Character jet printing mistake proofing system suitable for thick plate

Abstract

The invention discloses a character jet printing mistake proofing system suitable for a thick plate, and belongs to the field of thick plate finishing. The system comprises a communication module, an image acquisition module, an image processing module, a character recognition module and a man-machine interaction module. The core of the method is as follows: after steel plate jet printing is completed, a system automatically triggers image acquisition, interference such as on-site visible light change, steel plate temperature and material reflection is effectively overcome through unique image processing procedures (including geometric correction based on camera calibration, strong light suppression, adaptive gray processing and binaryzation), and a clear image is obtained; and then, the OCR algorithm is used for recognizing the content of the jet printing characters, and the recognition result is doubly compared with the jet printing information required by the production system and the actual size measurement value of the steel plate, so that automatic and real-time accurate mistake proofing is realized. According to the invention, manual inspection is comprehensively replaced by machine vision, the detection accuracy and efficiency of the thick plate jet printing information are remarkably improved, and a closed-loop error-proofing mechanism is established.

Description

Technical Field

[0001] This invention relates to the finishing of thick plates, and more particularly to a character printing error prevention system suitable for thick plates. Background Technology

[0002] In the thick plate manufacturing industry, for ease of identification and management, characters need to be printed on the surface of steel plate products using online fully automated marking devices or offline equipment. Common marking methods include surface printing, engraving, and side printing. These markings contain key information such as the steel plate's specifications, type, and batch number, serving as the steel plate's "identity ID," and their accuracy is paramount. Currently, the industry commonly uses manual sampling to monitor printing quality, that is, manually checking whether the printed information matches the actual production results at a certain frequency. This method has significant limitations:

[0003] (1) Low reliability and high quality risk: Manual inspection is easily affected by factors such as fatigue and subjective judgment, and there is a risk of missed inspection and wrong inspection. It cannot guarantee 100% accuracy and may lead to quality accidents where the information is inconsistent with the actual product.

[0004] (2) Inefficient and unable to provide real-time feedback: The sampling frequency is limited, and it is impossible to inspect each steel plate in real time. When an error is discovered, a large number of unqualified products may have already been produced, resulting in rework costs and production efficiency losses.

[0005] (3) Lack of objective records: Manual inspection makes it difficult to form systematic and traceable test records, which is not convenient for quality traceability and analysis.

[0006] At the character recognition algorithm level, domestic and international research has attempted to apply deep learning and image processing technologies to thick plate inkjet character recognition. For example, improved BP neural networks, template matching methods, multi-level cutting projection algorithms, or support vector machines (SVMs) have been used. While these technologies can achieve high recognition rates in laboratory or ideal environments, they still face challenges in actual industrial applications. The complex environment of thick plate production sites, including variations in visible light, high temperatures (hot state) of the steel plate itself, and surface reflectivity, severely affects image acquisition quality, leading to a decrease in the recognition accuracy of existing algorithms. Existing research mainly focuses on optimizing the character recognition algorithm itself, lacking a complete system solution that automatically and in real-time compares the character recognition results with the production management system to achieve dual error prevention.

[0007] Therefore, the existing technology has the following core problems that urgently need to be solved: it relies on manual spot checks and cannot achieve fully automatic, real-time and highly reliable printing information detection; it lacks a robust image acquisition and processing solution that can effectively resist interference from complex industrial environments (such as light changes, high temperatures and reflections); and it lacks a closed-loop error prevention mechanism that automatically verifies the character recognition results with the actual information of the steel plate (such as the printing content required by the system and the size measurement values). Summary of the Invention

[0008] Purpose of the invention: The purpose of this invention is to provide an automated error prevention system that can overcome the above-mentioned defects, replace manual labor with machine vision technology, achieve intelligent error prevention, and fundamentally eliminate the occurrence of printing errors.

