A method and related equipment for detecting black spots on galvanized automotive steel sheets

Abstract

This application discloses a method and related equipment for detecting black spots on galvanized automotive steel sheets, relating to the field of metallurgical machinery technology. The method includes: acquiring the specification information of the strip steel; determining the upper and lower surface detection areas of the strip steel based on the specification information and a preset edge offset; acquiring a first surface image and a second surface image; extracting features from the first and second surface images to generate first and second black spot feature data; identifying the first and second black spot feature data based on a deep learning model, and counting the number of first black spots in the upper surface detection area and the number of second black spots in the lower surface detection area; and determining the quality grade of the strip steel based on the number of first and second black spots and a preset black spot threshold. This application achieves automated, full-process detection of black spot defects.

Description

Technical Field

[0001] This application relates to the field of metallurgical machinery technology, and in particular to a method and related equipment for detecting black spots on galvanized automotive steel sheets. Background Technology

[0002] Galvanized automotive steel sheets are a crucial material in automobile manufacturing, and their surface quality directly impacts subsequent processing performance and product appearance. Black spot defects are a common quality issue in the galvanizing process of cold-rolled strip steel, primarily caused by zinc inclusions, residual oxidation, or process fluctuations. These defects lead to uneven coating, and in severe cases, product downgrading or even scrapping. Currently, traditional black spot detection mainly relies on manual visual inspection or offline sampling analysis. This requires cutting samples after production line shutdown and using a microscope to locally count defects. However, manual methods are inefficient, subjective, and difficult to achieve comprehensive coverage in high-intensity continuous production, resulting in high missed detection rates and significant safety hazards. Therefore, a novel method for detecting black spots in galvanized automotive steel sheets is urgently needed to address the aforementioned technical problems. Summary of the Invention

[0003] The summary section introduces a series of simplified concepts, which will be further explained in detail in the detailed description section. This summary section is not intended to limit the key and essential technical features of the claimed technical solutions, nor is it intended to determine the scope of protection of the claimed technical solutions.

[0004] Firstly, this application provides a method for detecting black spots on galvanized automotive steel sheets, including:

[0005] Obtain the specifications of the strip steel;

[0006] Based on specification information and preset edge offset, the upper surface detection area and lower surface detection area of ​​the strip are determined. The upper surface detection area includes a first drive side detection area, a first middle detection area and a first operating side detection area, and the lower surface detection area includes a second drive side detection area, a second middle detection area and a second operating side detection area.

[0007] Acquire a first surface image of the upper surface detection area and a second surface image of the lower surface detection area;

[0008] Feature extraction is performed on the first surface image and the second surface image to generate first black dot feature data corresponding to the first surface image and second black dot feature data corresponding to the second surface image.

[0009] The first black dot feature data and the second black dot feature data are identified based on the deep learning model, and the number of first black dots in the upper surface detection area and the number of second black dots in the lower surface detection area are counted.

[0010] The quality grade of the strip steel is determined based on the number of first black spots, the number of second black spots, and the preset black spot threshold.

[0011] In some implementations, the detection area on the upper surface of the strip is determined based on specification information and a preset edge offset, including:

[0012] Based on the strip length and preset edge offset information, the upper surface is divided into a first drive side region, a first operating side region and a first middle region along the length direction. The preset edge offset includes drive side offset and operating side offset.

[0013] In the first driving side region, a first driving side detection region is selected along the length direction based on a preset fixed side length value. The first driving side detection region includes the center point of the first driving side.

[0014] In the first operating side region, a first operating side detection region is selected along the length direction based on a preset fixed side length value. The first operating side detection region includes a first operating side center point, wherein the first driving side center point and the first operating side center point are located on the same horizontal straight line.

[0015] Based on the first driving side center point, the first operating side center point, and the first central region, determine the first central center point of the first central detection region;

[0016] The first central detection area is determined based on the first central point.

[0017] In some embodiments, based on the strip length and preset edge offset information, the upper surface is divided along the length direction into a first drive side region, a first operating side region, and a first middle region, including:

[0018] Based on the strip length and the drive-side offset, the drive-side start position and drive-side end position of the first drive-side region are determined sequentially along the first direction.

[0019] The first driving side region of the upper surface is determined based on the starting position and ending position of the driving side.

[0020] Based on the strip length and the operating side offset, the starting position and ending position of the operating side of the first operating side region are determined sequentially along the second direction opposite to the first direction.

[0021] The first operating side region of the upper surface is determined based on the starting position and ending position of the operating side.

[0022] Based on the end position of the drive side and the end position of the operation side, a first central region of the upper surface is determined, wherein the starting position of the center of the first central region is the end position of the drive side, and the ending position of the center is the end position of the operation side.

[0023] In some implementations, the first surface image includes a first driving side image, a first operating side image, and a first central image. Acquiring the first surface image of the upper surface detection area includes:

[0024] Based on the center point of the first driving side and the preset fixed side length value, the first driving side image acquisition range of the first driving side detection area is determined;

[0025] Based on the center point of the first operating side and the preset fixed side length value, the first operating side image acquisition range of the first operating side detection area is determined;

[0026] Based on the first central point and the preset fixed side length value, the first central image acquisition range of the first central detection area is determined;

[0027] Based on a preset scanning resolution, the first drive-side image acquisition range, the first operation-side image acquisition range, and the first middle image acquisition range are scanned line by line using a line scan camera to generate the first drive-side image, the first operation-side image, and the first middle image.

[0028] In some implementations, feature extraction is performed on the first surface image to generate first black dot feature data corresponding to the first surface image, including:

[0029] Based on the first driving side image, the first operating side image, and the first central image, adaptive light correction and distortion correction are performed respectively to generate the corresponding first driving side corrected image, first operating side corrected image, and first central image.

[0030] Based on a preset filtering algorithm, noise is removed from the first driving side correction image, the first operating side correction image, and the first middle correction image to generate the first driving side denoised image, the first operating side denoised image, and the first middle denoised image.

[0031] Based on the contrast enhancement algorithm, the black dot region of the first driving side denoised image, the first operating side denoised image and the first middle denoised image are enhanced to generate the first driving side enhanced image, the first operating side enhanced image and the first middle enhanced image.

[0032] Based on preset segmentation rules, the first driving side enhanced image, the first operating side enhanced image and the first middle enhanced image are each segmented into multiple sub-images, wherein the size of each sub-image is smaller than a preset fixed side length value;

[0033] Based on the size threshold, shape features and color differences of black dots, candidate regions of black dots corresponding to the first driving side image, the first operating side image and the first middle image are extracted from multiple sub-images;

[0034] Based on the grayscale distribution and edge gradient features of the candidate black spot region, the first black spot feature data is determined, which includes the black spot location, black spot size, and black spot shape parameters.

[0035] In some implementations, the first black dot feature data and the second black dot feature data are identified based on a deep learning model, and the number of first black dots in the upper surface detection area and the number of second black dots in the lower surface detection area are counted, including:

[0036] Based on the feature data of the first black point and the feature data of the second black point, a deep learning model is used to perform feature matching on the candidate black point region to determine the confidence parameter of the candidate black point.

[0037] Based on the comparison results between the confidence parameter and the preset confidence threshold, valid black points are selected.

