A method and device for detecting cracks in fiberglass board

Abstract

The present invention relates to the technical field of material testing and analysis, and in particular to a glass fiber board crack detection method and equipment, which specifically involves collecting a visible light surface image of a glass fiber board to be detected by optical means, and then obtaining a pre-processed visible light surface image, testing or analyzing the material based on the pre-processed visible light surface image, and testing each crack defect existing in the pre-processed visible light surface image, thereby determining the crack information of the glass fiber board to be detected, the crack information including whether there is a crack area in the glass fiber board to be detected and each initial crack area. The present invention determines the crack information of the glass fiber board to be detected by performing material testing and analysis on the visible light surface image of the glass fiber board to be detected, thereby effectively improving the detection accuracy of the glass fiber board crack detection, and can be used to realize new material detection, measurement, related standardization, certification and accreditation services, etc.

Description

Technical Field

[0001] The invention relates to the technical field of material testing and analysis, and in particular to a method and equipment for detecting cracks in a fiberglass board. Background Art

[0002] Fiberglass board is also called fiberglass insulation board, fiberglass board, fiberglass composite board, etc. It is made of fiberglass material and high heat-resistant composite material. It does not contain asbestos components that are harmful to the human body. It has high mechanical and dielectric properties, good heat resistance and moisture resistance, and good processability. Fiberglass board can not only be used for plastic molds, injection molds, machinery manufacturing, molding machines, drilling machines, injection molding machines, motors, table grinding pads, etc., but also can be used for air-conditioned room walls, ventilation channels, trench walls as insulation materials, and broadcast recording studio double walls as sound insulation and insulation materials. When cracks appear on the surface of fiberglass board, the sound insulation and heat preservation effects of fiberglass board will be reduced. Therefore, detecting cracks on the surface of fiberglass board is a key link to ensure the use effect of fiberglass board.

[0003] The patent document with the publication number CN114359195A in the prior art discloses a glass curtain wall crack detection method, which specifically collects a glass curtain wall image and converts it into a grayscale image. Based on the improved ULBP (Uniform Local Binary Pattern) descriptor, the first feature of the grayscale image is calculated, and the first feature of the grayscale image is reduced in dimension to obtain a macro feature. The macro feature is input into a classifier, and the classifier is used to detect whether there is a crack in the glass curtain wall image. The ULBP descriptor in the detection method cannot accurately detect crack pixels with shallow depth and weak intensity, which easily causes false crack detection. In addition, the detection method uses a classifier to identify cracks. The process of classifier identifying cracks is relatively complicated and requires a large number of data samples. In addition, the classifier has high training accuracy requirements for identifying cracks, and it is difficult to accurately identify weak cracks with unclear image features. Therefore, the crack detection accuracy of the detection method is low. Summary of the invention

[0004] In order to solve the above-mentioned problem of low accuracy of existing crack detection, the purpose of the present invention is to provide a method and device for detecting cracks in a fiberglass board.

[0005] The present invention provides a method for detecting cracks in a glass fiber board, comprising the following steps:

[0006] Acquire a visible light surface image of the glass fiber board to be inspected, and preprocess the visible light surface image to obtain a first eigenvalue and a second eigenvalue corresponding to each pixel point in the preprocessed visible light surface image;

[0007] Determine, according to the first eigenvalue and the second eigenvalue corresponding to each pixel point, a linear eigenvalue corresponding to each pixel point in the preprocessed visible light surface image;

[0008] Determine an optimization factor corresponding to each pixel point in the preprocessed visible light surface image according to the first eigenvalue, the second eigenvalue and the gray value corresponding to each pixel point;

[0009] Determine the linearity determination index value corresponding to each pixel point in the preprocessed visible light surface image according to the first eigenvalue, the second eigenvalue, the linearity eigenvalue and the optimization factor corresponding to each pixel point;

[0010] According to the linear judgment index value corresponding to each pixel point in the preprocessed visible light surface image, the feature enhanced significant image of the glass fiber board to be detected is determined, and then the crack information of the glass fiber board to be detected is determined, and the crack information includes: whether there is a crack area in the glass fiber board to be detected and each initial crack area of ​​the glass fiber board to be detected.

