Intelligent defect detection method and system for stamping die

Abstract

The invention discloses an intelligent defect detection method and system for a stamping die, and the method comprises the steps: obtaining a stamping excess material image through preset azimuth parameters after a stamping process is completed, and the azimuth parameters comprise a collection angle and a space coordinate; dividing the stamping excess material image into a plurality of sub-region excess material images according to coordinate information of the positioning reference hole in the stamping excess material image; inputting each sub-regional excess material image into a perspective correction model, and outputting a sub-regional excess material image with a normalized size; and extracting local stamping contour features by using a contour analysis algorithm, and outputting a stamping deformation early warning signal by comparing geometric deviation values of the local stamping contour features and the standard stamping contour template and combining with a preset tolerance threshold interval. Real-time monitoring of the size and the burr problem of the stamping part is achieved through the plate excess materials, the detection precision and the production efficiency are remarkably improved, and manual intervention and the rejection rate are reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of machine vision detection, and in particular to an intelligent defect detection method and system for a stamping die. Background Art

[0002] In modern manufacturing, small, continuous stamping dies are widely used in the electronics, automotive, and home appliance industries to produce a variety of tiny parts. However, due to the small size of finished stamped parts and their disorganized position after being removed from the sheet metal, traditional inspection methods struggle to efficiently and accurately inspect their quality. Specifically, finished stamped parts often scatter on conveyor belts after being removed from the sheet metal, leaving them in an unstable position. This makes manual inspection inefficient and prone to errors. Furthermore, subtle burrs and dimensional deviations on the surface of small parts are often difficult to detect with the naked eye or simple tools, complicating quality control.

[0003] Traditional quality inspection of stamped parts relies primarily on manual visual inspection or the use of contact measuring tools (such as calipers and micrometers). These methods are not only time-consuming and labor-intensive, but also susceptible to operator subjective factors, leading to inconsistent inspection results. With increasing production speeds, manual inspection can no longer meet the real-time and accuracy requirements of large-scale production. Furthermore, contact measuring tools can cause secondary damage to the product during the inspection process, further affecting product quality. Summary of the Invention

[0004] In order to solve at least one of the technical problems mentioned above, the present invention provides an intelligent defect detection method and system for a stamping die.

[0005] In a first aspect, the present invention provides an intelligent defect detection method for a stamping die, the method comprising:

[0006] After the stamping process is completed, the stamping residual material image is obtained through the preset orientation parameters, and the orientation parameters include the acquisition angle and the spatial coordinates;

[0007] According to the coordinate information of the positioning reference hole in the stamping residual material image, the stamping residual material image is divided into a plurality of sub-region residual material images;

[0008] Input the residual material images of each sub-region into the perspective correction model, and output the residual material images of each sub-region with normalized size;

[0009] The contour analysis algorithm is used to extract local stamping contour features. By comparing the geometric deviation of the local stamping contour features with the standard stamping contour template and combining the preset tolerance threshold range, a stamping deformation warning signal is output.

[0010] Preferably, the perspective correction model is expressed as:

[0011] ,

[0012] ,

[0013] Where, and Respectively represent the width and length of the adjusted area residual image, and Respectively represent the width and length of the residual image of the divided area before adjustment, Indicates the vertical height between the industrial camera and the stamping plate. Indicates the horizontal distance from the center point of the corresponding area residual material image to the center point of the industrial camera. Indicates the tilt angle of the industrial camera shooting downwards. Indicates the azimuth angle of the center point of the corresponding sub-area residual material image relative to the industrial camera's visual axis. The default is 0.

[0014] Preferably, after dividing the stamping waste material image into a plurality of sub-region waste material images, the method further comprises:

[0015] Select the largest area of ​​the sub-region residual material image as the burr analysis object;

[0016] The local stamping contour features corresponding to the burr analysis object are imported into the texture recognition model to obtain the burr quantitative index, and stamping burr abnormality warning information is generated based on the index.