[0038] Based on the position and size of the effective black dots, duplicate effective black dots in the upper and lower surface detection areas are removed to determine the target black dot position set.

[0039] Based on the number of black dots in the target black dot location set, determine the number of first black dots in the upper surface detection area and the number of second black dots in the lower surface detection area.

[0040] In some implementations, the preset black spot threshold includes a first preset threshold and a second preset threshold. Based on the number of first black spots, the number of second black spots, and the preset black spot threshold, the quality grade of the strip steel is determined, including:

[0041] Based on the comparison result of the number of first black dots and the first preset threshold, the quality status of the upper surface of the upper surface detection area is determined.

[0042] Based on the comparison result between the number of second black dots and the second preset threshold, the quality status of the lower surface of the lower surface detection area is determined;

[0043] The quality grade of the strip steel is determined based on the logical combination relationship between the quality states of the upper and lower surfaces.

[0044] Secondly, this application proposes a device for detecting black spots on galvanized automotive steel sheets, comprising:

[0045] Specification information acquisition unit, used to acquire the specification information of strip steel;

[0046] The detection area determination unit determines the upper surface detection area and the lower surface detection area of ​​the strip steel based on the specification information and the preset edge offset. The upper surface detection area includes a first drive side detection area, a first middle detection area and a first operation side detection area, and the lower surface detection area includes a second drive side detection area, a second middle detection area and a second operation side detection area.

[0047] A surface image acquisition unit is used to acquire a first surface image of the upper surface detection area and a second surface image of the lower surface detection area;

[0048] The black dot feature generation unit is used to extract features from the first surface image and the second surface image to generate first black dot feature data corresponding to the first surface image and second black dot feature data corresponding to the second surface image.

[0049] The black dot count unit identifies the first black dot feature data and the second black dot feature data based on a deep learning model, and counts the number of first black dots in the upper surface detection area and the number of second black dots in the lower surface detection area.

[0050] The quality grade determination unit determines the quality grade of the strip steel based on the number of first black spots, the number of second black spots, and a preset black spot threshold.

[0051] Thirdly, an electronic device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program stored in the memory to implement the steps of the method for detecting black spots on galvanized automotive sheet as described in any of the first aspects.

[0052] Fourthly, this application proposes a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the method for detecting black spots on galvanized automotive steel sheets according to any one of the first aspects.

[0053] In summary, this application achieves automated, full-process detection of black spot defects by dynamically adjusting the multi-region detection range on the upper and lower surfaces of the strip steel, combined with the efficient image acquisition of a line scan camera and the accurate identification of a deep learning model. Specifically, the detection area is adaptively determined based on the strip steel specification information, avoiding manual intervention; double-sided synchronous image processing and black spot quantity statistics improve detection efficiency and coverage; and the intelligent analysis of black spot feature data using a deep learning model effectively reduces the false detection rate and the missed detection rate, ensuring the objectivity and accuracy of quality judgment results. This application not only improves production efficiency but also reduces the safety risks of manual operation, providing reliable technical support for the quality control of galvanized automotive steel sheets. Attached Figure Description

[0054] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit this specification. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0055] Figure 1This is a schematic flowchart of a method for detecting black spots on galvanized automotive steel sheets, provided in an embodiment of this application.

[0056] Figure 2 This is a schematic diagram illustrating the detection results of black spots on a galvanized automotive sheet provided in an embodiment of this application.

[0057] Figure 3 A schematic diagram of a device for detecting black spots on galvanized automotive steel sheets provided in this application embodiment;

[0058] Figure 4 This is a schematic diagram of an electronic device for detecting black spots on galvanized automotive steel sheets, provided as an embodiment of this application. Detailed Implementation

[0059] The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus. The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them.

[0060] Please see Figure 1 This is a schematic flowchart of a method for detecting black spots on galvanized automotive steel sheets provided in an embodiment of this application, which may specifically include:

[0061] S110. Obtain the specifications of the strip steel;

[0062] For example, specification information refers to the basic parameters recorded during the strip steel production process, including key physical properties such as the strip's width, length, and thickness. This information is typically collected in real time by sensors or control systems on the production line and stored in a database. The main purpose of obtaining specification information is to provide a basis for the dynamic division of subsequent inspection areas, ensuring that the inspection range can adapt to strip steel of different sizes and avoiding blind spots or duplicate coverage caused by specification differences.

[0063] By acquiring specification information and combining it with preset parameters such as edge offset, the initial positions and distribution rules of the detection areas on the upper and lower surfaces can be quickly determined. For example, the strip width directly affects the division range of the detection areas on the drive side, operation side, and middle, while the length parameter is used to dynamically adjust the sampling interval of the detection area along the strip's extension direction. This step provides the necessary data foundation for subsequent image acquisition and black spot recognition, ensuring that the detection process accurately matches the actual production status of the strip.

[0064] S120. Based on specification information and preset edge offset, determine the upper surface detection area and lower surface detection area of ​​the strip. The upper surface detection area includes a first drive side detection area, a first middle detection area and a first operating side detection area. The lower surface detection area includes a second drive side detection area, a second middle detection area and a second operating side detection area.

[0065] For example, the detection areas on the upper and lower surfaces are dynamically divided based on the strip specifications and preset edge offsets. Based on the actual dimensions of the strip, the drive-side offset and the operating-side offset define the start and end positions of the edge detection area. Through their combined action, the drive-side, middle, and operating-side detection areas are sequentially divided along the strip length, ensuring that the detection range covers critical quality risk areas on the strip surface and avoiding detection blind spots caused by width variations.

[0066] Furthermore, the division of the detection area on the lower surface follows the same logic as the upper surface, determining the positions of the detection areas on the drive side, middle, and operation side through a symmetrical mapping method. This dynamic division mechanism can adapt to the production needs of strip steel of different specifications, ensuring the synchronization and consistency of the detection areas on the upper and lower surfaces, and providing a standardized spatial reference for subsequent image acquisition and black spot recognition.

[0067] S130: Obtain the first surface image of the upper surface detection area and the second surface image of the lower surface detection area;

[0068] For example, a line scan camera is used to acquire high-precision images of the preset detection areas on the upper and lower surfaces of the strip. Based on the dynamic movement of the strip, the line scan camera scans line by line along the length of the detection area, generating surface images covering the drive side, middle section, and operating side detection range. Simultaneous acquisition of images from the upper and lower surfaces ensures consistency of black dot data on both sides, avoiding detection deviations caused by strip displacement or process fluctuations.

[0069] The first and second surface images serve as raw data, providing input for subsequent black spot feature extraction and recognition. Preset scanning resolution and dynamic exposure control ensure image clarity and detail capture capabilities, adapting to detection needs under different lighting conditions. The acquired images are directly transmitted to the processing system and correlated with strip steel specifications and detection area coordinates to establish a mapping relationship for black spot location and quantity statistics.

[0070] S140. Extract features from the first surface image and the second surface image to generate first black dot feature data corresponding to the first surface image and second black dot feature data corresponding to the second surface image.