[0011] Furthermore, it also includes:

[0012] Obtaining first related pixel points and second related pixel points corresponding to each crack edge pixel point in each initial crack region of the glass fiber board to be detected, and determining a crack determination coefficient corresponding to each crack edge pixel point in each initial crack region according to the linear determination index values ​​corresponding to each crack edge pixel point and the first related pixel point and the second related pixel point corresponding to each crack edge pixel point in each initial crack region;

[0013] According to the crack determination coefficient corresponding to each crack edge pixel point in each initial crack area, each final crack area of ​​the glass fiber board to be detected is determined.

[0014] Furthermore, the calculation formula for determining the linear feature value corresponding to each pixel point in the preprocessed visible light surface image is:

[0015]

[0016] in, is the linear eigenvalue corresponding to the i-th pixel point in the preprocessed visible light surface image, λ1(i) is the first eigenvalue corresponding to the i-th pixel point in the preprocessed visible light surface image, and λ2(i) is the second eigenvalue corresponding to the i-th pixel point in the preprocessed visible light surface image.

[0017] Furthermore, the calculation formula for determining the optimization factor corresponding to each pixel in the preprocessed visible light surface image is:

[0018]

[0019] Among them, τ i is the optimization factor corresponding to the i-th pixel in the preprocessed visible light surface image, c i is the grayscale variance value of the i-th pixel and its eight neighboring pixels in the preprocessed visible light surface image, λ1(i) is the first eigenvalue corresponding to the i-th pixel in the preprocessed visible light surface image, λ2(i) is the second eigenvalue corresponding to the i-th pixel in the preprocessed visible light surface image, and k is a hyperparameter.

[0020] Furthermore, the calculation formula for determining the linear determination index value corresponding to each pixel point in the preprocessed visible light surface image is:

[0021]

[0022] Among them, γ i is the linearity judgment index value corresponding to the i-th pixel point in the preprocessed visible light surface image, λ1(i) is the first eigenvalue corresponding to the i-th pixel point in the preprocessed visible light surface image, λ2(i) is the second eigenvalue corresponding to the i-th pixel point in the preprocessed visible light surface image, τ i is the optimization factor corresponding to the i-th pixel in the preprocessed visible light surface image, is the linear feature value corresponding to the i-th pixel in the preprocessed visible light surface image, is the preset linear feature value threshold.

[0023] Furthermore, the step of determining the crack information of the glass fiber board to be detected includes:

[0024] According to the position of each pixel point in the feature enhanced significant image of the glass fiber board to be detected and the linear determination index value corresponding to each pixel point, each pixel point in the feature enhanced significant image is clustered to obtain M clustering regions of the feature enhanced significant image;

[0025] If the number M of clustered regions of the feature enhanced significant image is equal to 1, it is determined that there is no crack region in the glass fiber board to be detected. If the number M of clustered regions of the feature enhanced significant image is greater than 1, it is determined that there is a crack region in the glass fiber board to be detected, and according to the mean of the linear judgment index values ​​corresponding to the M clustered regions of the feature enhanced significant image, the clustered region corresponding to the mean of the larger M-1 linear judgment index values ​​is selected from the M clustered regions, and the M-1 clustered regions are used as the initial crack regions of the glass fiber board to be detected.