[0017] Preferably, the texture recognition model is expressed as:

[0018] ,

[0019] ,

[0020] ,

[0021] ,

[0022] Where, and They are the gradient components along the X-axis and Y-axis directions, which are calculated by the Sobel operator. Represents the gradient magnitude of each pixel in the image, Represents the gradient direction of each pixel in the image, represents the gradient change index, Represents a local window used to traverse the entire image, Indicates the total number of pixels in the window, represents the gradient magnitude of each pixel in the window, Represents the average value of the gradient amplitude of all pixels in the window, represents the glitch quantification index, is a measure of gradient direction consistency, expressed as standard deviation, The area of ​​the detected burr region is extracted from the binary gradient image through morphological operations. 、 and Represents the weight coefficient.

[0023] In a second aspect, the present invention further provides an intelligent defect detection system for a stamping die, the system comprising:

[0024] An image acquisition module is used to acquire an image of the stamping residue using preset orientation parameters after the stamping process is completed. The orientation parameters include an acquisition angle and spatial coordinates.

[0025] An image partitioning module is used to divide the stamping waste image into a plurality of sub-region waste images according to the coordinate information of the positioning reference hole in the stamping waste image;

[0026] An image correction module is used to input the residual material images of each sub-region into the perspective correction model and output the residual material images of each sub-region with normalized sizes;

[0027] The defect detection module is used to extract local stamping contour features using a contour analysis algorithm, and output a stamping deformation warning signal by comparing the geometric deviation of the local stamping contour features with the standard stamping contour template and combining it with a preset tolerance threshold range.

[0028] Preferably, the perspective correction model is expressed as:

[0029] ,

[0030] ,

[0031] Where, and Respectively represent the width and length of the adjusted area residual image, and Respectively represent the width and length of the residual image of the divided area before adjustment, Indicates the vertical height between the industrial camera and the stamping plate. Indicates the horizontal distance from the center point of the corresponding area residual material image to the center point of the industrial camera. Indicates the tilt angle of the industrial camera shooting downwards. Indicates the azimuth angle of the center point of the corresponding sub-area residual material image relative to the industrial camera's visual axis. The default is 0.

[0032] Preferably, the image partitioning module is further used for:

[0033] Select the largest area of ​​the sub-region residual material image as the burr analysis object;

[0034] The local stamping contour features corresponding to the burr analysis object are imported into the texture recognition model to obtain the burr quantitative index, and stamping burr abnormality warning information is generated based on the index.

[0035] In a third aspect, the present invention also provides an electronic device comprising a processor and a memory, wherein the memory is used to store computer program code, and the computer program code comprises computer instructions. When the processor executes the computer instructions, the electronic device executes the method as described in the first aspect above and any possible implementation thereof.

[0036] In a fourth aspect, the present invention also provides a computer-readable storage medium, in which a computer program is stored. The computer program includes program instructions, and when the program instructions are executed by a processor of an electronic device, the processor executes the method as described in the first aspect above and any possible implementation method thereof.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] The present invention uses an industrial camera to capture high-resolution images of scraps, segments and marks potential burr areas based on positioning holes, ensures dimensional consistency and analyzes stamping marks through perspective adjustment and edge detection, and further applies a burr detection algorithm to accurately identify and quantify burrs and generate early warning information. Compared with directly detecting finished stamping parts, this method utilizes the characteristics that sheet material scraps are usually continuous and connected. These scraps have more stable and predictable structural characteristics, making image recognition and processing simpler and more efficient. It not only significantly improves the accuracy and speed of detection and reduces the need for manual intervention, but also improves product quality and production efficiency through automated feedback control. In addition, the indirect monitoring of scrap can continuously provide immediate quality feedback without affecting the normal production process, helping manufacturers to promptly discover and solve potential problems, and ultimately reduce scrap rates and maintenance costs.

[0039] It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the background technology, the drawings required for use in the embodiments of the present invention or the background technology will be described below.

