Online detection method for damage of printing scraper based on machine vision

Abstract

The invention relates to the technical field of image processing, in particular to a printing scraper damage online detection method based on machine vision. The method comprises the steps of collecting a surface image of a scraper, performing edge detection on the surface image to obtain an edge image, and obtaining a connected domain and a to-be-detected area corresponding to the connected domain in the edge image; analyzing the connected domains and the to-be-detected areas corresponding to the connected domains to obtain edge continuity characteristic values and shape regularity characteristic values of the to-be-detected areas, and screening all the to-be-detected areas to obtain candidate areas; determining main radiation directions by taking the central point of one candidate area as an original point, and acquiring a pixel point sequence corresponding to each main radiation direction; based on the pixel point sequence corresponding to each main radiation direction of one candidate area, obtaining a texture periodicity quantitative index of the candidate area, and finally screening all the candidate areas to obtain an abrasion and corrosion area. According to the invention, the damage of the printing scraper can be accurately detected.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and more specifically to an online detection method for damage to printing scrapers based on machine vision. Background Technology

[0002] As a core component of the printing process, the doctor blade's edge shape and surface integrity directly determine the ink transfer accuracy and printing stability. During continuous high-speed printing, the doctor blade surface inevitably experiences wear and corrosion due to the combined effects of mechanical friction and the chemical media of the ink. This damage is essentially corrosion caused by vibration or sliding of the contact surface under external loads in an oxygen-rich environment. Initially, this damage often manifests as micro-grooves, which can easily expand into macroscopic defects under dynamic high-pressure environments. Failure to achieve online identification and intervention can directly lead to doctor blade breakage, batch printing defects, or even unplanned downtime, significantly impacting production efficiency and resource costs. Therefore, the detection of wear and corrosion on the doctor blade surface is extremely important. Currently, the printing industry typically uses traditional machine vision technology for inspection: acquiring images of the doctor blade surface using an industrial camera and preprocessing them, then using morphological and edge detection techniques for defect segmentation, and finally locating and identifying defects based on shape and grayscale features.

[0003] In the scenario of detecting wear and corrosion of printing blades, due to the characteristics of the printing process and the influence of the blade's working environment, ink residue often interferes with the surface. Ink adhesion or drying during continuous brushing operations can form irregular dark areas and broken granular residues on the surface. These interference characteristics are highly similar to actual wear and corrosion in several dimensions: in terms of morphological features, both appear as irregularly contoured, non-uniform areas; in terms of optical properties, both exhibit abrupt changes in edge grayscale and locally isolated dark areas. Traditional algorithms based on edge detection and morphological processing can only make judgments based on superficial features such as shape and grayscale differences, making it difficult to distinguish the deeper differences between wear and corrosion and ink residue, leading to frequent misjudgments by the detection system. Summary of the Invention

[0004] To address the aforementioned technical problems, the present invention aims to provide an online detection method for damage to printing squeegees based on machine vision. The specific technical solution adopted is as follows: One embodiment of the present invention provides an online detection method for damage to printing squeegees based on machine vision, the method comprising: The surface image of the scraper is acquired, and edge detection is performed on the surface image to obtain the edge image; connected component analysis is performed on the edge image to obtain different connected components and the edges corresponding to each connected component; closing operation is performed on the edges corresponding to each connected component to obtain the detection area corresponding to each connected component; Edge continuity feature values ​​of each region to be detected are obtained based on the edges corresponding to each connected component and the edges of the regions to be detected corresponding to each connected component; shape regularity feature values ​​are obtained based on the area of ​​the region to be detected and the area of ​​the convex hull of the region to be detected; candidate regions are obtained by filtering all regions to be detected based on the edge continuity feature values ​​and the shape regularity feature values. The main radiation direction is determined by taking the center point of a candidate region as the origin. Pixels are taken in sequence along the main radiation direction to form the pixel sequence corresponding to the main radiation direction. The texture periodicity quantization index of the candidate region is obtained based on the distance of each pixel from the origin and the gray value of each pixel in the pixel sequence corresponding to each main radiation direction of the candidate region. The wear and corrosion region is obtained by screening all candidate regions based on the texture periodicity quantization index.