[0071] For example, the feature extraction process first preprocesses the acquired surface image to eliminate environmental interference and enhance the recognizability of black spot regions. Adaptive lighting correction and distortion correction techniques are used to adjust brightness differences and geometric distortions caused by shooting angles or uneven lighting, generating a standardized corrected image. Subsequently, filtering algorithms are employed to remove image noise, and contrast enhancement techniques are used to improve the brightness difference between black spots and the background, ensuring the visibility of minor defects. Based on this, candidate black spot regions are further extracted through image segmentation and feature selection. The enhanced image is segmented into multiple sub-images to increase the relative size proportion of black spots, facilitating subsequent analysis. Potential black spot regions are selected from the sub-images by combining black spot size thresholds, shape features, and color differences, and black spot feature data containing location, size, and morphological parameters are generated based on grayscale distribution and edge gradient features. This process provides structured input for the deep learning model, supporting accurate statistics and quality assessment of the number of black spots.

[0072] S150. Based on a deep learning model, identify the feature data of the first black dot and the feature data of the second black dot, and count the number of the first black dot in the upper surface detection area and the number of the second black dot in the lower surface detection area.

[0073] For example, a pre-trained deep learning model is used to analyze black spot feature data to accurately identify and statistically analyze black spot defects. The model, based on a convolutional neural network architecture, performs feature matching and classification on the input candidate black spot regions, calculating the confidence parameter for each candidate region. Valid black spots are selected by using a pre-set confidence threshold, eliminating false positives caused by noise or interference. Simultaneously, by combining the location and size information of the black spots, duplicate or overlapping black spots in the detection areas on the upper and lower surfaces are removed, generating a unique set of black spot locations to ensure the accuracy of the statistical results. This process achieves robust handling of complex background noise, ensuring reliable detection of minute defects while avoiding subjective bias from human intervention.

[0074] Based on the number of black dots in the target black dot location set, the total number of black dots in the upper and lower surface detection areas is counted separately. The statistical results are directly related to the strip steel quality judgment logic, providing core data support for subsequent quality grade classification. Through the efficient computation and adaptive learning capabilities of the deep learning model, the coverage and judgment accuracy of black dot recognition are improved, and the reliance on manual re-inspection is reduced.

[0075] S160. Determine the quality grade of the strip steel based on the number of first black spots, the number of second black spots, and the preset black spot threshold.

[0076] For example, by comparing a preset black spot threshold with the number of black spots in the detection areas of the upper and lower surfaces, an objective determination of the strip steel quality grade can be achieved. Specifically, based on the comparison result of the first black spot count and the first preset threshold, the quality state of the upper surface (e.g., qualified, unqualified, or critical state) is determined; simultaneously, based on the comparison result of the second black spot count and the second preset threshold, the quality state of the lower surface is determined. Through preset logical combination rules, the quality states of the upper and lower surfaces are combined to finally output the comprehensive quality grade of the strip steel (e.g., Level 1 qualified, Level 2 warning, or Level 3 unqualified).

[0077] This judgment mechanism, through a layered threshold design and state combination strategy, avoids the risk of misjudgment that may result from a single threshold. Simultaneously, by combining the synchronous detection results of the upper and lower surfaces, it ensures that the quality grade reflects the overall defect distribution characteristics of the strip steel, providing a reliable basis for subsequent production decisions.

[0078] In summary, this embodiment of the application achieves fully automated detection of black spot defects on galvanized automotive steel sheets by dynamically adapting to the strip steel specifications and automatically dividing the upper and lower surfaces into multiple detection areas. Combined with the efficient image acquisition of a line scan camera and the analysis of a deep learning model, this achieves fully automated detection of black spot defects. Based on the actual dimensions of the strip steel, this embodiment dynamically adjusts the position and range of the detection areas on the drive side, middle, and operating side to ensure coverage of key quality risk areas and avoid missed or duplicate detections due to width variations. Through dual-sided synchronous image acquisition and preprocessing technology, uneven illumination and geometric distortion interference are eliminated, enhancing the contrast between black spots and the background and providing a high-quality data foundation for subsequent feature extraction. A deep learning model is used to identify and deduplicate black spot candidate areas, effectively distinguishing between real defects and noise interference, reducing false positive and false negative rates. Finally, based on a layered threshold judgment rule, the number of black spots on the upper and lower surfaces is comprehensively evaluated, outputting an objective quality level result. This embodiment of the application improves detection efficiency and accuracy, reduces the safety hazards of manual intervention, and provides real-time quality monitoring capabilities for continuous production lines, meeting the surface defect control requirements of automotive outer panels.

[0079] In some instances, the upper surface inspection area of ​​the strip is determined based on specification information and preset edge offset, including:

[0080] Based on the strip length and preset edge offset information, the upper surface is divided into a first drive side region, a first operating side region, and a first middle region along the length direction. The preset edge offset includes drive side offset and operating side offset, and includes:

[0081] Based on the strip length and the drive-side offset, the drive-side start position and drive-side end position of the first drive-side region are determined sequentially along the first direction.

[0082] The first driving side region of the upper surface is determined based on the starting position and ending position of the driving side.

[0083] Based on the strip length and the operating side offset, the starting position and ending position of the operating side of the first operating side region are determined sequentially along the second direction opposite to the first direction.

[0084] The first operating side region of the upper surface is determined based on the starting position and ending position of the operating side.

[0085] Based on the end position of the drive side and the end position of the operation side, a first central region of the upper surface is determined, wherein the starting position of the center of the first central region is the end position of the drive side, and the ending position of the center is the end position of the operation side.

[0086] For example, based on the strip length recorded in the strip specifications, and combined with a preset drive-side offset (i.e., a fixed distance inward from the drive-side edge of the strip), the drive-side start position and drive-side end position of the first drive-side region are calculated sequentially along a first direction of the strip length (e.g., the direction extending from the head to the tail of the strip). The drive-side offset is a preset fixed value or proportional value used to define the initial offset distance of the drive-side region relative to the edge of the strip. For example, if the strip length is L and the drive-side offset is D1, then the drive-side start position is the position where the strip head starts offset by D1 along the first direction, and the drive-side end position is the termination point after extending the drive-side start position along the first direction by a preset fixed distance (e.g., 70cm). Through the above steps, it is ensured that the coverage area of ​​the drive-side region matches the actual size of the strip, avoiding invalid detection caused by exceeding the actual size.

[0087] Based on the starting and ending positions of the drive side obtained in the above steps, a first drive side region on the upper surface is determined. The first drive side region extends along the length of the strip, and its width is determined by the strip width specified in the strip specifications. Specifically, the first drive side region is a rectangular area extending along the width direction of the strip (i.e., perpendicular to the length direction) from the starting position to the ending position. This region division logic ensures that the drive side range is always located at the edge of the strip, covering the critical risk area prone to black spot defects during the galvanizing process.

[0088] Based on the strip length and a preset operating side offset, the starting and ending positions of the first operating side region are sequentially determined along a second direction opposite to the first direction (e.g., the direction extending from the tail end of the strip to the head end). The operating side offset D2 is a preset value independent of the drive side offset and is used to define the initial offset distance of the operating side region relative to the other side edge of the strip. For example, the starting position is the position where the end point of the strip tail is offset by D2 along the second direction, and the ending position is the termination point after extending the starting position along the second direction by the same preset fixed distance (e.g., 70 cm). The above steps cover the potential defect area on the other side edge of the strip.

