Adhesive tape edge point detection method

Abstract

The invention discloses an adhesive tape edge point detection method, which is carried out according to the following steps in sequence: firstly, acquiring an image of an adhesive tape, then segmenting the image into a plurality of sub-images with unit lengths along the length direction of the adhesive tape, and carrying out image processing operation on each sub-image, edge information extraction, edge information filtering processing, edge pair and glue width calculation, and calculation result comparison. According to the steps, performing operations of rectangular ROI establishment, edge information extraction, edge information filtering processing, edge pair and adhesive width calculation and calculation result comparison on each sub-image until all the sub-images in one adhesive tape image are processed, outputting a result, and if an error is reported, outputting the result. A worker can find out the adhesive tape with the adhesive width not meeting the requirement for repairing according to the serial number of the adhesive tape image, and if an error is not reported, it is indicated that the width of the adhesive tape meets the requirement and does not need to be processed.

Description

Technical Field

[0001] The invention relates to the field of image processing for visual detection, in particular to a method for detecting edge points of a rubber strip. Background Art

[0002] The automated gluing process has been widely used in industrial production due to its high efficiency and precision. Especially in the field of automobile manufacturing, there are many places where gluing is required, and the quality of the rubber strips is directly related to the performance and life of the product. The rubber strips not only have multiple functions such as anti-corrosion, adhesion, connection and fixation, shock absorption, and waterproofing, but are also the key factor in ensuring the overall performance of the car. If the quality of the glue coating is not up to standard, it will cause a series of quality problems such as leakage, abnormal noise, and corrosion, which will seriously weaken the performance of the product. Therefore, in the modern gluing process, it is particularly important to carry out strict quality inspection and control of the glued strips. However, the existing gluing detection methods have obvious deficiencies in terms of calculation amount, running speed, and quantitative analysis, which are prone to missed detection and false detection, and cannot meet the needs of fast and accurate detection.

[0003] Therefore, a method or device capable of solving the above problems is now needed. Summary of the invention

[0004] The present invention aims to solve the above-mentioned deficiencies in the prior art and proposes a method for detecting edge points of rubber strips. The method subdivides the rubber strips into multiple sub-regions and performs refined edge point filtering on the region of interest (ROI) of each sub-region, thereby achieving accurate detection of the edge of the rubber strips.

[0005] The technical solution is as follows: A method for detecting edge points of a rubber strip, characterized in that the method is performed in sequence according to the following steps:

[0006] First, the image of the rubber strip is collected, and then the image is divided into multiple sub-images of unit length along the length direction of the rubber strip, and image processing operations are performed on each sub-image.

[0007] The image processing operations include: establishing a rectangular ROI, extracting edge information, filtering edge information, calculating edge alignment and glue width, and comparing calculation results.

[0008] A. The specific operation of the step of establishing a rectangular ROI is as follows:

[0009] First, create a rectangular ROI and determine the coordinates of the center position of the rectangular ROI (x0, y0), as well as the half length X and half width Y of the rectangular ROI.

[0010] Then, based on the coordinates of the above center position and the values ​​of half-length X and half-width Y, the coordinates of the four vertices F, Q, E, and C of the rectangular ROI, as well as the coordinates of the midpoints B and D of its wide side, are calculated.

[0011] F(x0+Y*sinθ-X*cosθ, y0-Y*cosθ-X*sinθ)

[0012] Q(x0+X*cosθ+Y*sinθ, y0+X*sinθ-Y*cosθ)

[0013] E(x0+X*cosθ-Y*sinθ, Y0+X*sinθ+Y*cosθ)

[0014] C(x0-X*cosθ-Y*sinθ, y0-X*sinθ+Y*cosθ)

[0015] B(x0-X*cosθ,y0-X*sinθ)

[0016] D(x0+X*cosθ,y0+X*sinθ)

[0017] Using the boundary coordinates calculated by the above formula, a mask image with the same size as the original image is created, where the grayscale value within the rectangular ROI area is set to 1, and the grayscale value outside the area is set to 0.

[0018] The created mask image is applied to the original image, and only the grayscale values ​​in the ROI area within the mask will be retained, thereby achieving operations that focus only on the ROI area.