[0005] Preferably, the detection region corresponding to each connected component is obtained by performing a closing operation on the edges of each connected component, including: A morphological closing operation is performed on the edges of a connected component to obtain the morphologically connected edges of that connected component. The region enclosed by the morphologically connected edges is the region to be detected corresponding to that connected component.

[0006] Preferably, the edge continuity feature values ​​of each region to be detected are obtained based on the edges corresponding to each connected component and the edges of the regions to be detected corresponding to each connected component, including: The difference between the number of edge pixels on the edge of a connected component and the number of edge pixels on the edge of the corresponding region to be detected is obtained, and compared with the number of edge pixels on the edge of the corresponding region to be detected to obtain the edge continuity feature value of the region to be detected.

[0007] Preferably, the shape regularity feature value is obtained based on the area of ​​the region to be detected and the area of ​​the convex hull of the region to be detected, including: The area ratio is obtained by comparing the area of ​​a region to be detected with the area of ​​the convex hull of the region to be detected; the difference between the first preset value and the area ratio is used as the shape regularity feature value of the region to be detected.

[0008] Preferably, candidate regions are obtained by filtering all regions to be detected based on edge continuity feature values ​​and shape regularity feature values, including: The geometric morphology comprehensive index of a region to be detected is obtained by weighted summing of the normalized values ​​of the edge continuity feature value and the normalized values ​​of the shape regularity feature value. If the geometric morphology comprehensive index of a region to be detected is greater than or equal to the first reference threshold, then the region to be detected is a candidate region.

[0009] Preferably, determining the main radiation direction using the center point of a candidate region as the origin includes: In a Cartesian coordinate system with the center point of a candidate region as the origin, the four directions of 0°, 45°, 90° and 135° are determined as the main axis radiation directions.

[0010] Preferably, the texture periodicity quantization index of a candidate region is obtained based on the distance of each pixel from the origin and the gray value of each pixel in the pixel sequence corresponding to each main radiation direction of the candidate region, including: Calculate the distance of each pixel in the pixel sequence corresponding to a main radiation direction of a candidate region from the origin, and construct a scatter plot with the distance of each pixel from the origin as the horizontal axis and the gray value of each pixel as the vertical axis. Connect adjacent points in the scatter plot to obtain a connected image. Smooth the connected image to obtain a smoothed image. Take the extreme points in the smoothed image in order of the horizontal axis to form an extreme point sequence corresponding to the main radiation direction. Calculate the difference of the horizontal coordinates of every two adjacent extreme points in the extreme point sequence and average them to obtain the texture periodicity feature value of the main radiation direction. The difference of the horizontal coordinates of every two adjacent extreme points is the difference of the horizontal coordinates of the latter extreme point and the former extreme point in every two adjacent extreme points. Remove the main radiation direction with the largest texture periodicity feature value in the candidate region, and calculate the average of the texture periodicity feature values ​​of other main radiation directions as the texture periodicity quantification index of the candidate region.

[0011] Preferably, the wear and corrosion regions are obtained by screening all candidate regions based on the texture periodicity quantization index, including: Candidate regions whose texture periodicity quantization index is less than or equal to the second reference threshold are denoted as wear and corrosion regions.

[0012] The embodiments of the present invention have at least the following beneficial effects: This application obtains the connected components and edges of the connected components by analyzing the surface image of the scraper, and then performs a closing operation on the edges corresponding to each connected component to obtain the detection area corresponding to each connected component. Next, it analyzes each connected component, the edges corresponding to each connected component, the edges of the detection area corresponding to each connected component, and the area of ​​the detection area to obtain the edge continuity feature value and shape regularity feature value of the detection area. Finally, it filters all the detection areas based on the edge continuity feature value and shape regularity feature value to obtain candidate areas. The detection areas are initially filtered based on the differences between wear corrosion and ink in macroscopic geometric morphology, which narrows the scope of analysis and reduces the overall computational load. Then, the main radiation direction is determined by taking the center point of a candidate region as the origin, and pixels are taken sequentially along the main radiation direction to form a pixel sequence corresponding to the main radiation direction. Based on the distance of each pixel in the pixel sequence corresponding to each main radiation direction of a candidate region from the origin and the gray value of each pixel, the texture periodicity quantification index of the candidate region is obtained. Based on the texture periodicity quantification index, all candidate regions are screened to obtain the wear and corrosion region. Starting from the essential difference between wear and corrosion and ink micro-texture features, the consistency characteristics of radial texture distribution are evaluated, the regularity of texture diffusion from the center to the surrounding area is quantified, the subtle differences between wear and corrosion and ink residue are accurately captured, and the accuracy of wear and corrosion region detection of the doctor blade is improved. Attached Figure Description