[0089] Based on the starting and ending positions of the operating side obtained in the above steps, the first operating side region on the upper surface is determined. The width of the first operating side region is also determined by the strip width specified in the strip specification information, and its extension along the strip length is limited by the starting and ending positions of the operating side. Through the above steps, the logic for dividing the operating side region ensures that the operating side range is always located on the other side of the strip, covering another risk area prone to black spot defects in the galvanizing process.

[0090] Based on the end positions of the drive side and the operation side, a first central region on the upper surface is determined. The starting position of the center of the first central region is the end position of the drive side, and the ending position of the center is the end position of the operation side. It covers the middle area between the drive side and the operation side regions along the length of the strip. Through this step, the drive side, central region, and operation side regions together cover the full width of the upper surface of the strip, providing a spatial reference for the subsequent division of inspection areas and the comprehensive identification of black spot defects.

[0091] In the first driving side region, a first driving side detection region is selected along the length direction based on a preset fixed side length value. The first driving side detection region includes the center point of the first driving side.

[0092] For example, in the first drive-side region, a first drive-side detection area is selected along the length direction of the strip (i.e., the first direction, such as the direction extending from the head to the tail of the strip) based on a preset fixed side length value (e.g., 10 cm). Specifically, the first drive-side detection area is a square region with a side length equal to the preset fixed side length value, and its starting position along the length direction of the strip is determined by the drive-side starting position and a preset drive-side offset. By offsetting from the drive-side starting position along the first direction by a first preset step, the coordinates of the upper left corner vertex of the first drive-side detection area are determined, and a square region is generated by extending along the length and width directions based on the preset fixed side length value. The first drive-side center point of the first drive-side detection area is located at the geometric center of the square region, and its coordinates are determined by calculating the average of the vertex coordinates of the square region. The position of the first drive-side center point is associated with the strip specifications and the preset offset to ensure that the detection area covers the critical quality risk areas on the drive-side edge.

[0093] In the first operating side region, a first operating side detection region is selected along the length direction based on a preset fixed side length value. The first operating side detection region includes a first operating side center point, wherein the first driving side center point and the first operating side center point are located on the same horizontal straight line.

[0094] For example, in the first operating side region, a first operating side detection area is selected along a second direction opposite to the first direction (such as the direction extending from the tail end to the head end of the strip) based on the same preset fixed side length value (e.g., 10 cm). The first operating side detection area is also a square area with a side length equal to the preset fixed side length value, and its starting position is determined by the operating side starting position and the preset operating side offset. By offsetting the same first preset step length from the operating side starting position along the second direction, the coordinates of the upper left corner vertex of the first operating side detection area are determined, and a square area is generated by extending along the length and width directions based on the preset fixed side length value. The first operating side center point of the first operating side detection area is located at the geometric center of the square area, and its coordinates are determined by calculating the average of the vertex coordinates. The first driving side center point and the first operating side center point are ensured to be on the same horizontal straight line through coordinate calibration. The horizontal straight line is parallel to the length direction of the strip, thereby ensuring the symmetry and consistency of the detection areas on the upper and lower surfaces. By combining the preset offset with dynamic calculation, a single square detection area is located, avoiding the efficiency loss of traditional multi-area detection.

[0095] Based on the first driving side center point, the first operating side center point, and the first central region, determine the first central center point of the first central detection region.

[0096] The central axis of the first central region is determined by the line connecting the center point of the first drive side and the center point of the first operation side.

[0097] Based on the central axis and the first central region, the position of the first central point is determined. The first central point is the intersection of the central axis and the symmetrical center line of the first central region, and the symmetrical center line is the vertical center line.

[0098] Based on the first central point, a first central detection area is determined, wherein each detection area is square in shape and has a side length that is a preset fixed side length value;

[0099] For example, the central axis of the first central region is determined based on the line connecting the center point of the first drive side and the center point of the first operating side. The central axis is a straight line connecting the center points of the first drive side and the first operating side, and its direction is parallel to the length direction of the strip. The first central point of the first central detection area is determined by calculating the intersection of the central axis and the symmetrical center line of the first central region (i.e., a straight line perpendicular to the central axis and bisecting the first central region). The first central detection area is a square area with a side length equal to a preset fixed side length value, and its starting position is determined by the coordinates of the first central point. Specifically, a square area is generated by extending half of the preset fixed side length value along both the length and width directions, with the first central point as the center. The position of the first central point is bound to the spatial relationship between the center points of the drive side and the operating side to ensure that the central detection area covers the potential defect area in the middle of the strip and does not overlap with the edge detection area.

[0100] The process of determining the lower surface detection area is completely symmetrical to the division logic of the upper surface detection area, and is based on the same preset parameters and dynamic adjustment mechanism. Specifically, based on the strip specifications and preset edge offset, the lower surface is dynamically divided into a second drive side detection area, a second middle detection area, and a second operation side detection area along the strip length. First, based on the strip length and drive side offset, the drive side start position and drive side end position of the second drive side area are determined along the first direction, generating a rectangular area covering the drive side edge of the lower surface. Simultaneously, based on the strip length and operation side offset, the operation side start position and operation side end position of the second operation side area are determined along the second direction opposite to the first direction, forming a rectangular area on the operation side edge of the lower surface. Subsequently, based on the drive side end position and operation side end position, the middle start position and end position of the second middle area are determined, covering the middle area of ​​the lower surface between the drive side and operation side areas. Based on this, unique square detection areas are selected along the length direction in the second driving side region and the second operating side region, respectively, using preset fixed side length values. The coordinates of the upper left corner vertex and the geometric center point of each detection area are determined using the same offset step size, ensuring that both are on the same horizontal straight line. Furthermore, by connecting the center lines of the second driving side center point and the second operating side center point, and combining the intersection of the symmetrical center line of the lower surface middle region, the center point of the second middle detection area is determined. Using this center, the preset fixed side length values ​​are extended along the length and width directions to generate unique square detection areas. This process, through symmetrical mapping of the coordinates and parameter configuration of the upper surface detection area, ensures strict consistency between the lower surface detection area and the upper surface in terms of spatial distribution, coverage, and detection accuracy, thereby achieving synchronous and accurate identification of defects on both sides.

[0101] In some instances, the first surface image includes a first driving-side image, a first operating-side image, and a first central image. Acquiring the first surface image of the upper surface detection region includes:

[0102] Based on the center point of the first driving side and the preset fixed side length value, the first driving side image acquisition range of the first driving side detection area is determined;

[0103] Based on the center point of the first operating side and the preset fixed side length value, the first operating side image acquisition range of the first operating side detection area is determined;

[0104] Based on the first central point and the preset fixed side length value, the first central image acquisition range of the first central detection area is determined;

[0105] Based on a preset scanning resolution, the first drive-side image acquisition range, the first operation-side image acquisition range, and the first middle image acquisition range are scanned line by line using a line scan camera to generate the first drive-side image, the first operation-side image, and the first middle image.

[0106] For example, when acquiring the first surface image of the upper surface detection area, the image acquisition range of each detection area is first defined based on the preset fixed side length value and the coordinates of the determined first driving side center point, first operating side center point, and first central center point. Specifically, for the first driving side detection area, a square image acquisition range with a side length of 10cm is formed by extending half of the preset fixed side length value along both the length and width directions of the strip, centered on the first driving side center point. The vertex coordinates of this range are calculated using the center point coordinates and offset to ensure complete coverage of the driving side detection area. Similarly, the first operating side image acquisition range of the first operating side detection area extends by the same distance along the second direction and width direction, centered on the first operating side center point, forming a square area with equal side lengths; the first central image acquisition range of the first central detection area is symmetrically expanded along the length and width directions based on the first central center point, generating an independent square coverage area. The above process ensures consistent spatial coverage of each detection area through the standardized configuration of the preset fixed side length value, avoiding black dot recognition errors caused by size differences.