[0019] B. The specific operation of the edge information extraction step is as follows:

[0020] First convert the image to grayscale.

[0021] Then the grayscale image is subjected to edge enhancement processing. During the edge enhancement processing, two Gaussian filters with different standard deviations are applied to the grayscale image, and the difference between the two filter results is calculated to obtain a Gaussian difference edge enhanced image to highlight the edges in the image.

[0022] The grayscale image is ternary processed. By setting the threshold, the image is converted into an image consisting of only three grayscale values: 0, 125, and 255, thereby further strengthening the edge of the grayscale image, simplifying the image and reducing the complexity of subsequent processing.

[0023] The ternary image is used as the input of edge detection, and the edges of the image are identified through noise reduction, gradient calculation, non-maximum suppression and double threshold detection steps.

[0024] C. The specific operation of the edge information filtering and processing step is as follows:

[0025] First, perform a deduplication operation on the boundary coordinate data obtained in the step of establishing a rectangular ROI to ensure that all edge point coordinates are unique.

[0026] Determine a circle O for filtering edge points. The center of circle O is located on line segment BD. The diameter of circle O is the relatively smaller value between the X value and the Y value. The radius of circle O is recorded as r.

[0027] Let the circle O move from point B to point D. The initial position of the center of the circle is at point B + r, and the end position is at point D - r. The step length of each movement of the center of the circle is one pixel.

[0028] The moving trajectory of circle O is perpendicular to and interferes with the two boundary lines L1 and L2 of the rubber strip on the grayscale image. When circle O intersects L1, circles O at different positions are marked as circle O-L1-1, circle O-L1-2, ... circle O-L1-m in the order of movement. Similarly, when circle O intersects L2, circles O at different positions are marked as circle O-L2-1, circle O-L2-2, ... circle O-L2-m in the order of movement, where m is a natural number.

[0029] In the circle O-L1 group and the circle O-L2 group, the number of points on L1 or L2 covered by them will increase from 0 to the maximum and then decrease from the maximum to 0. The circle O with the most covered points is selected from the circle O-L1 group and the circle O-L2 group, and marked as circle O1 and circle O2. If there are multiple circles with the most covered points in the same group, the circle O with the smallest serial number is selected;

[0030] D. The specific operations of the edge pair and glue width calculation steps are as follows:

[0031] The average value of the points in circle O1 and circle O2 is calculated according to the following formula:

[0032]

[0033] Where n is the number of coordinate points;

[0034] According to the above calculation results, a point A1 corresponding to the circle O1 and a point A2 corresponding to the circle O2 are determined respectively.

[0035] Then calculate the slope and intercept of the straight line between points B and D according to the following formula:

[0036] Slope k BD The calculation formula is:

[0037]

[0038] Intercept b BD The calculation formula is:

[0039] b BD=y B -k BD x B

[0040] Then the equation of the line segment BD is:

[0041] y BD =k BD x+b BD

[0042] Then use the following formula to find the coordinates of the perpendicular point between point A1 and line segment BD, recorded as A1-perpendicular, and the coordinates of the perpendicular point between point A2 and line segment BD, recorded as A2-perpendicular. The above two perpendicular points are the only edge pair finally obtained.

[0043]

[0044] Then use the following formula to calculate the distance d between the two new coordinate points, which is the actual glue width.

[0045]

[0046] Where x1 and y1 are the coordinates of point A1-vertical, and x2 and y2 are the coordinates of point A2-vertical.

[0047] E. Compare the obtained d value with the preset theoretical glue width value. If the difference between the d value and the theoretical glue width value is within the allowable error range, the actual glue width meets the requirements. If it exceeds the allowable error range, the actual glue width does not meet the requirements and an error is reported.