[0013] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 A flowchart illustrating an online detection method for damage to printing squeegees based on machine vision, provided in an embodiment of the present invention. Figure 2 A schematic diagram of the area to be detected for an online detection method for damage to printing squeegees based on machine vision provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the main radiation direction of an online detection method for printing blade damage based on machine vision, provided in an embodiment of the present invention. Detailed Implementation

[0015] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of an online detection method for printing blade damage based on machine vision proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0016] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0017] The following description, in conjunction with the accompanying drawings, details a specific scheme for an online detection method for damage to printing squeegees based on machine vision, provided by the present invention.

[0018] Example: The main application scenario of this invention is as follows: This application proposes a processing idea and method based on "macro-geometric screening-micro-texture subdivision" by combining image processing related algorithms. By deeply exploring the essential differences between wear corrosion and ink in macro-geometric morphology and micro-texture features, it specifically solves the problems and shortcomings of traditional methods for detecting wear corrosion on the surface of the scraper, and achieves efficient and accurate defect detection processing.

[0019] Please see Figure 1 The diagram illustrates a flowchart of an online detection method for printing blade damage based on machine vision, according to an embodiment of the present invention. The method includes the following steps: Step S1: Acquire a surface image of the scraper and perform edge detection on the surface image to obtain an edge image; perform connected component analysis on the edge image to obtain different connected components and the edges corresponding to each connected component; perform a closing operation on the edges corresponding to each connected component to obtain the detection area corresponding to each connected component.

[0020] First, an image of the scraper surface needs to be acquired. Specifically, a light source system needs to be set up: a highly uniform coaxial LED light source (wavelength 450-650nm visible light band, power 20-40W) should be used, along with an adjustable polarizer to eliminate ink reflection on the scraper surface; a diffuser should be equipped simultaneously to weaken the problem of overexposure of the sharp edges of the scraper blade; specifically, for the reflection of dried ink on the metal scraper surface (reflectivity 30%-40%), a low-frequency pulsed coaxial light source (frequency 200-500Hz) should be used to enhance the grayscale contrast between the dark areas of wear and corrosion and the ink residue, so that the grayscale difference between the wear and corrosion area and the background is increased by 40-50%, and the ink reflection is prevented from covering the tiny corrosion points.

[0021] Furthermore, when selecting an industrial camera, the following parameters are required: resolution ≥ 16MP (pixel size ≤ 1.6μm), frame rate ≥ 50fps, and support for global shutter and HDR function (dynamic range ≥ 85dB).

[0022] Its lens configuration is as follows: when the flatness of the scraper blade edge is ±0.2mm, a low-distortion industrial telecentric lens (magnification 0.5X-2.0X, distortion rate ≤0.1%) is selected, with a working distance of 150-300mm, a depth of field ≥1.5mm, and equipped with a laser focusing module (accuracy ±0.01mm).

[0023] Shooting environment control: ambient light intensity ≤30 lux, installed in a closed light-shielding enclosure; equipment vibration amplitude <0.01mm to avoid image ghosting. Operating temperature 20-28℃ to reduce the impact of temperature on image quality.

[0024] Therefore, an image of the scraper surface is acquired. The acquisition needs to cover the entire working surface of the scraper, and the clarity of the acquired image is checked (grayscale contrast ≥ 30% is acceptable). This way, an image of the scraper surface can be obtained.

[0025] Next, the image of the scraper surface needs to be preprocessed. First, the image of the scraper surface is converted to a grayscale image using a conventional grayscale conversion method. Median filtering is used to optimize the quality of the grayscale image, reducing the interference of environmental noise on the printing scraper surface on feature extraction. Global histogram equalization is used to enhance the grayscale difference between the worn and corroded areas and the normal scraper surface, reducing visual redundancy in uniform background areas. Then, morphological operations such as cap transformation are combined to highlight the surface depressions and irregular texture features caused by wear and corrosion, weakening invalid information in flat areas, thus obtaining the scraper surface image. Finally, the Canny operator is used to perform edge detection on the preprocessed scraper surface image to obtain the edge image.

