreading are acquired by performing multiple powder spreading operations in advance.

[0100] In this embodiment, the abnormal powder spreading image includes an abnormal powder spreading image containing small areas of powder collapse. Its pixel value is very close to that of the image after powder spreading, and traditional algorithms cannot accurately distinguish it.

[0101] S42: Train the AI ​​large model using normal powder spreading images and labeled abnormal powder spreading images to obtain a powder spreading defect recognition model;

[0102] Optionally, the large AI model includes an input layer, a downsampling layer, an upsampling layer, and an upsampling layer, specifically configured as follows:

[0103] Input layer: 1) 3x3 convolutional layer, 2) 3x3 convolutional layer;

[0104] Downsampling layers: 1) 2x2 max pooling layer; 2) 3x3 convolutional layer; 3) 3x3 convolutional layer; 4) 2x2 max pooling layer; 5) 3x3 convolutional layer; 6) 3x3 convolutional layer; 7) 2x2 max pooling layer; 8) 3x3 convolutional layer; 9) 3x3 convolutional layer; 10) 2x2 max pooling layer; 11) 3x3 convolutional layer; 12) 3x3 convolutional layer;

[0105] Upsampling layers: 1) 2x2 upsampling; 2) 3x3 convolutional layer; 3) 3x3 convolutional layer; 4) 2x2 upsampling; 5) 3x3 convolutional layer; 6) 3x3 convolutional layer; 7) 2x2 upsampling; 8) 3x3 convolutional layer; 9) 3x3 convolutional layer; 10) 2x2 upsampling; 11) 3x3 convolutional layer; 12) 3x3 convolutional layer;

[0106] Output layer: 1) 1x1xk convolutional layer, where k is the number of convolutional kernels.

[0107] Furthermore, the training of large AI models includes:

[0108] Set training parameters:

[0109] Learning rate: 1e-5; Optimizer: adam; Batch size: 4; Number of epochs: 128; Activation function: ReLU;

[0110] Training process:

[0111] 1. Preparation stage:

[0112] 1.1 Input images (normal powder spreading images and marked abnormal powder spreading images);

[0113] 1.2 Labels: Each pixel in the mask is labeled with its category (0 is background);

[0114] 1.3 Dataset partitioning: training set, validation set, and test set;

[0115] 2. Iterative training phase:

[0116] 2.1 Forward propagation: Inputting data from the training set into the AI ​​model in batches;

[0117] 2.2 Calculate the loss: Compare the predicted segmentation map output by the model with the corresponding true labels;

[0118] 2.3 Backpropagation and Optimization: Based on the loss value, the backpropagation algorithm is used to calculate the gradient of the loss function with respect to all weights and biases of the model, and then the weights are updated;

[0119] 3. Evaluation Phase:

[0120] 3.1 Use the validation set for forward propagation, calculate the model's evaluation metrics on the validation set, and save the best weight file;

[0121] 4. Termination and Testing:

[0122] 4.1 Training continues until the set total number of batches is reached, at which point training is stopped, and the powder spreading defect recognition model is obtained; the input of the powder spreading defect recognition model is the powder spreading image of the current layer, and the output is the defect recognition result.

[0123] S43: Use the powder spreading defect recognition model to perform AI recognition on the powder spreading image of the current layer to obtain the recognition result;

[0124] By training a large AI model, a powder-laying defect recognition model was obtained, which improved the detection capability of powder-laying image defects. By effectively identifying minute defects, it supplemented the blind spots of conventional machine vision detection, expanded the scope of defect detection, and improved the comprehensiveness of detection.

[0125] S5: Compare the identification results with the area feature information of the abnormal powder spreading area;

[0126] S51: When the area feature information of the abnormal toner application area is within the predetermined area range and the identification result indicates that the toner application is normal, execute the next layer printing work;

[0127] In this embodiment, the machine vision method is used to determine whether the current layer has been successfully coated by comparing the area feature information of the abnormal coating area with the predetermined area range, and the AI ​​method is used to determine whether the current layer has been successfully coated by the recognition result.

[0128] S52: When the area feature information of the abnormal powder spreading area is not within the predetermined area range, or the identification result indicates that the powder spreading is abnormal, execute the next powder spreading operation and return to the acquisition step;

[0129] S521: If the toner is successfully applied to the current layer for the second time, the toner application equipment will continue to print the next layer.

[0130] S522: If the current layer fails to spread toner twice in a row, stop printing.

[0131] By setting a rule for two failed powder spreading attempts, material waste and potential hazards caused by repeated ineffective powder spreading are avoided. An alarm is issued to prompt manual intervention, ensuring production safety.