[0041] The accompanying drawings herein are incorporated into and constitute a part of the specification. These drawings illustrate embodiments consistent with the present disclosure and, together with the specification, are used to explain the technical solutions of the present disclosure.

[0042] Figure 1 A schematic flow chart of an intelligent defect detection method for a stamping die provided by an embodiment of the present invention;

[0043] Figure 2 A schematic structural diagram of an intelligent defect detection system for a stamping die provided by an embodiment of the present invention;

[0044] Figure 3 This is a stamping waste image provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0045] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0046] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present invention. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute a separate or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0047] The existing quality inspection methods for small continuous stamping dies have problems such as low inspection efficiency, poor accuracy, and susceptibility to human factors. It is difficult to efficiently and accurately detect the size and burrs of tiny finished parts.

[0048] refer to Figure 3 To address the issues of small, disorganized finished stamping parts and their disorganized placement after being removed from the sheet metal during small continuous stamping die operations, a machine vision-based stamping die quality inspection method has been proposed. This method not only addresses the difficulty of traditional inspection methods in efficiently and accurately inspecting small parts, but also, by innovatively utilizing sheet metal remnants (the remaining portion of the sheet metal cut from the original sheet metal surrounding the finished stamping part after the stamping operation) as an indirect monitoring target, enables rapid assessment of finished product dimensions and burr issues, thereby improving production efficiency and product quality.

[0049] See also Figure 1 , Figure 1The following is a flow chart of an intelligent defect detection method for a stamping die provided by an embodiment of the present invention. Figure 1 As shown, the method includes:

[0050] S100, after the stamping process is completed, obtaining a stamping residual material image using preset orientation parameters, wherein the orientation parameters include an acquisition angle and spatial coordinates;

[0051] After each stamping process, an industrial camera captures images of the remaining stamped sheet material using preset capture parameters. These capture parameters, optimized through multiple experiments tailored to the specific geometry of the stamping die and the sheet material properties, encompass both the capture angle (the relative angle between the camera and the sheet material) and the camera coordinates (the camera's position in three-dimensional space). To ensure that the captured images accurately reflect the remaining material, the industrial camera is equipped with a high-resolution sensor and an appropriate optical lens. Furthermore, a uniform and sufficiently intense light source is used to minimize the impact of shadows and reflections on image quality. Furthermore, the camera's position and orientation must remain constant throughout the production process to ensure consistent and comparable image data.

[0052] S200, dividing the stamping waste image into a plurality of sub-region waste images according to the coordinate information of the positioning reference hole in the stamping waste image;

[0053] To segment the image based on the location of the locating holes in the stamped sheet metal scrap image and mark the largest sub-residue image as the burr detection image, precise identification of the locating holes is required. These holes are pre-designed and punched into the sheet metal, with a clear geometric shape (usually circular) and relatively fixed size, facilitating automatic identification. Denoising and smoothing the original captured image can be performed using methods such as mean filtering, Gaussian filtering, or bilateral filtering to reduce interference from background noise and irregular textures on subsequent analysis. The Canny edge detection operator or similar edge enhancement algorithms are applied to highlight image boundaries and make the locating hole outlines clearer. Morphological operations such as dilation and erosion are used to optimize edge detection results, eliminate small pseudo-contours, and preserve the true shape of the locating holes. Using a known standard template for locating holes, combined with normalized cross-correlation or other similarity metrics, the image is searched for the region that best matches the template, enhancing recognition robustness under varying lighting conditions.

[0054] After successfully identifying the locations of all the locating holes, the entire scrap image is segmented into multiple independent sub-residue images based on this location information. A connected component analysis algorithm is used to determine the sub-residue region to which each locating hole belongs, based on the distance and relative position between the locating holes. This algorithm groups image pixels into connected regions, each corresponding to a potential sub-residue image. For each connected region, its minimum bounding rectangle is calculated. This rectangle defines the sub-residue image boundary, simplifying subsequent perspective adjustments and other processing tasks.