[0009] Technical Solution: A character printing error-proofing system suitable for thick plates, comprising a communication module, an image acquisition module, an image processing module, a character recognition module, and a human-computer interaction module; the communication module is configured to acquire steel plate information and a start recognition signal from the production line system and send the detection result; the image acquisition module is configured to acquire an image of the steel plate after receiving the start recognition signal; the image processing module is configured to perform distortion correction, grayscale adjustment, and binarization processing on the acquired image; the character recognition module is configured to recognize the character content in the processed image using an OCR algorithm; and the human-computer interaction module is configured to display the recognition result.

[0010] A method for error-proofing character printing on thick plates includes the following steps:

[0011] S1. After the steel plate moves to the identification position and completes the printing, the production line system sends a start identification signal, which is received by the communication module.

[0012] S2. The image acquisition module responds to the start recognition signal and acquires an image of the steel plate;

[0013] S3. The image processing module performs image correction processing to eliminate distortion, and then performs grayscale adjustment and binarization processing.

[0014] S4. The character recognition module uses the OCR algorithm to recognize character content and compares the recognition results with production performance data.

[0015] S5. Store the recognition results and perform statistical analysis;

[0016] S6. Send the recognition results to the production line system and display them.

[0017] Furthermore, the image processing module employs a geometric correction method based on camera intrinsic and extrinsic parameters to correct distortion by estimating the transformation matrix through camera parameter calibration.

[0018] Furthermore, the grayscale adjustment employs an automatic thresholding method, calculating the optimal threshold based on the image grayscale histogram to maximize the inter-class variance. The threshold selection is based on the formula:

[0019]

[0020] Where t is the threshold, L is the gray level, N is the total number of pixels, θ1 and θ2 are the area ratios of the regions, and u1 and u2 are the average gray levels of the regions; The largest value is the optimal threshold.

[0021] Furthermore, in step S3, the distortion correction process performs distortion correction by calculating the camera's intrinsic and extrinsic parameters, and the grayscale processing uses an automatic thresholding method for binarization.

[0022] Furthermore, in step S4, the comparison includes comparing the identified character content with the steel plate size measurement value to achieve double error prevention and trigger an alarm when there is a discrepancy.

[0023] Furthermore, the communication module is also configured to buffer the recognition signal and send the recognition result to the production line system in a specified message format; the human-machine interaction module is also configured to count the number of correct recognitions, the number of failures, and the recognition rate on a daily basis.

[0024] Furthermore, in step S1, the communication module will begin to identify signal flag position 1 to indicate that there is a signal; in step S5, the identification result is stored in the database and statistically analyzed.

[0025] Furthermore, the system is applicable to various environments with different visible light, steel plate temperatures, and steel plate materials, and obtains clear images through strong light suppression and image processing.

[0026] Furthermore, the OCR algorithm of the character recognition module is configured to process contiguous characters, and to improve recognition accuracy by combining a multi-level cutting projection algorithm or a support vector machine algorithm.

[0027] Beneficial effects: After the printed content is identified, the system compares the identified steel plate number with the corresponding printed information against the system's printing requirements, and simultaneously compares it with existing size detection results, performing a double information check to prevent printing errors and other problems. It overcomes environmental factors such as visible light, steel plate temperature, and steel plate material, ensuring the camera can capture clear images. Through correction, grayscale processing, and OCR character recognition, it accurately identifies the printed content on the steel plate and prevents missed or incorrect detections due to human factors, improving the recognition rate and work efficiency. Attached Figure Description

[0028] Figure 1 This is a diagram showing the connection of system modules;

[0029] Figure 2 This is a schematic diagram of the system architecture;

[0030] Figure 3 This is a flowchart illustrating the method. Detailed Implementation

[0031] To make the technical solution of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0032] A character printing error-proofing system suitable for thick plates consists of a communication module, an image acquisition module, a character recognition module, an image processing module, and a human-computer interaction module. Specific connections are as follows: Figure 1 As shown.

