Defect detection method and defect detection equipment for circular printed matter

Abstract

The present invention discloses a defect detection method and defect detection device for circular printed products. The method comprises: automatically extracting a minimum circumscribed rectangle from a sample circular printed product area to obtain an image correction ratio; fitting the circular outline of the circular printed product in the sample circular printed product area after the ratio correction, arbitrarily selecting two matching areas in the image area enclosed by the circular outline, and performing mark point positioning based on the two matching areas; obtaining an image of the circular printed product to be detected, performing ratio correction on the image of the circular printed product to be detected according to the image correction ratio, and selecting two matching areas for mark point positioning; then, inversely transforming the mark point positioning result into the ratio-corrected image of the circular printed product to be detected; aligning the image of the circular printed product to be detected with the sample circular printed product image, and performing defect detection using a difference image method. The present invention can improve the accuracy and efficiency of automated detection of circular printed products.

Description

Technical Field

[0001] The present invention relates to the technical field of defect detection, and in particular to a defect detection method and defect detection equipment for circular printed products. Background Art

[0002] Circular printed materials are printed on a circular substrate with patterns, letters, and text. After printing, circular printed materials need to be inspected for print quality, such as missing prints, false prints, scratches, stains, and color differences.

[0003] Currently, in order to improve detection efficiency, vision systems are often used to detect defects by collecting real-time images of circular printed products. However, existing detection methods still have the following shortcomings:

[0004] (1) Currently, line scan cameras are mainly used to capture real-time images of circular printed products. Due to the limitations of hardware conditions or detection environment, the horizontal and vertical resolutions of the images actually captured by line scan cameras are difficult to reach 1:1, resulting in deformation of the circular printed product image (for example, a perfect circle becomes an ellipse), which affects the detection accuracy.

[0005] (2) When the circular printed matter is conveyed to the line scan camera, the direction of the printed image cannot be fixed, resulting in the pattern, text, etc. in the circular printed matter image captured by the line scan camera to be deflected at different angles. Due to the special nature of the circle, the angle deflection is difficult to be discovered at the beginning of the detection. The deflection is often discovered in the middle of the detection, resulting in unsuccessful detection.

[0006] (3) In the actual printing production process, the image quality of circular printed products is affected by many factors, which may come from printing equipment, materials, process flow, etc. When identifying defects, the current detection method only uses qualified products for threshold training. The evaluation criteria are limited and there is a risk of missed detection. Summary of the Invention

[0007] The present invention aims to solve at least one of the problems existing in the prior art.

[0008] To this end, the present invention provides a defect detection method and defect detection equipment for circular printed products, which can improve the accuracy and detection efficiency of automated detection of circular printed products.

[0009] The technical solution adopted by the present invention to solve its technical problem is:

[0010] A method for detecting defects in circular printed products comprises the following steps:

[0011] S1, build detection template,

[0012] S11, extracting a sample circular printed product area from the sample circular printed product image, automatically extracting a minimum circumscribed rectangle on the sample circular printed product area, and obtaining an image correction ratio based on the actual size of the circular printed product. ;

[0013] S12, fitting the circular outline of the circular printed product in the sample circular printed product area after the scale correction, and recording the coordinates of the center of the circular outline ; arbitrarily select two matching areas in the image area surrounded by the circular outline, and perform Mark point positioning based on the two matching areas;

[0014] S2, performing defect detection on the circular printed product to be inspected according to the inspection template,

[0015] S21, obtaining an image of a circular printed matter to be detected, and correcting the ratio according to the image Performing scale correction on the circular printed matter image to be detected;

[0016] S22, selecting two matching areas in the printed image to be detected and performing Mark point positioning in the same manner as step S12; then inversely transforming the Mark point positioning results into the scale-corrected circular printed image to be detected;

[0017] S23 , aligning the image of the circular printed product to be inspected with the image of the sample circular printed product, and performing defect detection using a difference image method.

[0018] Furthermore, in step S12, Mark point positioning is performed based on the two matching areas, including:

[0019] To match the area and circle center coordinates The closest distance as the radius , to match the area and circle center coordinates The farthest distance as the radius , the radius As the inner radius, the radius Cut out the annular area containing the matching area as the outer circle radius;

[0020] The circular area is transformed into a rectangular area by polar coordinate expansion. The width of the rectangular area is , the length is ;

[0021] Mark point positioning is performed in the rectangular area.

[0022] Furthermore, any point is selected in the rectangular area. , Pixels in the circular print image after scale correction The corresponding relationship is:

[0023] , ,

[0024] Represents the pixel coordinates within the rectangular area. Represents pixel coordinates in the circular print image.