[0055] Preferably, after dividing the stamping waste material image into a plurality of sub-region waste material images, the method further comprises:

[0056] Select the largest area of ​​the sub-region residual material image as the burr analysis object;

[0057] The local stamping contour features corresponding to the burr analysis object are imported into the texture recognition model to obtain the burr quantitative index, and stamping burr abnormality warning information is generated based on the index.

[0058] After segmentation, the largest image is identified from the multiple sub-residue images and marked as the image for burr detection. Because larger sub-residue images often contain more edge information, particularly in areas prone to burrs, the area of ​​each sub-residue image corresponding to the MBR can be calculated using a simple method of multiplying width by height, or by considering more complex shape factors (such as the ratio of perimeter to area) to reflect actual shape characteristics. All sub-residue images are sorted by area, and the image with the largest area is selected as the target for burr detection. A minimum area threshold is also set to exclude sub-residue images that are too small and unlikely to contain significant burrs, improving detection efficiency.

[0059] S300, inputting the residual material images of each sub-region into the perspective correction model, and outputting the residual material images of each sub-region with normalized sizes;

[0060] Due to factors such as the camera's shooting angle, sheet metal deformation, or mold movement, the captured images of the sub-waste may exhibit perspective distortion, affecting the accuracy of subsequent dimensional measurements. Therefore, perspective adjustment must be performed on each sub-waste image to ensure consistent viewing angles and dimensional proportions across all images. In this step, Harris corner detection is used to extract stable feature points in each sub-residue image. The feature points are usually located at significant structures of the image, such as edge intersections, hole edges, etc., and have good repeatability under different viewing angles. By calculating the descriptors of the feature points (such as SIFT descriptors or BRIEF descriptors), the FLANN nearest neighbor algorithm is used to match the corresponding feature points on the same template image or standard reference image. The matching process combines the spatial position of the feature points and the local texture information to improve the matching accuracy. Based on the successfully matched feature point pairs, the homography matrix between the two images is estimated. This matrix describes the projective transformation relationship between planes and can correct the geometric distortion caused by perspective changes. The estimated homography matrix is ​​used to perform an inverse perspective transformation on the original sub-residue image, converting the image into an ideal front view so that the image elements have consistent proportions and directions. Furthermore, the quality of the transformed image is guaranteed by methods such as bilinear interpolation or cubic spline interpolation.

[0061] S400 uses the contour analysis algorithm to extract local stamping contour features, and outputs a stamping deformation warning signal by comparing the geometric deviation of the local stamping contour features with the standard stamping contour template and combining it with the preset tolerance threshold range.

[0062] After the perspective adjustment is completed, the sizes of all sub-residue images should be consistent to facilitate subsequent dimensional analysis. A strictly calibrated standard sub-residue image is selected as the dimensional benchmark. The standard image can be a high-precision template or an average image derived from a large amount of normal production data. For each adjusted sub-residue image, the actual size of key features (such as positioning hole diameter, side length, specific marker spacing, etc.) is measured and compared with the corresponding size of the standard image. If there is a significant difference, the system will issue a warning and record the data for subsequent analysis and calibration. Considering the accumulation of small deviations that may occur during long-term operation, the system regularly conducts a global assessment of the dimensional consistency of the entire production line. By collecting multiple batches of data, a statistical model is established to predict future dimensional change trends and take timely preventive measures, such as adjusting camera parameters or recalibrating the mold.