[0026] In surface images processed by edge detection algorithms, the apparent features of actual wear and corrosion are highly similar to those of ink residue, often resulting in insufficient screening accuracy. To address this issue, the core of this application lies in constructing a two-level discrimination strategy of "macro-geometric screening - micro-texture subdivision": First, combining the differences in edge continuity and shape regularity between the two, a geometric morphology index is constructed. This index reflects edge integrity by quantifying the degree of edge breakage and assesses the geometric regularity of the region by combining the degree of shape regularity, together forming a comprehensive discrimination criterion at the macro level. Based on the typical characteristics of wear and corrosion, a threshold range for the geometric morphology index is set to initially screen suspicious areas, directly excluding ink areas with continuous edges and regular shapes, significantly narrowing the scope of analysis and reducing the overall computational load.

[0027] Traditional methods for detecting scraper wear and corrosion often rely on single visual features for initial screening. Their core drawback lies in the difficulty of distinguishing between ink residue interference that is highly similar in macroscopic shape, leading to a high false positive rate. To address this issue, this application proposes a fusion discrimination strategy based on edge-shape dual geometric features. The core principle is that the physical causes of the two types of defects are fundamentally different: wear and corrosion, as damage due to missing substrate material, exhibits high degree of edge breakage and irregular shape; while ink residue, as surface adhesion, generally shows continuous, smooth edges and relatively regular shapes. By co-quantifying edge continuity and shape regularity, a discrimination index based on dual geometric features is constructed. This allows for the elimination of typical ink interference areas using geometric morphological features, significantly reducing the data volume required for subsequent processing.

[0028] Specifically, connected component analysis is performed on the edge image to obtain the different connected components and edges of the different connected components. A connected component is denoted as E. j Let E represent the j-th connected component, and traverse the connected component E.j The number of edge pixels in the original edge of the connected component is counted.

[0029] Furthermore, a morphological closing operation is performed on the edges of a connected component to obtain the morphologically connected edges of that connected component. The region enclosed by these morphologically connected edges is the region to be detected for that connected component. This allows us to obtain the edges of each connected component, the region to be detected for each connected component, and the edges of the region to be detected for each connected component, enabling subsequent edge continuity analysis and region shape analysis.

[0030] Step S2: Obtain edge continuity feature values ​​for each region to be detected based on the edges corresponding to each connected component and the edges of the regions to be detected corresponding to each connected component; obtain shape regularity feature values ​​based on the area of ​​the region to be detected and the area of ​​the convex hull of the region to be detected; and filter all regions to be detected based on the edge continuity feature values ​​and the shape regularity feature values ​​to obtain candidate regions.

[0031] Step S1 above obtains the edge of each connected component, the detection area corresponding to each connected component, and the edge of the detection area corresponding to each connected component. This step analyzes the edge pixel situation before and after the closing operation of the connected component and the shape of the detection area obtained after the closing operation.

[0032] Edge continuity feature values ​​for each detection region are obtained based on the edges corresponding to each connected component and the edges of the detection regions corresponding to each connected component. Specifically, the difference between the number of edge pixels on the edge of a connected component and the number of edge pixels on the edge of the detection region corresponding to that connected component is obtained, and this difference is compared with the number of edge pixels on the edge of the detection region corresponding to that connected component to obtain the edge continuity feature value of that detection region.

[0033] The specific calculation model for edge continuity eigenvalues ​​is as follows: , in, This represents the edge continuity feature value of the region to be detected corresponding to the j-th connected component; This represents the number of edge pixels on the edge of the region to be detected corresponding to the j-th connected component, which is also the number of edge pixels after morphological connection of the edge of the j-th connected component. The numerator represents the number of edge pixels on the edge of the j-th connected component. This reflects the number of pixels filling the edge breakpoints during the closing operation; a continuous, smooth edge will not show a significant change in the number of pixels after morphological processing; however, a highly fragmented edge requires a large number of pixels to connect it into a continuous whole. Ultimately... The closer the value is to 1, the worse the continuity of the edge of the area to be detected, and the more it matches the edge characteristics of the wear and corrosion area.