[0033] The communication module is used to acquire steel plate information from the production line system, initiate recognition signals, send detection results, and interface with the human-machine interface. Upon receiving the initiate recognition signal, the communication module acquires the image from the image processing module on the production line system and uploads it. After processing including distortion correction, grayscale adjustment, image binarization, and ROI region setting, the image is sent to the character recognition module. The OCR algorithm recognizes the characters, and the control system compares the results and determines whether an alarm should be triggered. After comparison, the result is returned to the communication module. Finally, the communication module uploads the result to the human-machine interface for display and sends the recognition result back to the production line system. The system architecture is as follows: Figure 2 As shown.

[0034] The automatic steel plate number verification method includes the following steps:

[0035] S1. The steel plate moves to the identification position and printing begins. After printing is completed, the production line system sends an identification signal, which is received by the communication module.

[0036] S2. After receiving the recognition signal, the image acquisition module calls the camera SDK library to capture the image and then sends it to the image processing module.

[0037] S3. Since the lens and the recognition plane are at a certain angle during image acquisition, the image processing module first performs correction processing after receiving the image. It uses a linear extension algorithm to suppress strong light and eliminate the optical effects caused by ambient light, temperature, material, etc. Then, it uses the grayscale processing interface to process the image to obtain a grayscale image for subsequent binarization processing. The binarization interface is used to obtain a binarized image, and after the recognition area is marked on the image, it is sent to the character recognition module.

[0038] S4. The character recognition module uses the OCR algorithm to recognize the character content in the marked area. After the recognition is completed, the recognition result is compared with the data in the production record, and it is determined whether to issue an alarm to remind the operator to intervene.

[0039] S5. After completing the verification of the identification results, store the identification results in the database, and statistically analyze the number of correct identifications, the number of failures, and the identification rate on a daily basis. Also, organize the identification verification results into a message according to the specified message format and send it to the production control system.

[0040] S6. Send the assembled identification result message from the identification and statistics module to the production control system server for reception. After receiving the message, the identification result is displayed on the control terminal.

[0041] The specific algorithm flow is as follows: Figure 3 As shown.

[0042] In S1, the communication module server sets the identification signal flag in the Start_Scan cache to 1 (0 for no signal, 1 for identification signal) for subsequent determination of whether an identification signal has been sent.

[0043] In S3, distortion correction utilizes a geometric correction method, a distortion correction method based on the camera's intrinsic and extrinsic parameters. The principle behind this method is to estimate the transformation matrix required for distortion correction by calculating the camera's internal and external parameters. In implementation, the camera is first calibrated by repeatedly photographing a specific calibration object to obtain the camera's internal and external parameters, and then these parameters are used for distortion correction.

[0044] In S3, grayscale processing is based on repeatedly testing to obtain the most ideal threshold parameter, which is then used to process the acquired color image to produce a grayscale image. Binarization, on the other hand, obtains a black and white image within a specified threshold range, which is more beneficial for text recognition. Grayscale processing better preserves image details and brightness information; while binarization simplifies the image and highlights edge information.

[0045] The grayscale processing employs an automatic thresholding method. Its basic principle is based on the image's grayscale histogram, using the maximum inter-class variance between the target and background as the threshold selection criterion, comprehensively considering the characteristics of pixel neighborhood and overall image grayscale distribution. The histogram determines the area ratio of regions 1 and 2 separated by the threshold t, as well as the average grayscale of the entire image, regions 1, and regions 2.

[0046] Let t be the threshold of the entire image, L be the gray levels of the image, and N be the total number of pixels. Therefore, the image can be divided into two regions: [0, t] and [t+1, L-1], named region 1 and region 2, respectively. i Let θ be the number of pixels with gray value i, and let θ1 and θ2 be the area proportions of region 1 and region 2 in the entire image, respectively.