[0026] Further, the step of determining each final crack area of ​​the glass fiber board to be tested includes:

[0027] Determining each first true crack edge pixel point and each pseudo crack edge pixel point in each initial crack region according to a crack determination coefficient corresponding to each crack edge pixel point in each initial crack region;

[0028] Obtain N first related pixel points and N second related pixel points corresponding to each pseudo crack edge pixel point in each initial crack region, and according to the crack determination coefficients corresponding to each pseudo crack edge pixel point and its corresponding N first related pixel points and N second related pixel points in each initial crack region, select the pixel point corresponding to the maximum crack determination coefficient from each pseudo crack edge pixel point and the N first related pixel points and N second related pixel points corresponding to each pseudo crack edge pixel point, and use the pixel point corresponding to the maximum crack determination coefficient as the second true crack edge pixel point;

[0029] According to each first true crack edge pixel point and each second true crack edge pixel point in each initial crack region, each final crack region of the glass fiber board to be detected is determined.

[0030] Furthermore, the calculation formula for determining the crack determination coefficient corresponding to each crack edge pixel point in each initial crack area is:

[0031]

[0032] Among them, ω i is the crack determination coefficient corresponding to the i-th crack edge pixel point in each initial crack area, γ i,a is the linear determination index value corresponding to the first related pixel point corresponding to the i-th crack edge pixel point in each initial crack area, γ i,b is the linear determination index value corresponding to the second related pixel point corresponding to the i-th crack edge pixel point in each initial crack area, γ i is the linear judgment index value corresponding to the i-th crack edge pixel point in each initial crack area.

[0033] Furthermore, the step of obtaining first related pixel points and second related pixel points corresponding to each crack edge pixel point of each initial crack area includes:

[0034] Determine the tangent line of each crack edge pixel point of each initial crack region according to each crack edge pixel point of each initial crack region;

[0035] According to the tangent lines of each crack edge pixel point of each initial crack area, one adjacent pixel point is taken from each side of the tangent line of each crack edge pixel point in the vertical direction, and one of the two adjacent pixel points is taken as the first related pixel point, and the other adjacent pixel point of the two adjacent pixel points is taken as the second related pixel point.

[0036] The present invention also provides a fiberglass board crack detection device, comprising a processor and a memory, wherein the processor is used to process instructions stored in the memory to implement a fiberglass board crack detection method.

[0037] The present invention has the following beneficial effects:

[0038] The present invention provides a glass fiber board crack detection method and equipment, which can be used to realize new material detection, measurement and related standardization, etc. Specifically, a visible light surface image of a glass fiber board to be detected is obtained through visible light, and the visible light surface image is subjected to material testing and analysis. Specifically, according to the visible light surface image of the glass fiber board to be detected, the first eigenvalue and the second eigenvalue corresponding to each pixel point in the preprocessed visible light surface image are determined, and then the linear eigenvalue and optimization factor corresponding to each pixel point are determined, so as to determine the linear judgment index value corresponding to each pixel point. According to the linear judgment index value corresponding to each pixel point, a feature enhanced significant image of the glass fiber board to be detected is obtained, and then the crack information of the glass fiber board to be detected is determined. If there are cracks in the glass fiber board to be detected, each initial crack area in the glass fiber board to be detected is corrected, so as to obtain each final accurate crack area in the glass fiber board to be detected.

[0039] The present invention mainly tests the existence of crack defects in the surface of the glass fiber board through material testing and analysis technology, can accurately determine the crack area in the surface of the glass fiber board, solves the problem of inaccurate glass crack detection, improves the accuracy of glass fiber board crack detection, and can be used to realize new material detection, measurement, etc. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0041] Figure 1 The figure is a flow chart of a glass fiber board crack detection method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0042] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation methods, structures, features and effects of the technical solutions proposed by the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures, or characteristics in one or more embodiments may be combined in any suitable form.

[0043] Unless defined otherwise, 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 belongs.

[0044] This embodiment provides a method for detecting cracks in a glass fiber board. Figure 1 As shown, the method comprises the following steps:

[0045] (1) Obtaining a visible light surface image of the glass fiber board to be inspected, and preprocessing the visible light surface image, thereby obtaining a first eigenvalue and a second eigenvalue corresponding to each pixel point in the preprocessed visible light surface image.

[0046] (1-1) Obtain a visible light surface image of the glass fiber board to be inspected.

