Hot-rolled steel surface defect analysis system

Abstract

The invention relates to the technical field of industrial automation, and discloses a hot rolled steel surface defect analysis system. According to the system, a surface image of newly produced steel is acquired in real time through the data acquisition module by using a high-definition camera, the image preprocessing module performs graying, noise reduction, zooming and color enhancement on the image, the image quality is improved, the defect detection and identification module performs edge detection and texture analysis on the processed image by using a computer vision technology, and the defect detection and identification module performs defect identification on the surface of the newly produced steel. The defect analysis module further analyzes the area, depth and length of the defects and provides a basis for classification of the defects, the statistical report module generates a detailed defect statistical report according to an analysis result, and the process adjustment module automatically adjusts production process parameters according to the report. Compared with a traditional method depending on manual detection, the system has the advantages that the steel detection efficiency and accuracy are remarkably improved, the interference of human factors is reduced, and the consistency of detection results is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial automation, and in particular to a hot-rolled steel surface defect analysis system. Background Art

[0002] In the modern steel industry, the production process of hot-rolled steel places increasingly higher demands on the quality of steel. Surface defects are an important factor affecting the performance and service life of steel. Common defects include scratches, bubbles, cracks, pits, etc. These defects not only affect the physical properties of steel, but also cause serious problems in subsequent processing and use. Therefore, timely and accurate detection and analysis of steel surface defects are of great significance for improving product quality, reducing production costs and enhancing market competitiveness.

[0003] Traditional steel defect detection methods mostly rely on manual visual inspection and simple instrument detection. This method is not only inefficient but also easily affected by human factors, making it difficult to ensure the accuracy and consistency of detection. Summary of the Invention

[0004] In response to the shortcomings of the existing technology, the present invention provides a hot-rolled steel surface defect analysis system. The data acquisition module uses a high-definition camera to obtain surface images of newly produced steel in real time, and transmits them to the image preprocessing module using a wireless network. The latter grayscales, reduces noise, scales and enhances the color of the image to improve the image quality. The defect detection and recognition module uses advanced computer vision technology to perform edge detection and texture analysis on the processed image, so as to accurately identify the type, location, size and number of defects, and transmits the results to the defect analysis module. The module further analyzes the area, depth and length of the defects to provide a basis for defect classification. The statistical reporting module generates a detailed defect statistical report based on the analysis results for reference by the process adjustment module. The latter can automatically adjust the production process parameters according to the report. Compared with the traditional method that relies on manual inspection, the system significantly improves the inspection efficiency and accuracy, reduces the interference of human factors, ensures the consistency of the inspection results, provides strong technical support for steel production enterprises, optimizes the production process, improves product quality and reduces production costs.

[0005] To achieve the above-mentioned object, the present invention provides the following technical solutions: a hot-rolled steel surface defect analysis system, comprising a data acquisition module, an image preprocessing module, a defect detection and recognition module, a defect analysis module, a statistical reporting module, and a process adjustment module;

[0006] The data acquisition module obtains the surface image of the newly produced steel material by connecting to a high-definition camera, and transmits the obtained surface image of the newly produced steel material to the image preprocessing module via a wireless network;

[0007] The image preprocessing module performs image grayscale conversion, image noise reduction, image scaling, and image color enhancement on the newly produced steel surface image, and the processed steel surface image is transmitted to the defect detection and recognition module;

[0008] The defect detection and recognition module uses computer vision technology to perform edge detection and texture analysis on the processed steel surface image and then identifies steel defects, and outputs the steel defect type, defect location, defect size and defect quantity to the defect analysis module;

[0009] The defect analysis module further analyzes and classifies the steel defects according to their type, location, size, and number, and calculates the defect area, depth, and length of the steel defects.

[0010] The statistical reporting module generates a steel defect statistical report based on the detailed analysis results of the defect analysis module and sends it to the process adjustment module;

[0011] The process adjustment module automatically adjusts steel production process parameters according to the steel defect statistical report.