[0047] The pixel area ratio of region 1 is:

[0048]

[0049] The pixel area ratio of region 2 is:

[0050]

[0051] The average gray level of the entire image is:

[0052]

[0053] The average gray level of region 1 is:

[0054]

[0055] The average gray level of region 2 is:

[0056]

[0057] The variance between regions can be expressed as:

[0058]

[0059] make The largest value is the optimal threshold. Typically, an image pixel size of 60*80 is used, and L is 25. The resulting threshold varies under different lighting intensities.

[0060] Once the printed content is identified, the corresponding printed information based on the identified steel plate number is compared with the printed content required by the system, and at the same time compared with the existing size detection results to perform double information checks to prevent problems such as printing errors or incorrect information.

[0061] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A character printing error-proofing system suitable for thick plates, characterized in that, The system includes a communication module, an image acquisition module, an image processing module, a character recognition module, and a human-computer interaction module. The communication module is configured to acquire steel plate information and a start recognition signal from the production line system and send the detection results. The image acquisition module is configured to acquire an image of the steel plate after receiving the start recognition signal. The image processing module is configured to perform distortion correction, grayscale adjustment, and binarization processing on the acquired image. The character recognition module is configured to recognize the character content in the processed image using an OCR algorithm. The human-computer interaction module is configured to display the recognition results.

2. A method for error-proofing character printing suitable for thick plates, characterized in that, Includes the following steps: S1. After the steel plate moves to the identification position and completes the printing, the production line system sends a start identification signal, which is received by the communication module. S2. The image acquisition module responds to the start recognition signal and acquires an image of the steel plate; S3. The image processing module performs image correction processing to eliminate distortion, and then performs grayscale adjustment and binarization processing. S4. The character recognition module uses the OCR algorithm to recognize character content and compares the recognition results with production performance data. S5. Store the recognition results and perform statistical analysis; S6. Send the recognition results to the production line system and display them.

3. The character printing error-proofing system for thick plates according to claim 1, characterized in that, The image processing module employs a geometric correction method based on camera intrinsic and extrinsic parameters, which corrects distortion by estimating the transformation matrix through camera parameter calibration.

4. The character printing error-proofing system for thick plates according to claim 1, characterized in that, The grayscale adjustment employs an automatic thresholding method, calculating the optimal threshold based on the image grayscale histogram to maximize the inter-class variance. The threshold selection is based on the formula: Where t is the threshold, L is the gray level, N is the total number of pixels, θ1 and θ2 are the area ratios of the regions, and u1 and u2 are the average gray levels of the regions; The largest value is the optimal threshold.

5. The character printing error prevention method for thick plates according to claim 2, characterized in that, In step S3, distortion correction is performed by calculating the camera's intrinsic and extrinsic parameters, and grayscale processing uses an automatic thresholding method for binarization.

6. The character printing error prevention method for thick plates according to claim 2, characterized in that, In step S4, the comparison includes comparing the identified character content with the steel plate size measurement value to achieve double error prevention and trigger an alarm when there is a discrepancy.

7. The character printing error-proofing system for thick plates according to claim 1, characterized in that, The communication module is also configured to buffer the recognition signal and send the recognition result to the production line system in a specified message format; the human-machine interaction module is also configured to count the number of correct recognitions, the number of failures, and the recognition rate on a daily basis.

8. The character printing error prevention method for thick plates according to claim 2, characterized in that, In step S1, the communication module will begin to identify signal flag position 1 to indicate that there is a signal; in step S5, the identification result is stored in the database and statistically analyzed.

9. The character printing error prevention system for thick plates according to claim 1, characterized in that, The system is applicable to various environments with different visible light, steel plate temperatures, and steel plate materials, and obtains clear images through strong light suppression and image processing.

10. The character printing error-proofing system for thick plates according to claim 1, characterized in that, The OCR algorithm of the character recognition module is configured to handle contiguous characters, and combines a multi-level cutting projection algorithm or a support vector machine algorithm to improve recognition accuracy.

Images