[0025] Furthermore, the alignment process in step S23 includes:

[0026] The two Mark points in the sample circular print image are recorded as and , and The connection line is ; Note the two Mark points in the circular printed image to be detected as and , and The connection line is ;

[0027] Connection and The angle between That is, the rotation angle of the circular printed image to be detected to the sample circular printed image;

[0028] and The center point between the two points is , and The center point between the two points is , with a point is the center of rotation, Rotation around the center of rotation After the point , Rotation around the center of rotation After the point , and Center point of the line The center point Rotation around the center of rotation The new coordinate point after and The difference in the X-axis and Y-axis directions is the displacement deviation of the circular printed image to be tested to the sample circular printed image. 、 ;

[0029] The circular printed matter image to be detected is rotated according to the rotation angle , displacement deviation 、 Perform affine transformation to align with the sample circular print image.

[0030] Furthermore, after the circular printed matter image to be detected is aligned with the sample circular printed matter image, the difference between corresponding pixels is calculated, and the difference is compared with a set threshold range to obtain a defect detection result.

[0031] Furthermore, the process of obtaining the threshold range includes:

[0032] Select a qualified product image,

[0033] Converting the qualified product image into HSV space and setting random perturbations of hue, saturation, and brightness to obtain a set of color difference perturbation images;

[0034] Gamma correction is performed on the qualified product image to obtain a light-dark transformation disturbance image;

[0035] Convolving the qualified product image with a bright spot Gaussian kernel and a shadow Gaussian kernel to generate a local disturbance image;

[0036] Performing affine transformation and motion blur kernel convolution processing on the qualified product image to obtain a disturbed image;

[0037] All the above perturbation images are used for training to obtain the threshold range.

[0038] Furthermore, the image correction ratio ,

[0039] Indicates the horizontal resolution of the image. Indicates the resolution of the image in the vertical direction, Indicates the actual diameter of a circular print. Indicates the width of the minimum enclosing rectangle, Indicates the height of the minimum enclosing rectangle.

[0040] The present invention also provides a circular printed product defect detection device, comprising:

[0041] A vision mechanism for capturing images of circular printed matter;

[0042] The host computer is used to execute the circular printed product defect detection method to perform defect detection on the circular printed product image.

[0043] The beneficial effects of the present invention are:

[0044] (1) The present invention adopts a unified image correction ratio, which can improve the problem that the vertical and horizontal resolution of the image cannot reach 1:1 due to the line scan camera, so that in subsequent inspections, the image of the circular printed product can reflect the actual size, reduce deformation, and help improve the inspection accuracy.

[0045] (2) The present invention adopts the method of "matching area + polar coordinate expansion + mark point inverse transformation" to locate feature points, which will not be affected by angle deflection and is conducive to improving matching positioning efficiency and positioning accuracy; the dual mark point alignment method can reduce the impact of angle deflection and improve alignment accuracy and efficiency.

[0046] (3) The present invention adopts a dynamic disturbance method to obtain the threshold range of defect detection, which can make the threshold range more accurate and have strong anti-interference ability, which is conducive to reducing the risk of missed detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The present invention will be further described below with reference to the accompanying drawings and examples.

[0048] Figure 1 It is a flow chart of the defect detection method of circular printed matter of the present invention.

[0049] Figure 2 Schematic diagram of the angle deflection of a circular printed matter according to the present invention.

[0050] Figure 3 It is a schematic diagram of the circular area of ​​the present invention being expanded into a rectangular area.

[0051] Figure 4 It is a schematic diagram of the double mark point alignment of the present invention. DETAILED DESCRIPTION

[0052] The present invention will now be described in further detail with reference to the accompanying drawings, which are simplified schematic diagrams that illustrate the basic structure of the present invention in a schematic manner.

[0053] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "radial", "circumferential" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, features defined as "first" or "second" may explicitly or implicitly include one or more of such features. In the description of the present invention, unless otherwise specified, "multiple" means two or more.

[0054] In the description of the present invention, it should be noted that, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed, detachable, or integral connections; mechanical or electrical connections; direct or indirect connections through an intermediate medium; and internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on the specific circumstances.

[0055] like Figure 1 As shown, the defect detection method for circular printed products of this embodiment includes the following steps:

[0056] S1. Build a detection template.