[0107] After determining the image acquisition range of each detection area, a line scan camera performs line-by-line scanning of the first driving side, first operating side, and first central image acquisition range. The line scan camera moves along the length of the strip steel based on a preset scanning resolution (50 pixels per centimeter), simultaneously triggering the image sensor to continuously capture images of the target area at high speed. For each detection area, the line scan camera starts from the initial vertex of the acquisition range and acquires pixel data line-by-line along the length direction at a preset scanning step size (0.02 mm per line) until the entire square area is covered. During the scanning process, the camera's dynamic exposure control module adjusts the illumination parameters in real time to adapt to local changes in the reflectivity of the strip steel surface, ensuring uniform image brightness. In this way, the line scan camera generates the first driving side image, the first operating side image, and the first central image, respectively. The resolution of each image is determined by the preset scanning resolution and the side length of the detection area (500×500 pixels for a 10cm×10cm area).

[0108] The generated surface images are strictly correlated with the coordinate information of the detection area. Specifically, the pixel coordinates of the first driving side image are mapped to the physical location of the first driving side detection area, and the first operating side image and the first central image are also bound to the corresponding detection areas through coordinate transformation matrices. This mapping relationship is dynamically calculated based on strip specifications (such as width and length) and preset offsets to ensure that the position of each pixel in the image accurately reflects the actual defect distribution on the strip surface. In addition, the line scan camera is linked with the strip motion control system through a time synchronization module to acquire images in real time during the continuous movement of the strip, avoiding image blurring or misalignment caused by strip displacement.

[0109] In summary, the first driving side image, the first operating side image, and the first central image are input as raw data into the image processing system, providing a foundation for subsequent feature extraction and black spot recognition. The acquisition range, resolution, and coordinate association information of each image are stored in the detection system's configuration file, supporting adjustments to preset parameters (such as side length and scanning resolution) according to the process requirements of different production lines, thereby achieving flexible adaptation of the detection method. Through the above steps, high-precision image data of the upper surface detection area is efficiently acquired, laying a reliable data foundation for fully automated black spot detection.

[0110] The process of acquiring the second surface image of the lower surface detection area is the same as that of the upper surface detection area. Based on the coordinates of the center point of the second drive side, the center point of the second operation side, and the center point of the second middle part, the image acquisition range of the second drive side, the second operation side, and the second middle part is defined with preset fixed side length values, respectively. The above ranges are scanned line by line by line by a line scan camera at a preset scanning resolution, and the second drive side image, the second operation side image, and the second middle part image are generated synchronously. They are dynamically associated with the physical coordinates of the lower surface detection area to ensure that the image data accurately maps the actual defect distribution of the lower surface of the strip.

[0111] In some instances, feature extraction is performed on the first surface image to generate first black dot feature data corresponding to the first surface image, including:

[0112] Based on the first driving side image, the first operating side image, and the first central image, adaptive light correction and distortion correction are performed respectively to generate the corresponding first driving side corrected image, first operating side corrected image, and first central image.

[0113] Based on a preset filtering algorithm, noise is removed from the first driving side correction image, the first operating side correction image, and the first middle correction image to generate the first driving side denoised image, the first operating side denoised image, and the first middle denoised image.

[0114] Based on the contrast enhancement algorithm, the black dot region of the first driving side denoised image, the first operating side denoised image and the first middle denoised image are enhanced to generate the first driving side enhanced image, the first operating side enhanced image and the first middle enhanced image.

[0115] Based on preset segmentation rules, the first driving side enhanced image, the first operating side enhanced image and the first middle enhanced image are each segmented into multiple sub-images, wherein the size of each sub-image is smaller than a preset fixed side length value;

[0116] Based on the size threshold, shape features and color differences of black dots, candidate regions of black dots corresponding to the first driving side image, the first operating side image and the first middle image are extracted from multiple sub-images;

[0117] Based on the grayscale distribution and edge gradient features of the candidate black spot region, the first black spot feature data is determined, which includes the black spot location, black spot size, and black spot shape parameters.

[0118] For example, adaptive lighting correction and distortion correction are performed based on the first driving-side image, the first operating-side image, and the first central image, respectively. Adaptive lighting correction adjusts the global brightness distribution of the image using a histogram equalization algorithm and employs locally contrast-limited adaptive histogram equalization (CLAHE) to eliminate local illumination unevenness. For distortion correction, based on a pre-calibrated intrinsic parameter matrix of the line-scan camera (including focal length, principal point coordinates, and radial distortion coefficients), the image is geometrically corrected using a perspective transformation model to generate the first driving-side corrected image, the first operating-side corrected image, and the first central image. The corrected images eliminate pixel shifts caused by shooting angle or lens distortion, ensuring the geometric accuracy of subsequent processing.

[0119] Noise suppression is performed on the corrected images using a preset filtering algorithm. Specifically, spatial filtering is applied to the first driving-side corrected image, the first operating-side corrected image, and the first centrally corrected image based on a median filter, with a preset window size of 5×5 pixels. Median filtering effectively removes salt-and-pepper noise and isolated noise points by replacing pixel values ​​with the median value within the neighborhood window, while preserving the edge details of black dots. The filtered images generate the first driving-side denoised image, the first operating-side denoised image, and the first centrally denoised image, providing clean input data for subsequent contrast enhancement.

[0120] A contrast enhancement algorithm is used to enhance the black point regions of the denoised image. Employing the CLAHE technique, the image is divided into multiple 8×8 pixel sub-blocks. Histogram equalization is performed independently on each sub-block, and a preset contrast limit threshold is used to prevent noise amplification. Simultaneously, gamma correction is combined to improve the contrast between dark areas (black points) and the bright background, generating a first driving-side enhanced image, a first operating-side enhanced image, and a first central enhanced image. In the enhanced image, the grayscale values ​​of the black point regions are significantly lower than the background, and the edge gradient features are more pronounced.

[0121] The enhanced image is divided into multiple sub-images according to a preset segmentation rule. The segmentation rule is defined as dividing each enhanced image (e.g., 500×500 pixels) into 25 sub-images (each sub-image is 100×100 pixels), with each sub-image smaller than a preset fixed side length (10cm corresponds to 500 pixels). During segmentation, the image is slid along the row and column directions with a fixed step size, ensuring no overlap. The segmented sub-images are stored independently, and their positions in the original image are recorded using coordinate labels to ensure the integrity of the black dot localization information.

[0122] Based on the size threshold (e.g., diameter ≥ 0.5 mm), shape features (circularity ≥ 0.7), and color difference (grayscale value ≤ 50) of black dots, candidate regions for black dots are extracted from the sub-images. Specifically, a connected component analysis algorithm is used to traverse all sub-images and filter connected regions that meet preset conditions as candidate black dots. For each candidate region, the coordinates of the center of its bounding rectangle are calculated as the black dot position, the length of the diagonal of the bounding rectangle is calculated as the black dot size, and its shape parameters are described using Hu moment features. This step outputs the candidate black dot sets corresponding to the first driving side image, the first operating side image, and the first central image.