[0012] Preferably, the formula for graying the image is as follows:

[0013] K(x,y)=0.299*R(x,y)+0.587*G(x,y)+0.114*B(x,y)

[0014] In the formula, K(x,y) represents the grayscale value at the pixel position (x,y), R(x,y) represents the red component in the color space, G(x,y) represents the green component in the color space, and B(x,y) represents the blue component in the color space.

[0015] Preferably, the image noise reduction formula is as follows:

[0016]

[0017] In the formula, I ′ (x, y) represents the pixel value after filtering, I(x+u, y+v) represents the pixel value of the original grayscale image at position (x+u, y+v), N represents the total number of pixels in the filtering window, and u and v represent the window offset parameters.

[0018] Preferably, the image scaling formula is as follows:

[0019]

[0020] In the formula, I s (x′, y′) represents the pixel value of the scaled image at the new pixel position, Indicates that the coordinates mapped to the original image x′ are calculated based on the scaling factor s. Indicates that the coordinates mapped to the original image y′ are calculated based on the scaling factor s, round(*) indicates that the coordinates are rounded to the nearest pixel point, and I(*,*) indicates the pixel value of the original image at the corresponding position.

[0021] Preferably, the formula for image color enhancement is as follows:

[0022] I c (x,y)=α*I g (x,y)+β

[0023] In the formula, I c (x,y) represents the enhanced color value, I g (x,y) represents the original color value, α represents the scaling factor, and β represents the offset.

[0024] Preferably, the edge detection formula is as follows:

[0025] Y(x,y)=|G x (x,y)|+|G y (x,y)|

[0026] In the formula, Y(x,y) represents the strength value of edge detection, G x (x,y) represents the horizontal gradient of the image, G y (x,y) represents the vertical gradient of the image.

[0027] Preferably, the formula for texture analysis is as follows:

[0028]

[0029] In the formula, Cl represents contrast, P(i,j) represents the probability of position (i,j) in the gray-level co-occurrence matrix, and N represents the number of gray levels.

[0030] Preferably, the formula for calculating the defect area of ​​the steel is as follows:

[0031]

[0032] In the formula, A defect Represents the defect area of ​​the steel, M(p,q) represents the defect pixel mask, 1 represents a defect, 0 represents a non-defect, S represents the actual area represented by each pixel, W and H represent the width and height of the defect mask, and p and q represent the position index of the pixel.

[0033] Preferably, the formula for calculating the defect depth of the steel is as follows:

[0034]

[0035] In the formula, D represents the defect depth of the steel, D ref Indicates the reference depth value, L ref represents the reference grayscale, and L represents the average grayscale of the defect area.

[0036] Preferably, the formula for calculating the defect length of the steel is as follows:

[0037] Lc defect =n pixel s*Sc x

[0038] In the formula, Lc defect Indicates the defect length of steel, n pixels Indicates the number of pixels in the defect outline, Sc x Indicates the actual length corresponding to each pixel.

[0039] Compared with the prior art, the present invention provides a hot-rolled steel surface defect analysis system with the following beneficial effects:

[0040] The present invention uses a high-definition camera through the data acquisition module to obtain surface images of newly produced steel in real time, and transmits them to the image preprocessing module using a wireless network. The latter grayscales, reduces noise, scales and color enhances the image to improve the image quality. The defect detection and recognition module uses advanced computer vision technology to perform edge detection and texture analysis on the processed image, so as to accurately identify the type, location, size and number of defects, and pass the results to the defect analysis module. The module further analyzes the area, depth and length of the defects to provide a basis for defect classification. The statistical reporting module generates a detailed defect statistical report based on the analysis results for reference by the process adjustment module. The latter can automatically adjust the production process parameters according to the report. Compared with the traditional method that relies on manual detection, the system significantly improves the detection efficiency and accuracy, reduces the interference of human factors, ensures the consistency of the detection results, provides strong technical support for steel production enterprises, optimizes the production process, improves product quality and reduces production costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 Schematic diagram of the system flow of the present invention. DETAILED DESCRIPTION