[0034] Next, the shape features of the region to be detected are analyzed, and the shape regularity feature value is obtained based on the area of ​​the region to be detected and the area of ​​its convex hull. Specifically, the area of ​​a region to be detected is compared with the area of ​​its convex hull to obtain the area ratio; the difference between a first preset value and the area ratio is taken as the shape regularity feature value of the region to be detected.

[0035] The specific calculation model for the shape regularity feature value is as follows: , in, This represents the shape regularity feature value of the region to be detected corresponding to the j-th connected component; This represents the area of ​​the region to be detected corresponding to the j-th connected component. This represents the area of ​​the convex hull of the region to be detected corresponding to the j-th connected component; This reflects the degree of fit between the actual shape of the area to be detected and the shape of the convex hull. The core idea is that a regularly shaped, full area has an actual area almost equal to the area of ​​its outermost convex hull; while an irregularly shaped area with depressions or defects will have an actual area significantly smaller than the convex hull area. Ultimately... The closer the value is to 0, the more regular the shape of the area to be detected (the area obtained after the edge closing operation of the connected domain) is, and the more it matches the shape characteristics of the ink residue area.

[0036] It should be noted that the schematic diagram of the region to be detected and its convex hull is as follows: Figure 2 As shown, Figure 2 The left side is a schematic diagram of the area to be detected, and the right side is a schematic diagram of the convex hull that encloses the area to be detected. The acquisition of the convex hull is a well-known technique and will not be described in detail here.

[0037] Finally, the edge continuity feature value and shape regularity feature value of the region to be detected need to be fused to obtain a comprehensive geometric morphology index. Specifically, the normalized value of the edge continuity feature value and the normalized value of the shape regularity feature value of the region to be detected are weighted and summed to obtain the comprehensive geometric morphology index of the region to be detected.

[0038] The specific calculation model for the geometric morphology comprehensive index is as follows: , Wherein, Norm[·] represents the maximum-minimum normalization function, which constrains the calculation result to the interval [0,1] to eliminate dimensional differences. α is a weighting parameter used to measure the contribution of edge continuity and shape regularity. In order to match the physical differences of scraper defects, the contributions of edge and shape features should be balanced to ensure stable discrimination capability for defects and interference of different shapes. Therefore, this application gives a reference value of α=0.5, which can be adjusted according to the needs of the scenario in actual application. and These represent the edge continuity feature value and shape regularity feature value of a region to be detected, respectively; geometric morphology comprehensive index. The value range is [0,1]. Its value directly reflects the possibility that the area belongs to wear and corrosion. When the value is close to 0, it represents an area with continuous edges and regular shape, which is consistent with the typical characteristics of ink residue. When the value is close to 1, it represents an area with broken edges and irregular shape, which matches the geometric properties of wear and corrosion.

[0039] In summary, a first reference threshold T1 is set to 0.4, which can be adjusted as needed: when the comprehensive geometric morphology index of a region to be detected is greater than or equal to the first reference threshold, the region is determined to meet the characteristics of wear and corrosion and is retained as a candidate region; when the comprehensive geometric morphology index is less than the first reference threshold, it is determined to be an area with ink residue interference and is excluded. A sequential traversal strategy is used to process each region to be detected in turn, and the discrimination is performed iteratively until all regions have been processed, ultimately obtaining a set of candidate regions. This significantly reduces the scale of subsequent processing data and provides high-quality input for accurate classification.

[0040] Step S3: Determine the main radiation direction with the center point of a candidate region as the origin, and take pixels in sequence along the main radiation direction to form a pixel sequence corresponding to the main radiation direction; obtain the texture periodicity quantization index of the candidate region based on the distance of each pixel from the origin and the gray value of each pixel in the pixel sequence corresponding to each main radiation direction of the candidate region; filter all candidate regions based on the texture periodicity quantization index to obtain the wear and corrosion region.

[0041] Step S2 constructs geometric morphology indicators based on the differences in edge continuity and shape regularity to initially screen the area to be detected and obtain candidate areas. For the candidate areas, it is necessary to further introduce micro-texture analysis. By evaluating the consistency characteristics of radial texture distribution, the regularity of texture diffusion from the center to the periphery is quantified to accurately capture the subtle differences between wear corrosion and ink residue. Based on this, a discriminant function is constructed to provide a quantitative basis for accurately distinguishing the two types of areas.