[0048] Compared with the prior art, the present invention has the following advantages:

[0049] The rubber strip edge point detection method disclosed in the present invention has significant advantages in many aspects compared with traditional methods. First, by applying circular filtering technology, the method can more accurately identify the effective edge points in the region of interest (ROI), and the circular shape better adapts to the curvature of the edge, thereby improving the recognition accuracy. In addition, circular filtering can effectively reduce unnecessary calculations, simplify the processing flow, and improve detection efficiency. At the same time, the method significantly suppresses the influence of noise, and further ensures the accuracy of the output edge point coordinates by calculating the mean of each group of edge points. Due to the accurate identification of effective edge points and the elimination of noise, the stability and robustness of the system in practical applications are enhanced, which is particularly important for high-precision application scenarios such as industrial automation and quality control. In summary, the improvements in accuracy, efficiency, noise suppression and system stability of this method make it not only suitable for rubber strip edge detection, but also have a wide range of application potentials, and can meet the needs of modern industrial and technological development. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 is a schematic diagram of the collected strip image.

[0051] Figure 2 It is a schematic diagram of establishing a rectangular ROI on a sub-image of the strip image (determining the center point and the side length).

[0052] Figure 3 It is a schematic diagram of establishing a rectangular ROI on a sub-image of the strip image (determining four vertices).

[0053] Figure 4 It is a schematic diagram of establishing a rectangular ROI on the sub-image of the strip image (determining line segment BD).

[0054] Figures 5 to 7 It is a schematic diagram of selecting different sub-images for processing on the rubber strip image.

[0055] The specific implementation of the present invention will be described below with reference to the accompanying drawings. Figures 1 to 7 As shown: A method for detecting edge points of a rubber strip is performed in the following steps:

[0056] First, the image of the rubber strip is collected, and then the image is divided into multiple sub-images of unit length along the length direction of the rubber strip, and image processing operations are performed on each sub-image.

[0057] The image processing operations include: establishing a rectangular ROI, extracting edge information, filtering edge information, calculating edge alignment and glue width, and comparing calculation results.

[0058] A. The specific operation of the step of establishing a rectangular ROI is as follows:

[0059] First, create a rectangular ROI and determine the coordinates of the center position of the rectangular ROI (x0, y0), as well as the half length X and half width Y of the rectangular ROI.

[0060] Then, based on the coordinates of the above center position and the values ​​of half-length X and half-width Y, the coordinates of the four vertices F, Q, E, and C of the rectangular ROI, as well as the coordinates of the midpoints B and D of its wide side, are calculated.

[0061] F(x0+Y*sinθ-X*cosθ, y0-Y*cosθ-X*sinθ)

[0062] Q(x0+X*cosθ+Y*sinθ, y0+X*sinθ-Y*cosθ)

[0063] E(x0+X*cosθ-Y*sinθ, y0+X*sinθ+Y*cosθ)

[0064] C(x0-X*cosθ-Y*sinθ, y0-X*sinθ+Y*cosθ)

[0065] B(x0-X*cos θ , y0-X*sinθ)

[0066] D(x0+X*cosθ,y0+X*sinθ)

[0067] Using the boundary coordinates calculated by the above formula, a mask image with the same size as the original image is created, where the grayscale value within the rectangular ROI area is set to 1, and the grayscale value outside the area is set to 0.

[0068] The created mask image is applied to the original image, and only the grayscale values ​​in the ROI area within the mask will be retained, thereby achieving operations that focus only on the ROI area.

[0069] B. The specific operation of the edge information extraction step is as follows:

[0070] First convert the image to grayscale.

[0071] Then the grayscale image is subjected to edge enhancement processing. During the edge enhancement processing, two Gaussian filters with different standard deviations are applied to the grayscale image, and the difference between the two filter results is calculated to obtain a Gaussian difference edge enhanced image to highlight the edges in the image.

[0072] The grayscale image is ternary processed. By setting the threshold, the image is converted into an image consisting of only three grayscale values: 0, 125, and 255, thereby further strengthening the edge of the grayscale image, simplifying the image and reducing the complexity of subsequent processing.

[0073] The ternary image is used as the input of edge detection, and the edges of the image are identified through noise reduction, gradient calculation, non-maximum suppression and double threshold detection steps.

[0074] C. The specific operation of the edge information filtering and processing step is as follows:

[0075] First, perform a deduplication operation on the boundary coordinate data obtained in the step of establishing a rectangular ROI to ensure that all edge point coordinates are unique.