[0057] S11. Extracting a sample circular print area from the sample circular print image, automatically extracting a minimum circumscribed rectangle on the sample circular print area, and obtaining an image correction ratio based on the actual size of the circular print. The collected sample circular print image is an original image, which contains not only the circular print but also background areas or other interference items. Therefore, it is necessary to extract the sample circular print area from the sample circular print image and remove the redundant interference items and background areas.

[0058] S12. Fitting the circular outline of the circular printed product in the sample circular printed product area after scale correction, and recording the coordinates of the center of the circular outline ; Select any two matching areas in the image area surrounded by the circular outline, and locate the mark points based on the two matching areas.

[0059] S2. Perform defect detection on the circular printed product to be inspected according to the inspection template.

[0060] S21, obtain the image of the circular printed matter to be detected, and correct the ratio according to the image Perform scale correction on the circular printed image to be detected.

[0061] S22 , selecting two matching areas in the printed image to be detected and performing Mark point positioning in the same manner as step S12 ; and then inversely transforming the Mark point positioning result into the circular printed image to be detected after scale correction.

[0062] S23 , aligning the image of the circular printed product to be inspected with the image of the sample circular printed product, and performing defect detection using a difference image method.

[0063] In other words, the inspection process in this embodiment first constructs a detection template based on a sample circular printed product image (an image of a qualified product), and then uses the detection template to inspect the real-time image. This detection template primarily consists of two parts: the first determines the image correction ratio, and the second locates the mark points based on the matching area.

[0064] For the first part, since the same line scan camera is used, the vertical and horizontal resolutions of the original images collected are the same. After the image correction ratio of the detection template is determined, the subsequent real-time images collected can directly use this image correction ratio. , Indicates the horizontal resolution of the image. Indicates the resolution of the image in the vertical direction, Indicates the actual diameter of a circular print. Indicates the width of the minimum enclosing rectangle, Indicates the height of the minimum bounding rectangle. When correcting, it is inconvenient to ensure the horizontal ratio of the image. Use the image correction ratio Perform bilinear interpolation correction on the image's vertical scale. Apply the same correction to subsequent images of circular printed products to be inspected. This ensures that the image maintains the same scale during subsequent defect inspections, minimizing image distortion and improving subsequent inspection accuracy.

[0065] For the second part, Figure 3 As shown, in step S12, the Mark point is positioned based on the two matching areas, including: The closest distance as the radius , to match the area and circle center coordinates The farthest distance as the radius , the radius As the inner radius, the radius The circular area containing the matching area is cut out as the outer circle radius; the circular area is transformed into a rectangular area by polar coordinate expansion, and the width of the rectangular area is , the length is ; Mark point positioning in the rectangular area.

[0066] It should be noted that due to the special nature of the circle, when the angle of the circular printed matter is deflected, it cannot be detected from the circular outline. However, the angle of the image printed on the circular printed matter will be deflected. If only fixed coordinate positions are selected during feature selection, the selected features may not be the same, which is detrimental to the accuracy of subsequent detection results. Therefore, this embodiment improves the positioning method of the Mark point. In this embodiment, the circular outline of the circular printed matter is first fitted in the sample circular printed matter area, and the center coordinates of the circular outline are recorded. Then, two matching areas are randomly selected in the circular image area surrounded by the circular outline (respectively marked as matching area A and matching area B), and the matching area is, for example, a square area. Then, the matching area and the circle center coordinates are The closest distance as the radius , to match the area and circle center coordinates The farthest distance as the radius ; Set the radius As the inner radius, the radius As the outer circle radius, the circular area containing the matching area is cut out, that is, two circular areas can be obtained. Then, the circular area is transformed into a rectangular area by polar coordinate expansion, and the Mark point is located in the rectangular area. Figure 2 As shown in the figure, when the circular printed matter is deflected, the position of the same feature in the image is different, but the feature will fall into the same circular area before and after deflection. Then expand the circular area into a rectangular area and locate the mark point in the rectangular area. When locating, select any point in the rectangular area. (as a Mark point), Pixels in the circular print image after scale correction The corresponding relationship is: , , Represents the pixel coordinates within the rectangular area. Represents the pixel coordinates in the circular print image. In other words, after locating a feature using a Mark point within a rectangular area, the Mark point coordinates can be reverse-transformed to the sample circular print area using the corresponding relationship. This embodiment uses the "matching area + polar coordinate expansion + Mark point reverse transformation" method to locate feature points. This method is not affected by angular deflection and does not require the establishment of multiple matching templates, greatly improving matching efficiency, positioning efficiency, and positioning success rate.