[0123] Based on the grayscale distribution and edge gradient features of the candidate black spot regions, the first black spot feature data is generated. For each candidate region, the average grayscale value, grayscale standard deviation, and gradient magnitude histogram of its internal pixels are extracted. Simultaneously, the edge contour is extracted using the Canny edge detection algorithm, and the average gradient direction and gradient intensity of the edge pixels are calculated. Finally, the black spot location (physical coordinates of the strip surface), black spot size (physical length), and black spot shape parameters (circularity, aspect ratio) are integrated to form structured feature data, which serves as the input to the deep learning model. The feature data is strictly correlated with the coordinates of the strip detection area, supporting the accurate statistical analysis of the number of black spots and the quality judgment.

[0124] When extracting features from the second driving side image, second operating side image, and second middle image of the lower surface detection area, the same processing flow as that for the upper surface detection area is adopted. First, adaptive light correction and distortion correction are used to eliminate uneven illumination and geometric distortion, generating the second driving side corrected image, the second operating side corrected image, and the second middle image. Then, noise removal is performed based on a preset filtering algorithm to generate the corresponding denoised image. Next, the contrast of the black spot area is enhanced using the CLAHE algorithm to generate the second driving side enhanced image, the second operating side enhanced image, and the second middle enhanced image. The enhanced image is further divided into multiple sub-images according to a preset segmentation rule. Candidate areas are selected based on the same black spot size threshold, shape features, and color differences. Second black spot feature data containing physical coordinates, size, and morphological features is generated through edge gradient analysis and grayscale distribution calculation, ensuring strict consistency with the upper surface processing logic, parameters, and data format, supporting the synchronous and accurate determination of double-sided defects.

[0125] In some instances, a deep learning model is used to identify the feature data of the first black dot and the feature data of the second black dot, and to count the number of the first black dot in the upper surface detection area and the number of the second black dot in the lower surface detection area, including:

[0126] Based on the feature data of the first black point and the feature data of the second black point, a deep learning model is used to perform feature matching on the candidate black point region to determine the confidence parameter of the candidate black point.

[0127] Based on the comparison results between the confidence parameter and the preset confidence threshold, valid black points are selected.

[0128] Based on the position and size of the effective black dots, duplicate effective black dots in the upper and lower surface detection areas are removed to determine the target black dot position set.

[0129] Based on the number of black dots in the target black dot location set, determine the number of first black dots in the upper surface detection area and the number of second black dots in the lower surface detection area.

[0130] For example, based on the first and second black spot feature data, a pre-trained deep learning model is used to perform feature matching and classification on black spot candidate regions. The deep learning model employs a convolutional neural network (CNN) architecture, with inputs including the position, size, shape parameters, and grayscale gradient features of the candidate regions. The model extracts high-order features through multi-layer convolution and pooling operations, outputting a confidence parameter (range 0 to 1) for each candidate region to be a true black spot. For instance, the confidence parameter is calculated using a Softmax function to normalize and reflect the probability that the candidate region belongs to a black spot defect. During model training, a labeled black spot dataset (labeling the location and category of true black spots) is used, and cross-validation is employed to optimize the weight parameters, ensuring sensitivity to minor defects and robustness to noise interference.

[0131] Based on a preset confidence threshold (e.g., 0.85), the confidence parameters of all candidate regions are filtered. Specifically, if the confidence parameter of a candidate region is greater than or equal to the preset confidence threshold, it is determined to be a valid black spot; if it is less than the preset confidence threshold, it is marked as noise or a false positive and removed. For example, for the candidate black spot set of the upper surface detection region, after threshold comparison, candidate regions with a confidence level greater than or equal to the preset confidence threshold of 0.8 are retained to generate the valid black spot set of the upper surface; the same threshold is used to filter the lower surface detection region to generate the valid black spot set of the lower surface. This step significantly reduces the false detection rate through threshold filtering, ensuring that subsequent statistical results only include high-confidence defects.

[0132] The effective black dots in the first driving side detection area, the first middle detection area, and the first operating side detection area on the upper surface are deduplicated. Based on the physical location and size of the black dots, the Euclidean distance algorithm is used to calculate the distance between adjacent black dots in the same detection area. If the distance is less than a preset deduplication threshold (e.g., 3mm) and the size difference is less than or equal to 10%, they are considered as the same black dot, and the one with higher confidence is retained. After deduplication, the number of black dots in the three sub-regions is counted and accumulated to obtain the first black dot count. The number of black dots in the first driving side detection area, the first middle detection area, and the first operating side detection area are denoted as N1_DS, N1_CT, and N1_OS, respectively. The same operation is performed on the second driving side detection area, the second middle detection area, and the second operating side detection area on the lower surface. The number of black dots in each sub-region is counted and accumulated to obtain the second black dot count. The number of black dots in the second driving side detection area, the second middle detection area, and the second operating side detection area are denoted as N2_DS, N2_CT, and N2_OS, respectively.

[0133] Based on the target black dot location set, the number of black dots in the upper and lower surface detection areas is counted separately. Specifically, each black dot in the target black dot location set is traversed, and the total number of black dots in the corresponding area is accumulated according to its detection area label. For example, if the target set contains 50 black dots, of which 30 are labeled as upper surface and 20 as lower surface, then the output black dot count N1 is 30 and the black dot count N2 is 20. The statistical results are stored in association with the coordinate information of the strip steel detection area and used as input for quality level determination. This step, through precise spatial mapping and logical counting, ensures that the number of black dots is consistent with the actual defect distribution, supporting the reliable execution of subsequent layered threshold determination rules.

[0134] In some instances, the preset black spot threshold includes a first preset threshold and a second preset threshold. Based on the number of first black spots, the number of second black spots, and the preset black spot threshold, the quality grade of the strip steel is determined, including:

[0135] Based on the comparison result of the number of first black dots and the first preset threshold, the quality status of the upper surface of the upper surface detection area is determined.

[0136] Based on the comparison result between the number of second black dots and the second preset threshold, the quality status of the lower surface of the lower surface detection area is determined;

[0137] The quality grade of the strip steel is determined based on the logical combination relationship between the quality states of the upper and lower surfaces.

[0138] For example, in the quality grade determination process, independent preset black spot thresholds are first set for each sub-region of the upper and lower surface detection areas. For the upper surface detection area, preset thresholds are set for the first drive-side detection area (A_DS), the first middle detection area (A_CT), and the first operating-side detection area (A_OS), for example, A_DS = 5, A_CT = 3, and A_OS = 5, indicating that the maximum number of black spots allowed in the drive-side and operating-side sub-regions is 5, and the maximum number of black spots allowed in the middle sub-region is 3, respectively. Similarly, preset thresholds are set for the lower surface detection area (B_DS), the second drive-side detection area (B_CT), and the second operating-side detection area (B_OS), for example, B_DS = 5, B_CT = 3, and B_OS = 5. The thresholds for each sub-region are configured independently according to actual process requirements to adapt to the sensitivity of different areas to defects.