[0042] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0043] In view of the fact that traditional steel defect detection methods mostly rely on manual visual inspection and simple instrument detection, this method is not only inefficient but also easily affected by human factors, making it difficult to ensure the accuracy and consistency of detection, a hot-rolled steel surface defect analysis system is proposed. Figure 1 ,The system includes a data acquisition module, an image preprocessing module, a defect detection and recognition module, a defect analysis module, a statistical reporting module and a process adjustment module;

[0044] The data acquisition module uses a high-resolution industrial HD camera, which ensures image clarity and detail capture through a dedicated optical lens. It is equipped with automatic focus and light source adjustment functions to adapt to different ambient lighting conditions. The camera is connected to the image acquisition hardware in the industrial control system and synchronously captures steel surface images through a high-speed data interface (such as GigE Vision or USB3.0). The built-in image processing chip performs preliminary preprocessing (such as brightness adjustment and exposure control). The collected image data is seamlessly transmitted via a wireless communication module (supporting Wi-Fi 6, 5G, or industrial-specific wireless communication protocols), ensuring fast and stable long-distance data transmission in the production site environment. The entire data acquisition system has anti-interference, data encryption and real-time transmission capabilities, ensuring the complete capture and lossless transmission of micro-defects on the steel surface during dynamic production processes, providing reliable basic data support for subsequent image preprocessing and defect analysis.

[0045] In the hot-rolled steel surface defect detection system, the image preprocessing module optimizes the surface images of newly produced steel by using a variety of advanced image processing technologies to improve the accuracy and robustness of subsequent defect detection. First, grayscale processing is used to convert the color image into a single-channel grayscale image. The formula is as follows:

[0046] K(x,y)=0.299*R(x,y)+0.587*G(x,y)+0.114*B(x,y)

[0047] In color images, color changes may interfere with defect detection. Grayscaling removes color information, highlights surface texture and brightness differences, and helps identify the boundaries and structures of defects such as cracks, trachoma, and dents, improving defect recognition accuracy.

[0048] Secondly, the random noise is removed by the noise reduction algorithm. The formula is as follows:

[0049]

[0050] Noise reduction can effectively suppress random interference caused by sensor noise or ambient light changes, reduce the occurrence of false edges and pseudo defects, and thus significantly improve the accuracy and stability of defect detection, ensuring that the static and dynamic performance of the system meets the requirements of industrial automation. At the same time, it removes stray noise in the image, which helps to clearly show subtle defects (such as microcracks and small inclusions). Especially in the edge detection and texture analysis stages, it reduces errors and improves the accuracy of defect positioning, providing more reliable basic data for subsequent quantitative analysis.

[0051] In order to adapt to different defect sizes and details, the image is scaled using the following formula:

[0052]

[0053] Scaling can adjust the spatial scale of the image, making tiny defects more obvious or large defects easier to observe as a whole. This facilitates multi-scale and multi-resolution defect analysis strategies, enabling unified detection and identification of defects of different sizes, thereby improving the overall adaptability of the system. Furthermore, by adjusting the image resolution through scaling, the image data size, computational complexity, and processing time can be reduced while ensuring detection results. This helps to realize a fast-response online monitoring system, which is particularly valuable in large-scale production lines.

[0054] At the same time, the color enhancement technology is achieved by linear transformation formula I c (x,y)=α*I g (x,y)+β, such as increasing contrast and brightness to make the difference between defective areas and normal areas more obvious, thereby improving the sensitivity and accuracy of defect recognition. The comprehensive application of these preprocessing steps makes the processed steel surface image have higher detail expression and lower noise interference, significantly enhancing the performance of subsequent defect detection and recognition modules, thereby achieving more accurate and stable steel surface quality control;