[0047] In this embodiment, according to the glass fiber board to be tested, the glass fiber board to be tested placed on the testing table is photographed by using an image acquisition device, so as to obtain a visible light surface image of the glass fiber board to be tested, and the visible light surface image can be used to perform crack defect detection on the glass fiber board to be tested. The image acquisition device includes a camera, a light source, a testing table, etc., and the implementer can set them according to the actual situation. In this embodiment, the camera head is placed directly above the glass fiber board to be tested, and the front view image of the surface of the glass fiber board to be tested can be collected through the camera's top-down angle.

[0048] (1-2) Based on the visible light surface image of the glass fiber board to be inspected, the visible light surface image is preprocessed.

[0049] In order to eliminate the influence of noise in the visible light surface image of the glass fiber board to be detected on the visible light surface image, this embodiment performs denoising on the visible light surface image by using a median filtering algorithm. At the same time, in order to achieve brightness equalization of the visible light surface image of the glass fiber board to be detected, this embodiment uses histogram equalization to equalize the visible light surface image. As a result, this embodiment obtains a pre-processed visible light surface image of the glass fiber board to be detected, and the pre-processed visible light surface image of the glass fiber board to be detected can be used for subsequent crack detection of the glass fiber board to be detected. The process of processing the visible light surface image by using a median filtering algorithm and histogram equalization is a prior art, which is not within the protection scope of the present invention and will not be elaborated in detail here.

[0050] (1-3) According to the preprocessed visible light surface image, a first eigenvalue and a second eigenvalue corresponding to each pixel point in the preprocessed visible light surface image are obtained.

[0051] It should be noted that, considering that the crack area on the surface of the fiberglass board has a low contrast feature, this embodiment will determine the first eigenvalue and the second eigenvalue corresponding to each pixel point in the preprocessed visible light surface image, which will be used to subsequently construct a crack significance model, that is, to subsequently determine the linear judgment index value. The linear judgment index value can highlight the crack defects on the surface of the fiberglass board and enhance the contrast of the crack area on the surface of the fiberglass board. There are many methods for determining the eigenvalue of a pixel point, which will not be described one by one here. This embodiment determines the eigenvalue of a pixel point through a two-dimensional Gaussian model, and the determination steps include:

[0052] (1-3-1) In this embodiment, a two-dimensional Gaussian model is first established. The two-dimensional Gaussian model is: Wherein, f is a two-dimensional Gaussian model, σ is a scale factor of the two-dimensional Gaussian model, x is a pixel abscissa of a pixel point in the two-dimensional Gaussian model, and y is a pixel ordinate of a pixel point in the two-dimensional Gaussian model. The process of establishing a two-dimensional Gaussian model is prior art and is not within the scope of protection of the present invention, and will not be elaborated on in detail herein.

[0053] (1-3-2) Then, the convolution process is performed on each pixel point in the preprocessed visible light surface image through the established two-dimensional Gaussian model, so as to obtain the Hessian matrix corresponding to each pixel point in the preprocessed visible light surface image. The Hessian matrix is: Among them, v is the Hessian matrix corresponding to each pixel in the preprocessed visible light surface image, and f xx 、f xy 、f yx 、f yyare the second-order partial derivatives of each pixel in the preprocessed visible light surface image in each direction, that is, the convolution response of the two-dimensional Gaussian model at each pixel in the preprocessed visible light surface image, where, σ is the scale factor of the two-dimensional Gaussian model, x is the pixel abscissa of the pixel point in the preprocessed visible light surface image, and y is the pixel ordinate of the pixel point in the preprocessed visible light surface image.

[0054] (1-3-3) Finally, according to the second-order partial derivatives of each pixel in the preprocessed visible light surface image in each direction, that is, f xx 、f xy 、f yx 、f yy , obtain two eigenvalues ​​corresponding to each pixel point in the preprocessed visible light surface image. In this embodiment, one of the two eigenvalues ​​is called the first eigenvalue, denoted as λ1. The other eigenvalue of the two eigenvalues ​​is called the second eigenvalue, denoted by λ2. in,

[0055] (2) Determine the linear eigenvalue corresponding to each pixel point in the preprocessed visible light surface image according to the first eigenvalue and the second eigenvalue corresponding to each pixel point.