[0063] The Canny edge detection algorithm is applied to highlight the boundary information of the sub-waste image and capture the contour of the stamping mark, especially the area that may indicate dimensional deviation or burr formation; based on the edge detection results, the findContours function in OpenCV is used to obtain all closed contour curves. These curves represent the boundaries formed by material cutting or extrusion during the stamping process, and their shape and size directly reflect the stamping quality; feature extraction is performed on each contour curve, including geometric features such as length, width, curvature radius, and endpoint position, as well as visual features such as grayscale histogram and texture features, to fully reflect the stamping mark information and facilitate subsequent classification and judgment; the extracted features are compared with the preset standard stamping mark features, and the standard features can be used to extract the features of the stamping mark. Based on historical data learning or experimental determination, if the difference between the actual feature and the standard feature exceeds the preset tolerance range, it is considered that there is a problem with the stamping mark, which may be inaccurate size or burrs; for each feature, the difference between the actual value and the standard value is calculated, the absolute error is calculated for linear features, and the relative error or standardized error is used for nonlinear features. These differences form the basis of dimensional warning information; based on the calculated dimensional difference, warning information is generated, and the warning information includes but is not limited to the specific dimensional deviation value. In a possible embodiment, corrective measures can also be suggested, such as adjusting the stamping pressure, modifying the mold design, or optimizing the production process parameters. At the same time, possible future problems can be predicted based on historical data and current conditions, and preparations can be made in advance.

[0064] In this embodiment, high-resolution images of residual materials are collected by an industrial camera, and potential burr areas are segmented and marked based on positioning holes. Perspective adjustment and edge detection are used to ensure dimensional consistency and analyze stamping marks. Furthermore, a burr detection algorithm is applied to accurately identify and quantify burrs and generate early warning information. Compared with direct detection of finished stamping parts, this method utilizes the characteristics that sheet material residual materials are usually continuous and connected. These residual materials have more stable and predictable structural characteristics, making image recognition and processing simpler and more efficient. It not only significantly improves the accuracy and speed of detection and reduces the need for manual intervention, but also improves product quality and production efficiency through automated feedback control. In addition, the indirect monitoring of residual materials can continuously provide immediate quality feedback without affecting the normal production process, helping manufacturers to promptly discover and solve potential problems, and ultimately reduce scrap rates and maintenance costs.

[0065] Preferably, the perspective correction model is expressed as:

[0066] ,

[0067] ,

[0068] Where, and Respectively represent the width and length of the adjusted area residual image, and Respectively represent the width and length of the residual image of the divided area before adjustment, Indicates the vertical height between the industrial camera and the stamping plate. Indicates the horizontal distance from the center point of the corresponding area residual material image to the center point of the industrial camera. Indicates the tilt angle of the industrial camera shooting downwards. Indicates the azimuth angle of the center point of the corresponding sub-area residual material image relative to the industrial camera's visual axis. The default is 0.

[0069] To determine the actual size of multiple identical objects on the ground based on the image being detected, a preset angle, and a preset height, a perspective adjustment algorithm based on perspective projection was constructed. This algorithm takes into account the camera's position (height), the shooting angle, and the distance between the object and the camera. The properties of similar triangles are used to establish a proportional relationship. For any object, its actual size is proportional to its apparent size in the image, and this proportionality factor depends on the distance D from the object to the camera and the camera's shooting angle θ.

[0070] Determine the distance D from each object to the camera. The distance D is calculated using the known image position and the camera's intrinsic parameter matrix (including information such as focal length). For width have: ,here, is the width of the object in the image (in pixels), H is the height of the camera, D is the horizontal distance from the object to the camera, and θ is the tilt angle of the camera. For the length L, , is the length of the object in the image (in pixels), is the azimuth of the object relative to the camera's line of sight, which affects the object's apparent length.

[0071] In this embodiment, a perspective adjustment algorithm is used to correct image deformation caused by shooting angle, distance and camera tilt, ensuring the consistency of size and viewing angle of all sub-residue images, thereby improving the accuracy of detection.