[0132] Specifically, if the toner application for the current layer fails twice in a row, the printing system will stop printing and issue an alarm until the alarm is cleared after manual troubleshooting.

[0133] Please refer to Figure 4 , Figure 4 This is a flowchart of a toner-spreading method based on machine vision and AI. In a specific embodiment, the printing system starts printing, and the toner-spreading device performs a toner-spreading operation in the forming chamber. After toner-spreading is completed, an image of the current layer's toner-spreading is captured. This image is then preprocessed to extract abnormal toner-spreading areas and calculate their area features. The image is then input into a pre-trained toner-spreading image recognition model to obtain a recognition result. The toner-spreading quality is then assessed by combining the area features of the abnormal areas with the recognition result. The criteria for assessment are whether the area features of the abnormal areas are within a predetermined area range and whether the recognition result indicates normal toner-spreading. If all conditions are met, the toner-spreading for the current layer is considered successful, and the printing system proceeds to the next layer. If any condition is not met, the toner-spreading for the current layer is considered a failure, and the toner-spreading operation is repeated. If the failure persists, both toner-spreading attempts are considered failures, and the printing system issues an alarm until the fault is manually resolved and the alarm is cleared.

[0134] Example 2

[0135] like Figure 5As shown, this embodiment provides a printing system that implements a powder spreading method based on machine vision and AI, including an acquisition module 501, a calculation module 502, and a detection and discrimination module 503.

[0136] Specifically, the acquisition module 501 is used to acquire the powder-spreading image of the current layer after the powder-spreading operation is completed;

[0137] Specifically, the calculation module 502 is used to preprocess the powder-spreading image of the current layer, extract the powder-spreading abnormal region and calculate the area feature information of the powder-spreading abnormal region; and input the powder-spreading image of the current layer into the pre-trained powder-spreading defect recognition model to obtain the recognition result;

[0138] Specifically, the detection and discrimination module 503 is used to compare the recognition result with the area feature information of the abnormal toner application area; when the area feature information of the abnormal toner application area is within the predetermined area range and the recognition result indicates that the toner application is normal, the next layer printing operation is performed; when the area feature information of the abnormal toner application area is not within the predetermined area range, or the recognition result indicates that the toner application is abnormal, the next toner application operation is performed and the process is returned to the acquisition module 501.

[0139] Example 3

[0140] Based on the same technical concept, embodiments of this application also provide a computer device, including a memory 1 and a processor 2, such as... Figure 6 As shown, memory 1 stores a computer program, and processor 2 executes the computer program to implement any of the above-mentioned powder spreading methods.

[0141] The memory 1 includes at least one type of readable storage medium, including flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 1 can be an internal storage unit of the printing system, such as a hard disk. In other embodiments, the memory 1 can be an external storage device of the printing system, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. Furthermore, the memory 1 can include both internal and external storage units of the printing system. The memory 1 can be used not only to store application software and various types of data installed in the printing system, such as the code of the printing system program, but also to temporarily store data that has been output or will be output.

[0142] In some embodiments, processor 2 may be a central processing unit (CPU), controller, microcontroller, microprocessor or other data processing chip, used to run program code stored in memory 1 or process data, such as a computer program that executes a powder spreading method.

[0143] The present invention also discloses a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the method described in the powder spreading method embodiments. The storage medium may be a volatile or non-volatile computer-readable storage medium.

[0144] The computer program product of the powder spreading method based on machine vision and AI provided in the embodiments of the present invention includes a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the steps of the method in the above method embodiments. For details, please refer to the above method embodiments, which will not be repeated here.

[0145] The present invention also discloses a computer program that, when executed by a processor, implements any of the methods described in the foregoing embodiments. This computer program product can be implemented specifically through hardware, software, or a combination thereof. In one optional embodiment, the computer program product is specifically embodied in a computer storage medium; in another optional embodiment, the computer program product is specifically embodied in a software product, such as a software development kit (SDK), etc.

[0146] It is understood that the same or similar parts in the above embodiments can be referred to each other, and the contents not described in detail in some embodiments can be referred to the same or similar contents in other embodiments.

[0147] It should be noted that in the description of this invention, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance. Furthermore, in the description of this invention, unless otherwise stated, "a plurality of" means at least two.

[0148] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process, and the scope of the preferred embodiments of the invention includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as will be understood by those skilled in the art to which embodiments of the invention pertain.

[0149] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0150] Those skilled in the art will understand that all or part of the steps of the methods described in the above embodiments can be implemented by a program instructing related hardware, and the program can be stored in a computer-readable storage medium. When executed, the program includes one or a combination of the steps of the method embodiments.

[0151] Furthermore, the functional units in the various embodiments of the present invention can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0152] The storage media mentioned above can be read-only memory, disk, or optical disk, etc.