[0042] Among the candidate regions selected through macroscopic geometric feature screening, some persistent ink interferences still exist, exhibiting edge and shape characteristics highly similar to actual wear and corrosion. Therefore, the core of this step is to introduce a consistency analysis of the microscopic texture structure: since wear and corrosion is a defective damage to the substrate material, the exposed material texture usually exhibits a clear directionality, with linear wear textures often distributed along the material's crystal orientation or the direction of stress within the region; while ink residue, as a surface adherence, has a random film formation or accumulation process, resulting in a uniformly disordered texture distribution. This step aims to effectively capture the essential differences in the microscopic texture structure of these two types of defects by quantifying the texture change patterns radiating from the region's centroid to the surrounding area, thereby constructing a discriminant function and providing a reliable quantitative basis for the final accurate classification of wear and corrosion and ink residue.

[0043] Therefore, the geometric centroid algorithm is used to determine the center point of the candidate region, and its coordinates are denoted as (x0, y0). Further, in a Cartesian coordinate system with (x0, y0) as the origin, eight radiation directions can be defined based on eight neighborhoods. However, due to the axial overlap of the direction vectors (e.g., the 0° and 180° directions are collinear), these directions are not all independent. Therefore, the four directions of 0°, 45°, 90°, and 135° are determined as the principal axis radiation directions. This set of directions is linearly independent on the plane and can effectively characterize its neighborhood structure, such as... Figure 3 The red arrows indicate the four main radiation directions, and the dashed lines in the diagram represent the linear wear texture within the region. The core challenge of the scene is to distinguish between the directional, patterned, periodic texture of wear corrosion and the disordered, aperiodic texture of ink residue. The periodic texture of wear corrosion can appear in any linear direction within the plane. The four directions of 0°, 45°, 90°, and 135° satisfy "linear independence" on the plane, meaning that the periodic features in any direction cannot be derived from other directions. This completely covers all possible linear periodic directions within the plane, ensuring that regardless of which main direction the periodic texture of wear corrosion is distributed along, it can be captured by the corresponding main axis direction. This prevents missed periodic textures due to missing directions and avoids misjudgments of periodicity caused by incomplete direction coverage.

[0044] Furthermore, starting from the center point of a candidate region, pixels are collected along a main radiation direction until the boundary of the candidate region is reached. The collected pixels are then arranged into a sequence according to the collection order, which serves as the pixel sequence corresponding to the main radiation direction. In this way, the pixel sequence corresponding to each main radiation direction of each candidate region can be obtained, ensuring that the texture information in each direction is completely covered.

[0045] Next, based on the distance of each pixel from the origin and the gray value of each pixel in the pixel sequence corresponding to each main radiation direction of a candidate region, the texture periodicity quantization index of the candidate region is obtained.

[0046] Specifically, the distance of each pixel in the pixel sequence corresponding to a main radiation direction of a candidate region from the origin is calculated, and a scatter plot is constructed with the distance of each pixel from the origin as the horizontal axis and the gray value of each pixel as the vertical axis. Adjacent points in the scatter plot are connected to obtain a connected image. The connected image is smoothed to obtain a smoothed image. The extreme points in the smoothed image are selected in order of the horizontal axis to form an extreme point sequence corresponding to the main radiation direction. The difference of the horizontal coordinates of every two adjacent extreme points in the extreme point sequence is calculated, and the average is calculated to obtain the texture periodicity feature value of the main radiation direction. The difference of the horizontal coordinates of every two adjacent extreme points is the difference of the horizontal coordinates of the latter extreme point and the former extreme point in every two adjacent extreme points. The main radiation direction with the largest texture periodicity feature value in the candidate region is removed, and the average of the texture periodicity feature values ​​of other main radiation directions is used as the texture periodicity quantification index of the candidate region.

[0047] The image obtained by connecting adjacent points in the scatter plot exhibits significant high-frequency sawtooth fluctuations due to factors such as imaging noise and microscopic inhomogeneities in the material. To visually reflect the overall trend of grayscale changes, Gaussian filtering is used to smooth the original data in the connected image, suppressing high-frequency noise and highlighting macroscopic variation patterns, resulting in a smoothed image. Furthermore, the distance from each pixel to the origin is the Euclidean distance.