[0072] Preferably, the texture recognition model is expressed as:

[0073] ,

[0074] ,

[0075] ,

[0076] ,

[0077] Where, and They are the gradient components along the X-axis and Y-axis directions, which are calculated by the Sobel operator. Represents the gradient magnitude of each pixel in the image, Represents the gradient direction of each pixel in the image, represents the gradient change index, Represents a local window used to traverse the entire image, Indicates the total number of pixels in the window, represents the gradient magnitude of each pixel in the window, Represents the average value of the gradient amplitude of all pixels in the window, represents the glitch quantification index, is a measure of gradient direction consistency, expressed as standard deviation, The area of ​​the detected burr region is extracted from the binary gradient image through morphological operations. 、 and Represents the weight coefficient.

[0078] In this embodiment, first, the original stamping mark image is preprocessed, including but not limited to grayscale conversion, noise removal (such as Gaussian filtering), contrast enhancement and other operations to ensure the accuracy of subsequent gradient calculation.

[0079] Use Sobel, Prewitt, or Canny edge detection operators to calculate the gradient in the image. These operators can estimate the gradient magnitude and direction for each pixel in the image. Let I(x, y) be the image grayscale value at position (x, y), then the gradient magnitude G(x, y) and direction θ(x, y) can be calculated by the following formula 、 To determine the presence and status of glitches, we focus on the gradient changes. Glitches usually appear as a sudden change in the local gradient, that is, the gradient amplitude increases significantly in a small area. Define a local window W(x, y) of size m×n to traverse the entire image. For each window, we can calculate the gradient change index In order to quantify the state of the burr, we can introduce a comprehensive evaluation index , which combines the gradient change index GI and gradient direction consistency and burr area, as shown in the formula .

[0080] In this embodiment, a burr detection algorithm is used to perform burr detection. Through numerical processing, multiple factors such as gradient amplitude, gradient direction, gradient change index and burr area are comprehensively considered, which significantly improves the accuracy of detection. Automated processing reduces manual intervention, enhances the adaptability and flexibility of the system, and can cope with different production environments and conditions. It can immediately discover and correct quality problems on high-speed production lines, reduce scrap rates, and improve product quality and production efficiency.

[0081] In a second aspect, the present invention further provides an intelligent defect detection system for a stamping die, the system comprising:

[0082] An image acquisition module is used to acquire an image of the stamping residue using preset orientation parameters after the stamping process is completed. The orientation parameters include an acquisition angle and spatial coordinates.

[0083] An image partitioning module is used to divide the stamping waste image into a plurality of sub-region waste images according to the coordinate information of the positioning reference hole in the stamping waste image;

[0084] An image correction module is used to input the residual material images of each sub-region into the perspective correction model and output the residual material images of each sub-region with normalized sizes;

[0085] The defect detection module is used to extract local stamping contour features using a contour analysis algorithm, and output a stamping deformation warning signal by comparing the geometric deviation of the local stamping contour features with the standard stamping contour template and combining it with a preset tolerance threshold range.

[0086] Preferably, the perspective correction model is expressed as:

[0087] ,

[0088] ,

[0089] Where, and Respectively represent the width and length of the adjusted area residual image, and Respectively represent the width and length of the residual image of the divided area before adjustment, Indicates the vertical height between the industrial camera and the stamping plate. Indicates the horizontal distance from the center point of the corresponding area residual material image to the center point of the industrial camera. Indicates the tilt angle of the industrial camera shooting downwards. Indicates the azimuth angle of the center point of the corresponding sub-area residual material image relative to the industrial camera's visual axis. The default is 0.

[0090] Preferably, the image partitioning module is further used for:

[0091] Select the largest area of ​​the sub-region residual material image as the burr analysis object;

[0092] The local stamping contour features corresponding to the burr analysis object are imported into the texture recognition model to obtain the burr quantitative index, and stamping burr abnormality warning information is generated based on the index.