[0056] It should be noted that, according to prior knowledge, the structure of the crack area on the surface of the glass fiber board is mostly linear structures of different shapes. This embodiment analyzes the pixels in the pre-processed visible light surface image according to the feature information of the pixels in the crack area on the surface of the glass fiber board, that is, according to the feature information of the pixels in the linear structure, so as to facilitate the subsequent enhancement of the feature information of the crack area on the surface of the glass fiber board to be detected, that is, to facilitate the subsequent determination of the feature-significant image corresponding to the glass fiber board to be detected.

[0057] This embodiment calculates the linear eigenvalues ​​corresponding to each pixel in the preprocessed visible light surface image based on the prior knowledge of the eigenvalues ​​of the pixel points. Taking the calculation of the linear eigenvalues ​​corresponding to the i-th pixel in the preprocessed visible light surface image as an example, the linear eigenvalue corresponding to the i-th pixel in the preprocessed visible light surface image is determined according to the first eigenvalue and the second eigenvalue corresponding to the i-th pixel point. The linear eigenvalue refers to the possibility that the pixel point is not a pixel point in the linear structure, that is, the possibility that the pixel point is not a pixel point in the crack area. The calculation formula is:

[0058]

[0059] in, is the linear eigenvalue corresponding to the i-th pixel point in the preprocessed visible light surface image, λ1(i) is the first eigenvalue corresponding to the i-th pixel point in the preprocessed visible light surface image, and λ2(i) is the second eigenvalue corresponding to the i-th pixel point in the preprocessed visible light surface image.

[0060] Referring to the calculation process of the linear eigenvalue corresponding to the i-th pixel in the preprocessed visible light surface image, the linear eigenvalue corresponding to each pixel in the preprocessed visible light surface image can be obtained. It should be noted that when the first eigenvalue λ1 corresponding to a certain pixel is larger and the second eigenvalue λ2 corresponding to the pixel is smaller, the linear eigenvalue corresponding to the pixel is The smaller it is, the greater the possibility that the pixel is a pixel in the crack area on the surface of the fiberglass board.

[0061] (3) Determine the optimization factor corresponding to each pixel point in the preprocessed visible light surface image according to the first eigenvalue, the second eigenvalue and the gray value corresponding to each pixel point.

[0062] It should be noted that, for the surface image of the glass fiber board without crack area, the grayscale distribution of each pixel in the surface image is relatively uniform. Since the grayscale distribution of each pixel in the surface image is relatively uniform, the curvature change of each pixel in each direction is small. Since the curvature change of each pixel in each direction is small, the element value in the Hessian matrix corresponding to each pixel is small, and then the first eigenvalue and the second eigenvalue corresponding to each pixel are also small. Therefore, this embodiment will optimize the linear determination index value determined later based on the pixel grayscale distribution and change status of each pixel in the preprocessed visible light surface image.

[0063] This embodiment takes the calculation of the optimization factor corresponding to the i-th pixel in the preprocessed visible light surface image as an example, and calculates the optimization factor corresponding to the i-th pixel in the preprocessed visible light surface image according to the first eigenvalue, the second eigenvalue and the gray value corresponding to the i-th pixel. The calculation formula is:

[0064]

[0065] Among them, τ i is the optimization factor corresponding to the i-th pixel in the preprocessed visible light surface image, c i is the grayscale variance value of the i-th pixel and its eight neighboring pixels in the preprocessed visible light surface image, λ1(i) is the first eigenvalue corresponding to the i-th pixel in the preprocessed visible light surface image, λ2(i) is the second eigenvalue corresponding to the i-th pixel in the preprocessed visible light surface image, and k is a hyperparameter.