[0048] The specific calculation model for the texture periodicity feature value of a main radiation direction is as follows: , in, Let M represent the texture periodic feature value of the d-th main radiation direction of a candidate region, and M represent the number of extreme points in the corresponding extreme point sequence of the main radiation direction. and These represent the x-coordinates of the (i+1)th and ith extreme points in the extreme point sequence, respectively, or the distances of the pixels corresponding to these two extreme points from the origin. It can represent the average interval of the periodic changes in grayscale, which is the interval between two extreme points. Its core logic is to distinguish the essential difference between the microscopic texture of wear and corrosion and ink residue by quantifying the pattern of texture recurrence.

[0049] The texture periodicity feature values ​​of the four main radiation directions are sorted, and the texture periodicity feature value with the largest value is identified and removed. This largest texture periodicity feature value usually corresponds to the main radiation direction parallel to the main direction of the linear texture of wear and corrosion, for example... Figure 3When the linear wear texture represented by the dashed line within the region is parallel to the 0° main radiation direction, its texture periodicity feature value will be significantly larger due to the consistency of the texture direction. Such components cannot effectively represent the true distribution law of the texture within the candidate region, and therefore need to be eliminated. After elimination, the average of the texture periodicity feature values ​​of the remaining three main radiation directions of a candidate region can be used to obtain the texture periodicity quantification index of the candidate region. That is, the texture periodicity feature values ​​of the effective main radiation directions are integrated into a single quantification index. In the effective main radiation direction, wear corrosion has strong periodicity of pixel sequence grayscale change and concentrated periodic values, resulting in a small arithmetic mean of texture periodicity feature values. Ink residue has no fixed period, and the periodic values ​​are randomly dispersed in the effective main radiation direction, resulting in a large arithmetic mean of texture periodicity feature values.

[0050] This allows us to obtain the texture periodicity quantification index for each candidate region. The value range of the texture periodicity quantification index is (0,1], and its value comprehensively quantifies the periodic distribution law of the texture in the candidate region: the wear and corrosion region has a strong periodicity of gray-level sequence along the vertical direction of wear due to the parallel distribution of internal linear wear texture; the ink residue region has a generally higher value of texture periodicity quantification index because the film formation process is random and the texture has no fixed period. In summary, setting the second reference threshold T2=0.5, which is in the middle of the value distribution, maximizes the discriminative power of micro-texture features while balancing the risk of false negatives and false negatives. In practical applications, it can be dynamically fine-tuned according to specific imaging conditions and detection requirements.

[0051] Therefore, all candidate regions are filtered based on the texture periodicity quantization index to obtain the wear and corrosion regions. Specifically, candidate regions with a texture periodicity quantization index less than or equal to the second reference threshold are recorded as wear and corrosion regions. Candidate regions with a texture periodicity quantization index equal to the second reference threshold are identified as ink residue interference regions and are removed. The verified wear and corrosion defect regions are marked and the detection results are output, completing the full online detection process for extracting scraper defects from the scraper surface image.

[0052] In summary, this application proposes a processing idea and method based on "macro-geometric screening - micro-texture subdivision" by combining image processing related algorithms. By deeply exploring the essential differences between wear and corrosion and ink in macro-geometric morphology and micro-texture features, it specifically solves the problems and shortcomings of traditional methods for detecting wear and corrosion on the surface of the scraper, and achieves efficient and accurate defect detection and processing.

[0053] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0054] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0055] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An online detection method for damaged printing squeegees based on machine vision, characterized in that, The method includes: The surface image of the scraper is acquired, and edge detection is performed on the surface image to obtain the edge image; connected component analysis is performed on the edge image to obtain different connected components and the edges corresponding to each connected component; closing operation is performed on the edges corresponding to each connected component to obtain the detection area corresponding to each connected component; Edge continuity feature values ​​of each region to be detected are obtained based on the edges corresponding to each connected component and the edges of the regions to be detected corresponding to each connected component; shape regularity feature values ​​are obtained based on the area of ​​the region to be detected and the area of ​​the convex hull of the region to be detected; candidate regions are obtained by filtering all regions to be detected based on the edge continuity feature values ​​and the shape regularity feature values. The main radiation direction is determined by taking the center point of a candidate region as the origin. Pixels are taken in sequence along the main radiation direction to form the pixel sequence corresponding to the main radiation direction. The texture periodicity quantization index of the candidate region is obtained based on the distance of each pixel from the origin and the gray value of each pixel in the pixel sequence corresponding to each main radiation direction of the candidate region. The wear and corrosion region is obtained by screening all candidate regions based on the texture periodicity quantization index.