[0093] In summary, the method provided in this embodiment can at least achieve the following effects:

[0094] The present invention uses an industrial camera to capture high-resolution images of scraps, segments and marks potential burr areas based on positioning holes, ensures dimensional consistency and analyzes stamping marks through perspective adjustment and edge detection, and further applies a burr detection algorithm to accurately identify and quantify burrs and generate early warning information. Compared with directly detecting finished stamping parts, this method utilizes the characteristics that sheet material scraps are usually continuous and connected. These scraps have more stable and predictable structural characteristics, making image recognition and processing simpler and more efficient. It not only significantly improves the accuracy and speed of detection and reduces the need for manual intervention, but also improves product quality and production efficiency through automated feedback control. In addition, the indirect monitoring of scrap can continuously provide immediate quality feedback without affecting the normal production process, helping manufacturers to promptly discover and solve potential problems, and ultimately reduce scrap rates and maintenance costs.

[0095] See also Figure 2 In one embodiment, an intelligent defect detection system for a stamping die is provided, the system comprising:

[0096] The image acquisition module 100 is used to acquire an image of the stamping residue using preset orientation parameters after the stamping process is completed. The orientation parameters include an acquisition angle and spatial coordinates.

[0097] An image partitioning module 200 is used to divide the stamping waste image into a plurality of sub-region waste images according to the coordinate information of the positioning reference hole in the stamping waste image;

[0098] An image correction module 300 is used to input the residual material images of each sub-region into a perspective correction model and output the residual material images of each sub-region with normalized sizes;

[0099] The defect detection module 400 is used to extract local stamping contour features using a contour analysis algorithm, and output a stamping deformation warning signal by comparing the geometric deviation of the local stamping contour features with the standard stamping contour template and combining it with a preset tolerance threshold range.

[0100] Preferably, the perspective correction model is expressed as:

[0101] ,

[0102] ,

[0103] Where, and Respectively represent the width and length of the adjusted area residual image, and Respectively represent the width and length of the residual image of the divided area before adjustment, Indicates the vertical height between the industrial camera and the stamping plate. Indicates the horizontal distance from the center point of the corresponding area residual material image to the center point of the industrial camera. Indicates the tilt angle of the industrial camera shooting downwards. Indicates the azimuth angle of the center point of the corresponding sub-area residual material image relative to the industrial camera's visual axis. The default is 0.

[0104] Preferably, the image partitioning module is further used for:

[0105] Select the largest area of ​​the sub-region residual material image as the burr analysis object;

[0106] The local stamping contour features corresponding to the burr analysis object are imported into the texture recognition model to obtain the burr quantitative index, and stamping burr abnormality warning information is generated based on the index.

[0107] It can be understood that the functions or modules included in the system provided in this embodiment can be used to execute the method described in the above method embodiment. Its specific implementation can refer to the description of the above method embodiment. For the sake of brevity, it will not be repeated here.

[0108] The present invention also provides an electronic device, including a processor and a memory, wherein the memory is used to store computer program code, and the computer program code includes computer instructions. When the processor executes the computer instructions, the electronic device executes a method as described in any one of the possible implementation modes.

[0109] The present invention also provides a computer-readable storage medium, in which a computer program is stored. The computer program includes program instructions. When the program instructions are executed by a processor of an electronic device, the processor executes a method as described in any one of the possible implementation methods described above.

[0110] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0111] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here. Those skilled in the art will also clearly understand that the descriptions of the various embodiments of the present invention have different focuses. For the convenience and brevity of description, the same or similar parts may not be repeated in different embodiments. Therefore, for parts not described or not described in detail in a certain embodiment, reference can be made to the descriptions of other embodiments.

Claims

1. An intelligent defect detection method for stamping dies, characterized in that: The method comprises: After the stamping process is completed, the stamping residual material image is obtained through the preset orientation parameters, and the orientation parameters include the acquisition angle and the spatial coordinates; According to the coordinate information of the positioning reference hole in the stamping residual material image, the stamping residual material image is divided into a plurality of sub-region residual material images; Input the residual material images of each sub-region into the perspective correction model, and output the residual material images of each sub-region with normalized size; The contour analysis algorithm is used to extract local stamping contour features. By comparing the geometric deviation of the local stamping contour features with the standard stamping contour template and combining the preset tolerance threshold range, a stamping deformation warning signal is output.