[0067] After constructing the inspection template, the circular printed product image can be inspected using the inspection template. First, the circular printed product image is scaled, and then the mark points are located. The operation process of these two steps is the same as step S1. It should be noted that the inspection step also includes a registration process, that is, the position of the circular printed product image to be inspected is aligned with the sample circular printed product image based on the mark points.

[0068] like Figure 4 As shown, the alignment process in step S23 includes: recording two Mark points in the sample circular print image as and , and The connection line is ; The two Mark points in the circular printed image to be detected are and , and The connection line is ; Connection and The angle between That is, the rotation angle of the circular printed image to be detected to the sample circular printed image; and The center point between the two points is , and The center point between the two points is , with a point is the center of rotation, Rotation around the center of rotation After the point , Rotation around the center of rotation After the point , and Center point of the line The center point Rotation around the center of rotation The new coordinate point after and The difference in the X-axis and Y-axis directions is the displacement deviation of the circular printed image to be tested to the sample circular printed image. 、 ; Rotate the circular printed image to be detected according to the rotation angle , displacement deviation 、 Perform affine transformation and align with the sample circular printed product image. This embodiment adopts a dual-mark point alignment method, which can improve the problem of inaccurate positioning caused by the rotation of the circular printed product angle, and is conducive to improving the subsequent defect detection accuracy.

[0069] After the circular printed image to be detected is aligned with the sample circular printed image, the difference between the corresponding pixels is calculated, and the difference is compared with the set threshold range to obtain the defect detection result. If the difference exceeds the threshold range, it is considered that there is a defect. The process of obtaining the threshold range includes: selecting a qualified product image, converting the qualified product image to the HSV space, and setting random perturbations of hue, saturation, and brightness to obtain a set of color difference perturbation images; performing Gamma correction on the qualified product image to obtain a light-dark conversion perturbation image; using the bright spot Gaussian kernel and shadow Gaussian kernel convolution to generate a local perturbation image for the qualified product image; performing affine transformation and motion blur kernel convolution on the qualified product image to obtain a perturbation image; using all the above perturbation images for training to obtain a threshold range. This embodiment adds dynamic perturbation to expand the threshold range when setting the threshold range for defect recognition, which can reduce the risk of missed detection.

[0070] For example, when perturbing color difference, (a) hue H represents the position of the color on the color wheel, that is, the type of color, and the value range of H is 0~360 degrees; the perturbation range of normal printing color difference simulation is within ±10, which is used to simulate relatively slight changes in light and dark; (b) pixel-by-pixel random perturbation: a set of random perturbation matrix data with the same size as the H space between [-10, 10] is generated, and superimposed with H to generate a new set of randomly perturbed images; (c) overall color deviation perturbation: a set of random constant matrix data with the same size as the H space between [-10, 10] is generated, and superimposed with H to generate a new set of unified perturbation images; (d) the value ranges of saturation S and lightness V are different in different image processing libraries, generally divided into 0~1 or 0~255. Taking 0~255 as an example, the perturbation range of saturation S and lightness V in normal printing color difference simulation is within ±15. Its perturbation method is similar to hue H, and can be divided into pixel-by-pixel random perturbation and overall deviation perturbation.

[0071] For example, when brightness is perturbed, gamma correction is a nonlinear adjustment method for image brightness. It is suitable for simulating variations in brightness and darkness caused by factors such as ink concentration, paper absorption, printing pressure, and uneven exposure during the printing process. The input image I is normalized to obtain image Iin. A gamma transformation is then performed on this image: Iout = Iinγ. A γ value range of 0.9 to 1.0 is used to simulate slight brightening changes, while a γ value range of 1.1 to 1.3 is used to simulate slight darkening changes, generating brightness perturbations.

[0072] Obtaining the defect detection threshold range through dynamic perturbation can save the training of a certain number of qualified product images, simplify the operation, and enhance practicality.

[0073] The present invention also provides a circular printed product defect detection device, comprising: a visual mechanism for capturing images of the circular printed product; and a host computer for executing a circular printed product defect detection method to detect defects in the circular printed product images. The visual mechanism is, for example, a line scan camera. The host computer can perform operations such as scale correction, matching and positioning, and defect detection on the images.

[0074] In summary, the circular printed product defect detection method and defect detection device of the present invention have the following advantages:

[0075] (1) The present invention adopts a unified image correction ratio, which can improve the problem that the vertical and horizontal resolution of the image cannot reach 1:1 due to the line scan camera, so that in subsequent inspections, the image of the circular printed product can reflect the actual size, reduce deformation, and help improve the inspection accuracy.