[0055] In the defect detection and recognition module, advanced computer vision technology is used to perform multi-faceted analysis on the pre-processed steel surface image to achieve high-precision defect recognition. First, through edge detection technology such as the Sobel operator, the horizontal and vertical gradient calculations are used to extract the defect edge. The formula is Y(x,y)=|G x (x,y)|+|G y(x,y)|, these technologies can quickly and accurately capture the boundaries of defects such as cracks, fissures, and deformations, helping the system to clearly identify the spatial location and contour information of the defects. Secondly, the system also uses texture analysis methods such as gray-level co-occurrence matrix (GLCM) to extract the texture features of the defect area, including contrast, energy, uniformity, etc. Taking contrast features as an example, the formula is Where P(i, j) represents the probability distribution of pixel pairs in the gray-level co-occurrence matrix. Texture analysis can effectively distinguish the microscopic differences between normal surface texture and defective areas, and identify the detailed features of defects such as fine cracks, inclusions, and looseness. After comprehensive analysis of these edge and texture information, the system can accurately determine the type of defect (such as cracks, holes, inclusions, etc.), location (coordinate information), size (area, length), and quantity (number statistics), and output the detailed detection results to the defect analysis module. This complete set of technical means greatly improves the automation level and accuracy of defect identification, provides a strong technical support for steel quality control, and also reduces the difficulty and error of manual inspection.

[0056] In the defect analysis module, based on the defect feature information obtained from the detection and identification stage, a variety of advanced calculation methods are used to conduct in-depth quantitative analysis and classification of defects on the steel surface to ensure accurate identification and evaluation of different defect types. First, the actual area of ​​the defect is calculated by counting the number of pixels in the detected defect area and combining the image scaling ratio. The formula is: Where M(p,q) represents the defect pixel mask, 1 represents a defect, 0 represents a non-defect, S represents the actual area represented by each pixel, W and H represent the width and height of the defect mask, and p and q represent the position index of the pixel. This step can accurately measure the size of the defect, which helps to formulate repair strategies or process improvements. Secondly, the depth map information obtained by the depth sensor is combined with the depth change of the defect area (such as crack depth or dent depth) to use the model estimation formula The depth of the defect is calculated to determine the severity of the defect and the possible failure risk. Finally, to evaluate the length of the defect, the continuous distance of the pixel points at the boundary of the defect outline is analyzed and the actual length of the defect is calculated by multiplying the number of pixels by the scale factor:

[0057] Lc defect =n pixels *Sc x

[0058] This indicator helps determine the extent of cracks or scratches and guides repair or preventive measures. By comprehensively analyzing the type of defects (such as cracks, holes, inclusions), location (coordinates in the image), size (area, depth, length), and quantity, a scientific assessment of the surface quality of steel is ensured. The use of these quantitative analysis methods based on image processing and geometric calculations significantly improves the objectivity and accuracy of the test results, providing reliable basic data for subsequent quality control, fault prediction, and production optimization, effectively avoiding errors caused by subjective judgment, and ensuring the high quality and stability of steel products.

[0059] In the entire hot-rolled steel quality control system, the statistical reporting module plays a key role as a connecting bridge. It uses data aggregation and visualization technology to organize the detailed defect information (including defect type, location, size and quantity) output by the defect analysis module into a structured statistical report, and uses database storage, chart generation and automatic formatting tools to ensure that the report content is comprehensive, clear and easy to understand. The statistical report is then transmitted to the process adjustment module, which uses automated decision-making algorithms (such as those based on rule bases or machine learning models) to parse defect data, automatically identify potential process anomalies or optimization space, and then adjust key process parameters of steel production, such as temperature, speed, pressure and cooling rate, to improve surface quality and reduce the incidence of defects. The process adjustment module The block adjusts equipment parameters in real time through the control system interface (such as PLC or industrial control platform) and quickly applies the optimization plan to the production line, thereby realizing continuous adaptive process optimization. The entire system forms a highly automated, closed-loop quality control process. From image preprocessing, defect detection, defect analysis, statistical reporting to process adjustment, each link is combined with efficient algorithms and automatic control technology, which not only greatly improves the speed and accuracy of detection and adjustment, but also reduces the risk of human intervention, and realizes intelligent manufacturing in the steel production process. The integration of this complete link not only ensures the stable and excellent surface quality of steel, but also greatly improves production efficiency and economic benefits, providing solid technical support for the digital transformation of the modern steel manufacturing industry.