2. The intelligent defect detection method for stamping dies according to claim 1, characterized in that: The perspective correction model is expressed as: ， ， Where, and Respectively represent the width and length of the adjusted area residual image, and Respectively represent the width and length of the residual image of the divided area before adjustment, Indicates the vertical height between the industrial camera and the stamping plate. Indicates the horizontal distance from the center point of the corresponding area residual material image to the center point of the industrial camera. Indicates the tilt angle of the industrial camera shooting downwards. Indicates the azimuth angle of the center point of the corresponding sub-area residual material image relative to the industrial camera's visual axis. The default is 0.

3. The intelligent defect detection method for stamping dies according to claim 1, characterized in that: After dividing the stamping waste material image into a plurality of sub-region waste material images, the method further includes: Select the largest area of ​​the sub-region residual material image as the burr analysis object; The local stamping contour features corresponding to the burr analysis object are imported into the texture recognition model to obtain the burr quantitative index, and stamping burr abnormality warning information is generated based on the index.

4. The intelligent defect detection method for stamping dies according to claim 3, characterized in that: The texture recognition model is expressed as: ， ， ， ， Where, and They are the gradient components along the X-axis and Y-axis directions, which are calculated by the Sobel operator. Represents the gradient magnitude of each pixel in the image, Represents the gradient direction of each pixel in the image, represents the gradient change index, Represents a local window used to traverse the entire image, Represents the total number of pixels in the window, represents the gradient magnitude of each pixel in the window, Represents the average value of the gradient amplitude of all pixels in the window, represents the glitch quantification index, is a measure of gradient direction consistency, expressed as standard deviation, The area of ​​the detected burr region is extracted from the binary gradient image through morphological operations. 、 and Represents the weight coefficient.

5. An intelligent defect detection system for stamping dies, characterized in that: The system comprises: An image acquisition module is used to acquire an image of the stamping residue using preset orientation parameters after the stamping process is completed. The orientation parameters include an acquisition angle and spatial coordinates. An image partitioning module is used to divide the stamping waste image into a plurality of sub-region waste images according to the coordinate information of the positioning reference hole in the stamping waste image; An image correction module is used to input the residual material images of each sub-region into the perspective correction model and output the residual material images of each sub-region with normalized sizes; The defect detection module is used to extract local stamping contour features using a contour analysis algorithm, and output a stamping deformation warning signal by comparing the geometric deviation of the local stamping contour features with the standard stamping contour template and combining it with a preset tolerance threshold range.

6. The intelligent defect detection system for stamping dies according to claim 5, characterized in that: The perspective correction model is expressed as: ， ， Where, and Respectively represent the width and length of the adjusted area residual image, and Respectively represent the width and length of the residual image of the divided area before adjustment, Indicates the vertical height between the industrial camera and the stamping plate. Indicates the horizontal distance from the center point of the corresponding area residual material image to the center point of the industrial camera. Indicates the tilt angle of the industrial camera shooting downwards. Indicates the azimuth angle of the center point of the corresponding sub-area residual material image relative to the industrial camera's visual axis. The default is 0.

7. The intelligent defect detection system for stamping dies according to claim 5, characterized in that: The image partitioning module is further configured to: Select the largest area of ​​the sub-region residual material image as the burr analysis object; The local stamping contour features corresponding to the burr analysis object are imported into the texture recognition model to obtain the burr quantitative index, and stamping burr abnormality warning information is generated based on the index.

8. An electronic device, characterized in that: include: A processor and a memory, the memory being used to store computer program code, the computer program code including computer instructions, and when the processor executes the computer instructions, the electronic device executes the intelligent defect detection method for stamping dies as described in any one of claims 1 to 4.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, which includes program instructions. When the program instructions are executed by a processor of an electronic device, the processor executes the intelligent defect detection method for stamping dies according to any one of claims 1 to 4.

Images