[0153] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0154] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for spreading powder, characterized in that, include: Acquisition steps: After the powder spreading operation is completed, acquire the powder spreading image of the current layer; The powder-spreading image of the current layer is preprocessed to extract powder-spreading abnormal regions, including: Multiple images of successfully applied powder are acquired, and a standard background image is generated by calculating the average of image pixels. The images of successfully applied powder are acquired by performing multiple powder application operations in advance. The powder-spreading image of the current layer is filtered and denoised, and the filtered and denoised powder-spreading image of the current layer is compared with the standard background image to obtain a difference image; The difference image is then subjected to opening operations to remove noise interference, and closing operations to connect neighboring regions to obtain the abnormal image. A two-pass scanning algorithm was used to segment the abnormal image into connected components, resulting in multiple independent powder-spreading abnormal regions. Calculate the area characteristics of areas with abnormal powder spreading; The powder-laying image of the current layer is input into a pre-trained powder-laying defect recognition model to obtain recognition results, including: Multiple images of successful powder application and images of abnormal powder application are acquired, and the abnormal areas in each of the abnormal powder application images are marked using tagging software. The normal powder application images and the abnormal powder application images are acquired by performing multiple powder application operations in advance. The AI ​​model was trained using the normal powder spreading images and the labeled abnormal powder spreading images to obtain a powder spreading defect recognition model; The powder spreading defect recognition model is used to perform AI recognition on the powder spreading image of the current layer to obtain the recognition result; Compare the identification results with the area feature information of the powder spreading abnormal area; When the area feature information of the abnormal powder spreading area is within the predetermined area range, and the identification result indicates that the powder spreading is normal, the next layer of printing is executed; If the area feature information of the abnormal powder application area is not within the predetermined area range, or if the identification result indicates an abnormal powder application, the next powder application operation is performed, and the process returns to the acquisition step.

2. The powder spreading method according to claim 1, characterized in that, The area feature information of the calculated powder-spreading anomaly region includes: The number of pixels contained in each of the powder-spreading abnormal regions is counted, and the number of pixels is used as the area feature information of the corresponding powder-spreading abnormal region and stored in memory.

3. The powder spreading method according to claim 1, characterized in that, The two-pass scanning algorithm is used to segment the abnormal image into connected components, resulting in multiple independent powder-spreading abnormal regions, including: The abnormal image is binarized to obtain a binary image; The binary image is scanned twice. In the first scan, each pixel is traversed row by row and column by column. If the current pixel is a foreground pixel and there are no labeled foreground neighbor pixels in the connected component of the current pixel, a first preset label is assigned to the current pixel. If the current pixel is a foreground pixel and there is only one labeled foreground neighbor pixel in the connected component of the current pixel, a second preset label is assigned to the current pixel. If the current pixel is a foreground pixel and there are two or more foreground neighbor pixels in the connected component of the current pixel, the labels of all pixels in the connected component are recorded as equal. During the second scan, pixels marked by labels with equal relationships are grouped into the same connected component, and a third preset label is assigned to pixels marked by labels with equal relationships. Based on the first preset label, the second preset label, and the third preset label, the binary image is segmented into multiple connected components to obtain multiple independent powder spreading anomaly regions.

4. The powder spreading method according to claim 1, characterized in that, The acquisition of multiple successfully applied powder images includes: Multiple powder spreading experiments were conducted using the same type of powder; Collect powder-spreading images that meet the preset requirements during the powder-spreading experiment to obtain multiple images of successful powder-spreading.

5. The powder spreading method according to claim 1, characterized in that, If the current layer fails to spread toner twice consecutively, the printing process will stop.

6. A printing system implementing the powder spreading method according to any one of claims 1 to 5, characterized in that, include: The acquisition module is used to acquire the powder-spreading image of the current layer after the powder-spreading operation is completed; The calculation module is used to preprocess the powder spreading image of the current layer, extract the powder spreading abnormal area, and calculate the area feature information of the powder spreading abnormal area; The powder-laying image of the current layer is input into a pre-trained powder-laying defect recognition model to obtain the recognition result; The detection and discrimination module is used to compare the identification result with the area feature information of the powder spreading abnormal area; If the area feature information of the abnormal toner application area is within the predetermined area range and the identification result indicates that the toner application is normal, the next layer printing operation is performed; if the area feature information of the abnormal toner application area is not within the predetermined area range, or the identification result indicates that the toner application is abnormal, the next toner application operation is performed, and the process returns to the acquisition module.

7. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the powder spreading method according to any one of claims 1 to 5.

8. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the powder spreading method according to any one of claims 1 to 5.

Images