[0060] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A hot-rolled steel surface defect analysis system, characterized by: It includes data acquisition module, image preprocessing module, defect detection and recognition module, defect analysis module, statistical reporting module and process adjustment module; The data acquisition module obtains the surface image of the newly produced steel material by connecting to a high-definition camera, and transmits the obtained surface image of the newly produced steel material to the image preprocessing module via a wireless network; The image preprocessing module performs image grayscale conversion, image noise reduction, image scaling, and image color enhancement on the newly produced steel surface image, and the processed steel surface image is transmitted to the defect detection and recognition module; The defect detection and recognition module uses computer vision technology to perform edge detection and texture analysis on the processed steel surface image and then identifies steel defects, and outputs the steel defect type, defect location, defect size and defect quantity to the defect analysis module; The defect analysis module further analyzes and classifies the steel defects according to their type, location, size, and number, and calculates the defect area, depth, and length of the steel defects. The statistical reporting module generates a steel defect statistical report based on the detailed analysis results of the defect analysis module and sends it to the process adjustment module; The process adjustment module automatically adjusts steel production process parameters according to the steel defect statistical report.

2. The hot-rolled steel surface defect analysis system according to claim 1, characterized in that: The formula for graying the image is as follows: K(x,y)=0.299*R(x,y)+0.587*G(x,y)+0.114*B(x,y) In the formula, K(x,y) represents the grayscale value at the pixel position (x,y), R(x,y) represents the red component in the color space, G(x,y) represents the green component in the color space, and B(x,y) represents the blue component in the color space.

3. The hot-rolled steel surface defect analysis system according to claim 2, characterized in that: The image denoising formula is as follows: In the formula, I ′ (x, y) represents the pixel value after filtering, I(x+u, y+v) represents the pixel value of the original grayscale image at position (x+u, y+v), N represents the total number of pixels in the filtering window, and u and v represent the window offset parameters.

4. The hot-rolled steel surface defect analysis system according to claim 3, characterized in that: The formula for image scaling is as follows: In the formula, I s (x′, y′) represents the pixel value of the scaled image at the new pixel position, Indicates that the coordinates mapped to the original image x′ are calculated based on the scaling factor s. Indicates that the coordinates mapped to the original image y′ are calculated based on the scaling factor s, round(*) indicates that the coordinates are rounded to the nearest pixel point, and I(*,*) indicates the pixel value of the original image at the corresponding position.

5. The hot-rolled steel surface defect analysis system according to claim 4, characterized in that: The formula for image color enhancement is as follows: I c (x,y)=α*I g (x,y)+β In the formula, I c (x,y) represents the enhanced color value, I g (x,y) represents the original color value, α represents the scaling factor, and β represents the offset.

6. The hot-rolled steel surface defect analysis system according to claim 5, characterized in that: The edge detection formula is as follows: Y(x,y)=|G x (x,y)|+|G y (x,y)| In the formula, Y(x,y) represents the strength value of edge detection, G x (x,y) represents the horizontal gradient of the image, G y (x,y) represents the vertical gradient of the image.

7. The hot-rolled steel surface defect analysis system according to claim 6, characterized in that: The formula for the texture analysis is as follows: In the formula, Cl represents contrast, P(i,j) represents the probability of position (i,j) in the gray-level co-occurrence matrix, and N represents the number of gray levels.

8. The hot-rolled steel surface defect analysis system according to claim 7, characterized in that: The formula for calculating the defect area of ​​steel is as follows: In the formula, A defect Represents the defect area of ​​the steel, M(p,q) represents the defect pixel mask, 1 represents a defect, 0 represents a non-defect, S represents the actual area represented by each pixel, W and H represent the width and height of the defect mask, and p and q represent the position index of the pixel.

9. The hot-rolled steel surface defect analysis system according to claim 8, characterized in that: The formula for calculating the defect depth of steel is as follows: In the formula, D represents the defect depth of the steel, D ref Indicates the reference depth value, L ref represents the reference grayscale, and L represents the average grayscale of the defect area.

10. The hot-rolled steel surface defect analysis system according to claim 9, characterized in that: The formula for calculating the defect length of steel is as follows: Lc defect =n pixels *Sc x In the formula, Lc defect Indicates the defect length of steel, n pixels Indicates the number of pixels in the defect outline, Sc x Indicates the actual length corresponding to each pixel.

Images