[0066] The optimization factor corresponding to each pixel in the preprocessed visible light surface image can be obtained by referring to the calculation process of the optimization factor corresponding to the i-th pixel in the preprocessed visible light surface image. It should be noted that the optimization factor can enhance the contrast between the crack area and the background area on the surface of the glass fiber board, that is, it can enhance the crack characteristics of the crack area, and can suppress the influence of the background area on the crack detection of the glass fiber board, thereby improving the accuracy of the crack detection of the glass fiber board.

[0067] (4) Determine the linearity determination index value corresponding to each pixel point in the preprocessed visible light surface image according to the first eigenvalue, the second eigenvalue, the linearity eigenvalue and the optimization factor corresponding to each pixel point.

[0068] In this embodiment, taking the calculation of the linear determination index value corresponding to the i-th pixel point in the preprocessed visible light surface image as an example, the linear determination index value corresponding to the i-th pixel point in the preprocessed visible light surface image is calculated according to the first eigenvalue, the second eigenvalue, the linear eigenvalue and the optimization factor corresponding to the i-th pixel point, and the calculation formula is:

[0069]

[0070] Among them, γ i is the linearity judgment index value corresponding to the i-th pixel point in the preprocessed visible light surface image, λ1(i) is the first eigenvalue corresponding to the i-th pixel point in the preprocessed visible light surface image, λ2(i) is the second eigenvalue corresponding to the i-th pixel point in the preprocessed visible light surface image, τ i is the optimization factor corresponding to the i-th pixel in the preprocessed visible light surface image, is the linear feature value corresponding to the i-th pixel in the preprocessed visible light surface image, The preset linear feature value threshold is set to 0.05 in this embodiment.

[0071] By referring to the linear determination index value corresponding to the i-th pixel point in the preprocessed visible light surface image, the linear determination index value corresponding to each pixel point in the preprocessed visible light surface image can be obtained. The linear determination index value refers to the index value after enhancing the feature information of each pixel point in the preprocessed visible light surface image.

[0072] It should be noted that in order to facilitate the subsequent intuitive detection and identification of the crack area in the glass fiber board to be detected, this embodiment normalizes the linear judgment index value corresponding to each pixel point in the preprocessed visible light surface image to ensure that the linear judgment index value corresponding to each pixel point is in [0,1].

[0073] (5) According to the linear judgment index value corresponding to each pixel point in the preprocessed visible light surface image, the feature enhanced significant image of the glass fiber board to be detected is determined, and then the crack information of the glass fiber board to be detected is determined, and the crack information includes: whether there is a crack area in the glass fiber board to be detected and each initial crack area of ​​the glass fiber board to be detected.

[0074] In this embodiment, first, according to the linear judgment index value corresponding to each pixel point in the preprocessed visible light surface image obtained in step (4), the feature enhanced significant image of the glass fiber board to be detected is determined, and the pixel value of each pixel point of the feature enhanced significant image is the linear judgment index value. The feature enhanced significant image enhances the crack feature information of the crack area in the glass fiber board to be detected, so as to facilitate the subsequent comprehensive and accurate detection of the crack condition on the surface of the glass fiber board to be detected.

[0075] Then, according to the characteristic enhanced significant image of the glass fiber board to be detected, crack information of the glass fiber board to be detected is determined, wherein the crack information includes whether there is a crack area in the glass fiber board to be detected and each initial crack area of ​​the glass fiber board to be detected, and the steps include:

[0076] (5-1) According to the position of each pixel point in the feature enhanced significant image of the glass fiber board to be detected and the linear judgment index value corresponding to each pixel point, each pixel point in the feature enhanced significant image is clustered to obtain M clustering areas of the feature enhanced significant image.

[0077] In this embodiment, according to the position of each pixel point in the feature enhancement significant image of the glass fiber board to be detected and the linear judgment index value corresponding to each pixel point, the DBSCAN algorithm is used to perform cluster analysis on each pixel point in the feature enhancement significant image of the glass fiber board to be detected, so as to obtain M clustering regions of the feature enhancement significant image. The process of clustering the feature enhancement significant image by the DBSCAN algorithm is a prior art, which is not within the protection scope of the present invention and will not be elaborated in detail here.