[0076] Determine a circle O for filtering edge points. The center of circle O is located on line segment BD. The diameter of circle O is the relatively smaller value between the X value and the Y value. The radius of circle O is recorded as r.

[0077] Let the circle O move from point B to point D. The initial position of the center of the circle is at point B + r, and the end position is at point D - r. The step length of each movement of the center of the circle is one pixel.

[0078] The moving trajectory of circle O is perpendicular to and interferes with the two boundary lines L and L2 of the rubber strip on the grayscale image. When circle O intersects L1, circles O at different positions are marked as circle O-L1-1, circle O-L1-2, ... circle O-L1-m in the order of movement. Similarly, when circle O intersects L2, circles O at different positions are marked as circle O-L2-1, circle O-L2-2, ... circle O-L2-m in the order of movement, where m is a natural number.

[0079] In the circle O-L1 group and the circle O-L2 group, the number of points on L1 or L2 covered by them will increase from 0 to the maximum and then decrease from the maximum to 0. The circle O with the most covered points is selected from the circle O-L1 group and the circle O-L2 group, and marked as circle O1 and circle O2. If there are multiple circles with the most covered points in the same group, the circle O with the smallest serial number is selected;

[0080] D. The specific operations of the edge pair and glue width calculation steps are as follows:

[0081] The average value of the points in circle O1 and circle O2 is calculated according to the following formula:

[0082]

[0083] Where n is the number of coordinate points;

[0084] According to the above calculation results, a point A1 corresponding to the circle O1 and a point A2 corresponding to the circle O2 are determined respectively.

[0085] Then calculate the slope and intercept of the straight line between points B and D according to the following formula:

[0086] Slope k BD The calculation formula is:

[0087]

[0088] Intercept b BD The calculation formula is:

[0089] b BD =y B -k BD x B

[0090] Then the equation of the line segment BD is:

[0091] y BD =k BD x+b BD

[0092] Then use the following formula to find the coordinates of the perpendicular point between point A1 and line segment BD, recorded as A1-perpendicular, and the coordinates of the perpendicular point between point A2 and line segment BD, recorded as A2-perpendicular. The above two perpendicular points are the only edge pair finally obtained.

[0093]

[0094] Then use the following formula to calculate the distance d between the two new coordinate points, which is the actual glue width.

[0095]

[0096] Where x1 and y1 are the coordinates of point A1-vertical, and x2 and y2 are the coordinates of point A2-vertical.

[0097] E. Compare the obtained d value with the preset theoretical glue width value. If the difference between the d value and the theoretical glue width value is within the allowable error range, the actual glue width meets the requirements. If it exceeds the allowable error range, the actual glue width does not meet the requirements and an error is reported.

[0098] According to the above steps, a rectangular ROI is established for each sub-image, edge information is extracted, edge information is filtered, edge alignment and glue width are calculated, and calculation results are compared, until all sub-images in a glue strip image are processed and the results are output.

[0099] If an error is reported, the staff can find the rubber strip whose width does not meet the requirements and repair it according to the number of the rubber strip image. If no error is reported, it means that the width of the rubber strip meets the requirements and no processing is required.