[0139] Based on the statistically obtained number of black dots in each sub-region (upper surface: N1_DS, N1_CT, N1_OS; lower surface: N2_DS, N2_CT, N2_OS), the quality status of each sub-region is determined. For the upper surface detection area, if the number of black dots N1_DS in the first driving side detection area is less than or equal to A_DS, the sub-region is considered qualified; if N1_DS exceeds A_DS, it is marked as unqualified. Similarly, the status of the first middle detection area and the first operating side detection area is determined based on the comparison results of N1_CT and A_CT, and N1_OS and A_OS, respectively. If all sub-regions of the upper surface meet the threshold requirements, the overall quality status of the upper surface is marked as "qualified"; if any sub-region exceeds the limit, the overall status of the upper surface is marked as "unqualified". The same logic is used for the lower surface detection area, and the overall status of the lower surface is determined as "qualified" or "unqualified" by comparing N2_DS, N2_CT, N2_OS with B_DS, B_CT, B_OS.

[0140] Further, based on the combination of quality states of the upper and lower surfaces, the comprehensive quality grade of the strip steel is determined. If the number of black spots in all sub-regions of the upper and lower surfaces does not exceed the limits (i.e., N1_DS≤A_DS, N1_CT≤A_CT, N1_OS≤A_OS, and N2_DS≤B_DS, N2_CT≤B_CT, N2_OS≤B_OS), the strip steel quality grade is determined to be "Level 1 Qualified". If one sub-region on the upper or lower surface exceeds the limit (e.g., N1_DS=6 on the upper surface exceeds A_DS=5, but all other sub-regions and the lower surface are qualified), it is determined to be "Level 2 Warning". If at least one sub-region on each of the upper and lower surfaces exceeds the limit, or two or more sub-regions on the same surface exceed the limit (e.g., N1_DS=6 and N1_CT=4 on the upper surface, or N1_DS=6 on the upper surface and N2_CT=4 on the lower surface), it is determined to be "Level 3 Unqualified".

[0141] The final quality grade result is stored in the database in association with the strip number, inspection time, and the number of black spots in each sub-region, and transmitted to the production control system in real time. For example, the output includes the quality grade (e.g., "Level 2 Warning"), the location of the out-of-limit sub-region (e.g., "Upper surface drive side non-conforming: N1_DS=6 / A_DS=5"), and detailed statistical data. Through independent regional judgment and hierarchical combination rules, this application accurately quantifies the impact of local defects on overall quality, providing high-granularity data support for process optimization and product downgrading decisions, while ensuring the comprehensiveness and traceability of the judgment logic. Please refer to [link / reference]. Figure 2 This is a schematic diagram of the detection results of black spots on a galvanized automotive sheet provided in an embodiment of this application. The detection results of the first drive side detection area, the first middle detection area, the first operating side detection area, the second drive side detection area, the second middle detection area, and the second operating side detection area are all qualified.

[0142] Please see Figure 3 The diagram below illustrates the structure of a device for detecting black spots on galvanized automotive steel sheets, as provided in this embodiment of the application. The device includes:

[0143] Specification information acquisition unit 21 is used to acquire the specification information of strip steel;

[0144] The detection area determination unit 22 determines the upper surface detection area and the lower surface detection area of ​​the strip steel based on the specification information and the preset edge offset. The upper surface detection area includes a first drive side detection area, a first middle detection area and a first operation side detection area, and the lower surface detection area includes a second drive side detection area, a second middle detection area and a second operation side detection area.

[0145] The surface image acquisition unit 23 is used to acquire a first surface image of the upper surface detection area and a second surface image of the lower surface detection area;

[0146] The black dot feature generation unit 24 is used to extract features from the first surface image and the second surface image to generate first black dot feature data corresponding to the first surface image and second black dot feature data corresponding to the second surface image.

[0147] The black dot count unit 25 identifies the first black dot feature data and the second black dot feature data based on a deep learning model, and counts the number of first black dots in the upper surface detection area and the number of second black dots in the lower surface detection area.

[0148] The quality grade determination unit 26 determines the quality grade of the strip steel based on the number of first black spots, the number of second black spots, and a preset black spot threshold.

[0149] Please see Figure 4This application also provides an electronic device 300, including a memory 310, a processor 320, and a computer program 311 stored in the memory 310 and executable on the processor. When the processor 320 executes the computer program 311, it implements the steps of any method for detecting black spots on galvanized automotive steel sheets.

[0150] Since the electronic device described in this embodiment is the device used to implement the detection device for black spots on galvanized automotive steel sheets in this application embodiment, those skilled in the art can understand the specific implementation method and various variations of the electronic device in this embodiment based on the method described in this application embodiment. Therefore, how the electronic device implements the method in this application embodiment will not be described in detail here. Any device used by those skilled in the art to implement the method in this application embodiment is within the scope of protection of this application.

[0151] In practice, when the computer program 311 is executed by the processor, it can implement any of the embodiments corresponding to the first aspect.

[0152] It should be noted that the descriptions of each embodiment in the above embodiments have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0153] Those skilled in the art will understand that embodiments of this application can provide methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-readable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-readable program code.

[0154] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0155] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0156] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0157] This application also provides a computer program product, which includes computer software instructions that, when executed on a processing device, cause the processing device to perform... Figure 1 The flowchart of a method for detecting black spots on galvanized automotive steel sheets in the corresponding embodiment.

[0158] A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another. For example, computer instructions may be transferred from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means. The computer-readable storage medium may be any usable medium that a computer can store or a data storage device such as a server or data center that integrates one or more usable media. The usable medium may be a magnetic medium, an optical medium, or a semiconductor medium, etc.

[0159] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0160] In the several embodiments provided in this application, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0161] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0162] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated units described above can be implemented in the form of hardware and / or software functional units.

[0163] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device to execute all or part of the steps of the methods in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory, magnetic disks, or optical disks.

[0164] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

[0165] Although preferred embodiments have been described in this specification, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this specification.

[0166] Obviously, those skilled in the art can make various modifications and variations to this specification without departing from its spirit and scope. Therefore, if such modifications and variations fall within the scope of the claims and their equivalents, this specification is also intended to include such modifications and variations.

Claims

1. A method for detecting black spots on galvanized automotive steel sheets, characterized in that, include: Obtain the specifications of the strip steel; Based on the specification information and the preset edge offset, the upper surface detection area and the lower surface detection area of ​​the strip are determined. The upper surface detection area includes a first drive side detection area, a first middle detection area and a first operation side detection area. The lower surface detection area includes a second drive side detection area, a second middle detection area and a second operation side detection area. The step of determining the upper surface detection area of ​​the strip steel based on the specification information and the preset edge offset includes: Based on the strip length and preset edge offset of the specification information, the upper surface is divided into a first driving side region, a first operating side region and a first middle region along the length direction, wherein the preset edge offset includes driving side offset and operating side offset. In the first driving side region, the first driving side detection region is selected along the length direction based on a preset fixed side length value, and the first driving side detection region includes the first driving side center point; In the first operation side region, the first operation side detection region is selected along the length direction based on the preset fixed side length value. The first operation side detection region includes a first operation side center point, wherein the first drive side center point and the first operation side center point are located on the same horizontal straight line. Based on the first driving side center point, the first operating side center point, and the first central region, determine the first central center point of the first central detection region; Based on the first central point, the first central detection area is determined; Acquiring a first surface image of the upper surface detection area and a second surface image of the lower surface detection area, wherein the first surface image includes a first driving side image, a first operating side image, and a first central image, and acquiring the first surface image of the upper surface detection area includes: Based on the center point of the first driving side and the preset fixed side length value, the first driving side image acquisition range of the first driving side detection area is determined; Based on the center point of the first operation side and the preset fixed side length value, the first operation side image acquisition range of the first operation side detection area is determined; Based on the first central point and the preset fixed side length value, the first central image acquisition range of the first central detection area is determined; Based on a preset scanning resolution, the first drive-side image acquisition range, the first operation-side image acquisition range, and the first middle image acquisition range are scanned line by line using a line scan camera to generate the first drive-side image, the first operation-side image, and the first middle image. Feature extraction is performed on the first surface image and the second surface image to generate first black dot feature data corresponding to the first surface image and second black dot feature data corresponding to the second surface image; The first black dot feature data and the second black dot feature data are identified based on a deep learning model, and the number of first black dots in the upper surface detection area and the number of second black dots in the lower surface detection area are counted. The quality grade of the strip steel is determined based on the number of the first black spots, the number of the second black spots, and the preset black spot threshold.