2. The online detection method for damage to printing squeegees based on machine vision according to claim 1, characterized in that, The step of performing a closing operation on the edges corresponding to each connected component to obtain the region to be detected for each connected component includes: A morphological closing operation is performed on the edges of a connected component to obtain the morphologically connected edges of that connected component. The region enclosed by the morphologically connected edges is the region to be detected corresponding to that connected component.

3. The online detection method for damage to printing squeegees based on machine vision according to claim 1, characterized in that, The step of obtaining edge continuity feature values ​​for each region to be detected based on the edges corresponding to each connected component and the edges of the regions to be detected corresponding to each connected component includes: The difference between the number of edge pixels on the edge of a connected component and the number of edge pixels on the edge of the corresponding region to be detected is obtained, and compared with the number of edge pixels on the edge of the corresponding region to be detected to obtain the edge continuity feature value of the region to be detected.

4. The online detection method for damage to printing squeegees based on machine vision according to claim 1, characterized in that, The method of obtaining shape regularity feature values ​​based on the area of ​​the region to be detected and the area of ​​the convex hull of the region to be detected includes: The area ratio is obtained by comparing the area of ​​a region to be detected with the area of ​​the convex hull of the region to be detected; the difference between the first preset value and the area ratio is used as the shape regularity feature value of the region to be detected.

5. The online detection method for damage to printing squeegees based on machine vision according to claim 1, characterized in that, The process of filtering all regions to be detected based on edge continuity feature values ​​and shape regularity feature values ​​to obtain candidate regions includes: The geometric morphology comprehensive index of a region to be detected is obtained by weighted summing of the normalized values ​​of the edge continuity feature value and the normalized values ​​of the shape regularity feature value. If the geometric morphology comprehensive index of a region to be detected is greater than or equal to the first reference threshold, then the region to be detected is a candidate region.

6. The online detection method for damage to printing squeegees based on machine vision according to claim 1, characterized in that, The step of determining the main radiation direction by taking the center point of a candidate region as the origin includes: In a Cartesian coordinate system with the center point of a candidate region as the origin, the four directions of 0°, 45°, 90° and 135° are determined as the main axis radiation directions.

7. The online detection method for damage to printing squeegees based on machine vision according to claim 1, characterized in that, The texture periodicity quantization index of a candidate region is obtained based on the distance of each pixel from the origin and the gray value of each pixel in the pixel sequence corresponding to each main radiation direction of the candidate region, including: Calculate the distance of each pixel in the pixel sequence corresponding to a main radiation direction of a candidate region from the origin, and construct a scatter plot with the distance of each pixel from the origin as the horizontal axis and the gray value of each pixel as the vertical axis. Connect adjacent points in the scatter plot to obtain a connected image. Smooth the connected image to obtain a smoothed image. Take the extreme points in the smoothed image in order of the horizontal axis to form an extreme point sequence corresponding to the main radiation direction. Calculate the difference of the horizontal coordinates of every two adjacent extreme points in the extreme point sequence and average them to obtain the texture periodicity feature value of the main radiation direction. The difference of the horizontal coordinates of every two adjacent extreme points is the difference of the horizontal coordinates of the latter extreme point and the former extreme point in every two adjacent extreme points. Remove the main radiation direction with the largest texture periodicity feature value in the candidate region, and calculate the average of the texture periodicity feature values ​​of other main radiation directions as the texture periodicity quantification index of the candidate region.

8. The online detection method for damage to printing squeegees based on machine vision according to claim 1, characterized in that, The wear and corrosion regions are obtained by filtering all candidate regions based on texture periodicity quantization indicators, including: Candidate regions whose texture periodicity quantization index is less than or equal to the second reference threshold are denoted as wear and corrosion regions.

Images