[0078] (5-2) If the number M of clustered regions of the feature enhanced significant image is equal to 1, it is determined that there is no crack region in the glass fiber board to be detected. If the number M of clustered regions of the feature enhanced significant image is greater than 1, it is determined that there is a crack region in the glass fiber board to be detected, and according to the mean of the linear judgment index values ​​corresponding to the M clustered regions of the feature enhanced significant image, the clustered region corresponding to the mean of the larger M-1 linear judgment index values ​​is selected from the M clustered regions, and the M-1 clustered regions are used as the initial crack regions of the glass fiber board to be detected.

[0079] (6) If crack regions exist in the glass fiber board to be inspected, determine each final crack region of the glass fiber board to be inspected.

[0080] If there is a crack region in the glass fiber board to be detected, this embodiment takes into account that when dividing the crack region of the feature enhanced significant image, the crack region division may be inaccurate for the condition where the crack edge is shallow and the strength is weak. Therefore, this embodiment corrects each initial crack region of the glass fiber board to be detected to improve the accuracy of identifying each crack region. Based on each initial crack region of the glass fiber board to be detected, each final crack region of the glass fiber board to be detected can be determined, and the determination steps include:

[0081] (6-1) Obtain the first related pixel point and the second related pixel point corresponding to each crack edge pixel point in each initial crack area of ​​the glass fiber board to be detected, and determine the crack determination coefficient corresponding to each crack edge pixel point in each initial crack area according to the linear determination index value corresponding to each crack edge pixel point in each initial crack area and its corresponding first related pixel point and second related pixel point.

[0082] It should be noted that, for the real crack area in the feature-significant image, the pixel value of each crack pixel point in the real crack area is greatly different from the pixel value of the pixel point in the background area. Based on this, the crack determination coefficient corresponding to each crack edge pixel point in each initial crack area is determined, and each initial crack area is corrected according to the crack determination coefficient corresponding to each crack edge pixel point in each initial crack area, so as to improve the accuracy of the crack area finally determined. The step of determining the crack determination coefficient corresponding to each crack edge pixel point in each initial crack area includes:

[0083] (6-1-1) The step of obtaining first related pixel points and second related pixel points corresponding to each crack edge pixel point of each initial crack region includes:

[0084] a. According to each crack edge pixel point of each initial crack region, determine the tangent line of each crack edge pixel point of each initial crack region.

[0085] b. According to the tangent lines of each crack edge pixel point of each initial crack area, take one adjacent pixel point on each side of the tangent line of each crack edge pixel point in the vertical direction, take one of the two adjacent pixel points as the first related pixel point, and take the other of the two adjacent pixel points as the second related pixel point.

[0086] (6-1-2) Determine the crack determination coefficient corresponding to each crack edge pixel point in each initial crack area based on the linear determination index values ​​corresponding to each crack edge pixel point and its corresponding first related pixel point and second related pixel point in each initial crack area.

[0087] This embodiment takes the calculation of the crack determination coefficient corresponding to the i-th crack edge pixel point in each initial crack area as an example, and calculates the crack determination coefficient corresponding to the i-th crack edge pixel point in each initial crack area according to the linear determination index value corresponding to the i-th crack edge pixel point in each initial crack area, the first related pixel point corresponding to the i-th crack edge pixel point, and the linear determination index value corresponding to the second related pixel point, and the calculation formula is:

[0088]

[0089] Among them, ω i is the crack determination coefficient corresponding to the i-th crack edge pixel point in each initial crack area, and the value of the determination coefficient is in [0,1), γ i,a is the linear determination index value corresponding to the first related pixel point corresponding to the i-th crack edge pixel point in each initial crack area, γ i,b is the linear determination index value corresponding to the second related pixel point corresponding to the i-th crack edge pixel point in each initial crack area, γ i is the linear judgment index value corresponding to the i-th crack edge pixel point in each initial crack area.