Claims

1. A method for detecting edge points of a rubber strip, characterized in that: The method described is carried out in the following steps: First, the image of the rubber strip is collected, and then the image is divided into multiple sub-images of unit length along the length direction of the rubber strip, and image processing operations are performed on each sub-image. The image processing operations include: establishing a rectangular ROI, extracting edge information, filtering edge information, calculating edge alignment and glue width, and comparing calculation results. A. The specific operation of the step of establishing a rectangular ROI is as follows: First, create a rectangular ROI and determine the coordinates of the center position of the rectangular ROI (x0, y0), as well as the half length X and half width Y of the rectangular ROI. Then, based on the coordinates of the above center position and the values ​​of half-length X and half-width Y, the coordinates of the four vertices F, Q, E, and C of the rectangular ROI, as well as the coordinates of the midpoints B and D of its wide side, are calculated. F(x0+Y*sinθ-X*cosθ, y0-Y*cosθ-X*sinθ) Q(x0+X*cosθ+Y*sinθ, y0+X*sinθ-Y*cosθ) E(x0+X*cosθ-Y*sinθ, y0+X*sinθ+Y*cosθ) C(x0-X*cosθ-Y*sinθ, y0-X*sinθ+Y*cosθ) B(x0-X*cosθ,y0-X*sinθ) D(x0+X*cosθ,y0+X*sinθ) Using the boundary coordinates calculated by the above formula, a mask image with the same size as the original image is created, where the grayscale value within the rectangular ROI area is set to 1, and the grayscale value outside the area is set to 0. The created mask image is applied to the original image, and only the grayscale values ​​in the ROI area within the mask will be retained, thereby achieving operations that focus only on the ROI area. B. The specific operation of the edge information extraction step is as follows: First convert the image to grayscale. Then the grayscale image is subjected to edge enhancement processing. During the edge enhancement processing, two Gaussian filters with different standard deviations are applied to the grayscale image, and the difference between the two filter results is calculated to obtain a Gaussian difference edge enhanced image to highlight the edges in the image. The grayscale image is ternary processed. By setting the threshold, the image is converted into an image consisting of only three grayscale values: 0, 125, and 255, thereby further strengthening the edge of the grayscale image, simplifying the image and reducing the complexity of subsequent processing. The ternary image is used as the input of edge detection, and the edges of the image are identified through noise reduction, gradient calculation, non-maximum suppression and double threshold detection steps. C. The specific operation of the edge information filtering and processing step is as follows: First, perform a deduplication operation on the boundary coordinate data obtained in the step of establishing a rectangular ROI to ensure that all edge point coordinates are unique. Determine a circle O for filtering edge points. The center of circle O is located on line segment BD. The diameter of circle O is the relatively smaller value between the X value and the Y value. The radius of circle O is recorded as r. Let the circle O move from point B to point D. The initial position of the center of the circle is at point B + r, and the end position is at point D - r. The step length of each movement of the center of the circle is one pixel. The moving trajectory of circle O is perpendicular to and interferes with the two boundary lines L and L2 of the rubber strip on the grayscale image. When circle O intersects L1, circles O at different positions are marked as circle O-L1-1, circle O-L1-2, ... circle O-L1-m in the order of movement. Similarly, when circle O intersects L2, circles O at different positions are marked as circle O-L2-1, circle O-L2-2, ... circle O-L2-m in the order of movement, where m is a natural number. In the circle O-L1 group and the circle O-L2 group, the number of points on L1 or L2 covered by them will increase from 0 to the maximum and then decrease from the maximum to 0. The circle O with the most covered points is selected from the circle O-L1 group and the circle O-L2 group, and marked as circle O1 and circle O2. If there are multiple circles with the most covered points in the same group, the circle O with the smallest serial number is selected; D. The specific operations of the edge pair and glue width calculation steps are as follows: The average value of the points in circle O1 and circle O2 is calculated according to the following formula: Where n is the number of coordinate points; According to the above calculation results, a point A1 corresponding to the circle O1 and a point A2 corresponding to the circle O2 are determined respectively. Then calculate the slope and intercept of the straight line between points B and D according to the following formula: Slope k BD The calculation formula is: Intercept b BD The calculation formula is: b BD =y B -k BD x B Then the equation of the line segment BD is: y BD =k BD x+b BD Then use the following formula to find the coordinates of the perpendicular point between point A1 and line segment BD, recorded as A1-perpendicular, and the coordinates of the perpendicular point between point A2 and line segment BD, recorded as A2-perpendicular. The above two perpendicular points are the only edge pair finally obtained. Then use the following formula to calculate the distance d between the two new coordinate points, which is the actual glue width. Where x1 and y1 are the coordinates of point A1-vertical, and x2 and y2 are the coordinates of point A2-vertical. E. Compare the obtained d value with the preset theoretical glue width value. If the difference between the d value and the theoretical glue width value is within the allowable error range, the actual glue width meets the requirements. If it exceeds the allowable error range, the actual glue width does not meet the requirements and an error is reported.

Images