[0076] (2) The present invention adopts the method of "matching area + polar coordinate expansion + mark point inverse transformation" to locate feature points, which will not be affected by angle deflection and is conducive to improving matching positioning efficiency and positioning accuracy; the dual mark point alignment method can reduce the impact of angle deflection and improve alignment accuracy and efficiency.

[0077] (3) The present invention adopts a dynamic disturbance method to obtain the threshold range of defect detection, which can make the threshold range more accurate and have strong anti-interference ability, which is conducive to reducing the risk of missed detection.

[0078] With the above-described preferred embodiments of the present invention as a guide, and with reference to the above description, relevant personnel are fully capable of making various changes and modifications without departing from the technical spirit of this invention. The technical scope of this invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A method for detecting defects in circular printed matter, characterized in that: The following steps are involved: S1, build detection template, S11, extracting a sample circular printed product area from the sample circular printed product image, automatically extracting a minimum circumscribed rectangle on the sample circular printed product area, and obtaining an image correction ratio based on the actual size of the circular printed product. ; S12, fitting the circular outline of the circular printed product in the sample circular printed product area after the scale correction, and recording the coordinates of the center of the circular outline ; Randomly select two matching areas in the image area surrounded by the circular outline, and perform Mark point positioning based on the two matching areas; S2, performing defect detection on the circular printed product to be inspected according to the inspection template, S21, obtaining an image of a circular printed matter to be detected, and correcting the ratio according to the image Performing scale correction on the circular printed matter image to be detected; S22, selecting two matching areas in the printed image to be detected and performing Mark point positioning in the same manner as step S12; then inversely transforming the Mark point positioning results into the scale-corrected circular printed image to be detected; S23, aligning the image of the circular printed product to be inspected with the image of the sample circular printed product, and performing defect detection using a differential image method; In step S12, Mark point positioning is performed based on the two matching areas, including: To match the area and circle center coordinates The closest distance as the radius , to match the area and circle center coordinates The farthest distance as the radius , the radius As the inner radius, the radius Cut out the annular area containing the matching area as the outer circle radius; The circular area is transformed into a rectangular area by polar coordinate expansion. The width of the rectangular area is , the length is ; Mark point positioning is performed in the rectangular area; Select any point in the rectangular area , Pixels in the circular print image after scale correction The corresponding relationship is: ， ， Represents the pixel coordinates within the rectangular area. represents the pixel coordinates in the circular print image; The alignment process in step S23 includes: The two Mark points in the sample circular print image are recorded as and , and The connection line is ; Note the two Mark points in the circular printed image to be detected as and , and The connection line is ; Connection and The angle between That is, the rotation angle of the circular printed image to be detected to the sample circular printed image; and The center point between the two points is , and The center point between the two points is , with a point is the center of rotation, Rotation around the center of rotation After the point , Rotation around the center of rotation After the point , and Center point of the line The center point Rotation around the center of rotation The new coordinate point after and The difference in the X-axis and Y-axis directions is the displacement deviation of the circular printed image to be tested to the sample circular printed image. 、 ; The circular printed matter image to be detected is rotated according to the rotation angle , displacement deviation 、 Perform affine transformation to align with the sample circular print image.

2. The defect detection method for circular printed matter according to claim 1, wherein: After the circular printed product image to be detected is aligned with the sample circular printed product image, the difference between corresponding pixels is calculated, and the difference is compared with a set threshold range to obtain a defect detection result.

3. The defect detection method for circular printed matter according to claim 2, wherein: The process of obtaining the threshold range includes: Select a qualified product image, Converting the qualified product image into HSV space and setting random perturbations of hue, saturation, and brightness to obtain a set of color difference perturbation images; Gamma correction is performed on the qualified product image to obtain a light-dark transformation disturbance image; Convolving the qualified product image with a bright spot Gaussian kernel and a shadow Gaussian kernel to generate a local disturbance image; Performing affine transformation and motion blur kernel convolution processing on the qualified product image to obtain a disturbed image; All the above perturbation images are used for training to obtain the threshold range.

4. The defect detection method for circular printed matter according to claim 1, wherein: The image correction ratio , Indicates the horizontal resolution of the image. Indicates the resolution of the image in the vertical direction, Indicates the actual diameter of a circular print. Indicates the width of the minimum enclosing rectangle, Indicates the height of the minimum enclosing rectangle.

5. A defect detection device for circular printed products, characterized in that: include: A vision mechanism for capturing images of circular printed matter; The host computer is used to execute the circular printed product defect detection method according to any one of claims 1 to 4 to perform defect detection on the circular printed product image.

Images