2. The method according to claim 1, characterized in that, Based on the strip length and preset edge offset information, the upper surface is divided into a first drive side region, a first operating side region, and a first middle region along the length direction, including: Based on the strip length and the drive-side offset, the drive-side start position and drive-side end position of the first drive-side region are determined sequentially along the first direction. Based on the starting position and ending position of the driving side, the first driving side region of the upper surface is determined; Based on the strip length and the operating side offset, the starting position and ending position of the operating side of the first operating side region are determined sequentially along a second direction opposite to the first direction. Based on the start position and end position of the operation side, the first operation side region of the upper surface is determined; Based on the drive-side end position and the operation-side end position, a first central region of the upper surface is determined, wherein the central starting position of the first central region is the drive-side end position, and the central ending position is the operation-side end position.

3. The method according to claim 1, characterized in that, Feature extraction is performed on the first surface image to generate first black dot feature data corresponding to the first surface image, including: Based on the first driving side image, the first operating side image, and the first central image, adaptive light correction and distortion correction are performed respectively to generate the corresponding first driving side correction image, first operating side correction image, and first central correction image. Based on a preset filtering algorithm, noise is removed from the first driving-side corrected image, the first operating-side corrected image, and the first mid-section corrected image to generate a first driving-side denoised image, a first operating-side denoised image, and a first mid-section denoised image. Based on the contrast enhancement algorithm, the black dot region of the first driving side denoised image, the first operating side denoised image and the first middle denoised image are enhanced to generate the first driving side enhanced image, the first operating side enhanced image and the first middle enhanced image. Based on a preset segmentation rule, the first driving-side enhanced image, the first operating-side enhanced image, and the first central enhanced image are each segmented into multiple sub-images, wherein the size of each sub-image is smaller than the preset fixed side length value; Based on the size threshold, shape features and color differences of black dots, candidate regions of black dots corresponding to the first driving side image, the first operating side image and the first middle image are extracted from the multiple sub-images; Based on the grayscale distribution and edge gradient features of the candidate black spot region, the first black spot feature data is determined, wherein the first black spot feature data includes the black spot position, black spot size, and black spot shape parameters.

4. The method according to claim 3, characterized in that, The step of identifying the first black dot feature data and the second black dot feature data based on a deep learning model, and counting the number of first black dots in the upper surface detection area and the number of second black dots in the lower surface detection area, includes: Based on the first black spot feature data and the second black spot feature data, a deep learning model is used to perform feature matching on the black spot candidate region to determine the confidence parameter of the candidate black spot. Based on the comparison results between the confidence parameters and the preset confidence threshold, valid black spots are selected. Based on the position and size of the effective black dots, duplicate effective black dots in the upper surface detection area and the lower surface detection area are removed to determine the target black dot position set. Based on the number of black dots in the target black dot location set, determine the first number of black dots in the upper surface detection area and the second number of black dots in the lower surface detection area.

5. The method according to claim 1, characterized in that, The preset black spot threshold includes a first preset threshold and a second preset threshold. Determining the quality grade of the strip steel based on the number of the first black spots, the number of the second black spots, and the preset black spot threshold includes: Based on the comparison result between the number of the first black spots and the first preset threshold, the quality status of the upper surface of the upper surface detection area is determined. Based on the comparison result between the number of the second black spots and the second preset threshold, the quality status of the lower surface of the lower surface detection area is determined; The quality grade of the strip steel is determined based on the logical combination relationship between the quality states of the upper surface and the lower surface.

6. A device for detecting black spots on galvanized automotive steel sheets, characterized in that, include: Specification information acquisition unit, used to acquire the specification information of strip steel; The detection area determination unit determines the upper surface detection area and the lower surface detection area of ​​the strip steel based on the specification information and the preset edge offset. The upper surface detection area includes a first drive side detection area, a first middle detection area and a first operation side detection area, and the lower surface detection area includes a second drive side detection area, a second middle detection area and a second operation side detection area. The step of determining the upper surface detection area of ​​the strip steel based on the specification information and the preset edge offset includes: Based on the strip length and preset edge offset of the specification information, the upper surface is divided into a first driving side region, a first operating side region and a first middle region along the length direction, wherein the preset edge offset includes driving side offset and operating side offset. In the first driving side region, the first driving side detection region is selected along the length direction based on a preset fixed side length value, and the first driving side detection region includes the first driving side center point; In the first operation side region, the first operation side detection region is selected along the length direction based on the preset fixed side length value. The first operation side detection region includes a first operation side center point, wherein the first drive side center point and the first operation side center point are located on the same horizontal straight line. Based on the first driving side center point, the first operating side center point, and the first central region, determine the first central center point of the first central detection region; Based on the first central point, the first central detection area is determined; A surface image acquisition unit is configured to acquire a first surface image of the upper surface detection area and a second surface image of the lower surface detection area. The first surface image includes a first driving side image, a first operating side image, and a first central image. Acquiring the first surface image of the upper surface detection area includes: Based on the center point of the first driving side and the preset fixed side length value, the first driving side image acquisition range of the first driving side detection area is determined; Based on the center point of the first operation side and the preset fixed side length value, the first operation side image acquisition range of the first operation side detection area is determined; Based on the first central point and the preset fixed side length value, the first central image acquisition range of the first central detection area is determined; Based on a preset scanning resolution, the first drive-side image acquisition range, the first operation-side image acquisition range, and the first middle image acquisition range are scanned line by line using a line scan camera to generate the first drive-side image, the first operation-side image, and the first middle image. The black spot feature generation unit is used to extract features from the first surface image and the second surface image to generate first black spot feature data corresponding to the first surface image and second black spot feature data corresponding to the second surface image. The black dot count unit identifies the first black dot feature data and the second black dot feature data based on a deep learning model, and counts the number of first black dots in the upper surface detection area and the number of second black dots in the lower surface detection area. The quality grade determination unit determines the quality grade of the strip steel based on the number of the first black spots, the number of the second black spots, and a preset black spot threshold.

7. An electronic device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program stored in the memory to implement the steps of the method for detecting black spots on galvanized automotive steel sheets as described in any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method for detecting black spots on galvanized automotive steel sheets as described in any one of claims 1 to 5.

Images