[0090] (6-2) Determining each final crack region of the glass fiber board to be inspected according to the crack determination coefficient corresponding to each crack edge pixel point in each initial crack region, the steps include:

[0091] (6-2-1) According to the crack determination coefficient corresponding to each crack edge pixel point in each initial crack region, each first true crack edge pixel point and each pseudo crack edge pixel point in each initial crack region are determined.

[0092] In this embodiment, if the crack determination coefficient corresponding to a certain crack edge pixel point in each initial crack area is greater than or equal to the preset crack determination coefficient threshold, the crack edge pixel point is determined to be the first true crack edge pixel point; if the crack determination coefficient corresponding to a certain crack edge pixel point in each initial crack area is less than the preset crack determination coefficient threshold, the crack edge pixel point is determined to be a pseudo crack edge pixel point. In this embodiment, the preset crack determination coefficient threshold is set to 0.5.

[0093] It should be noted that, when the crack determination coefficient corresponding to a crack edge pixel point is larger, the possibility that the crack edge pixel point is a true crack edge pixel point is greater; when the crack determination coefficient corresponding to a crack edge pixel point is smaller, the possibility that the crack edge pixel point is a true crack edge pixel point is smaller, and the classification of the crack edge pixel point is inaccurate.

[0094] (6-2-2) Obtain N first related pixel points and N second related pixel points corresponding to each pseudo-crack edge pixel point in each initial crack area, and based on the crack determination coefficients corresponding to each pseudo-crack edge pixel point and its corresponding N first related pixel points and N second related pixel points in each initial crack area, select the pixel point corresponding to the maximum crack determination coefficient from each pseudo-crack edge pixel point and the N first related pixel points and N second related pixel points corresponding to each pseudo-crack edge pixel point, and use the pixel point corresponding to the maximum crack determination coefficient as the second true crack edge pixel point.

[0095] In this embodiment, referring to the process of obtaining the first related pixel point and the second related pixel point corresponding to each crack edge pixel point in step (6-1-1), N first related pixel points and N second related pixel points corresponding to each pseudo crack edge pixel point in each initial crack region can be obtained, which will not be elaborated in detail here. The crack determination coefficients corresponding to each pseudo crack edge pixel point and its corresponding N first related pixel points and N second related pixel points in each initial crack region are sorted in order from small to large, so as to obtain each crack determination coefficient sequence, each pseudo crack edge pixel point has its corresponding crack determination coefficient sequence, and the pixel point corresponding to the maximum crack determination coefficient is selected from the crack determination coefficient sequence corresponding to each pseudo crack edge pixel point, and the pixel point is used as the second true crack edge pixel point, so as to obtain each second true crack edge pixel point in each initial crack region.

[0096] (6-2-3) Determine each final crack region of the glass fiber board to be inspected based on each first true crack edge pixel point and each second true crack edge pixel point in each initial crack region.

[0097] In this embodiment, each initial crack area of ​​the glass fiber board to be detected is corrected according to each first true crack edge pixel point in each initial crack area obtained in step (6-2-1) and each second true crack edge pixel point in each initial crack area obtained in step (6-2-2), so as to obtain each final crack area of ​​the glass fiber board to be detected. At this point, this embodiment realizes the accurate extraction and detection of the surface crack area of ​​the glass fiber board to be detected.

[0098] This embodiment also provides a fiberglass board crack detection device, including a processor and a memory, wherein the processor is used to process instructions stored in the memory to implement a fiberglass board crack detection method.

[0099] The present invention uses an image acquisition device and visible light means to obtain a visible light surface image of a glass fiber board to be detected, and performs material testing and analysis on the visible light surface image, thereby obtaining crack information of the glass fiber board to be detected. The present invention improves the characteristic significance of the crack area in the glass fiber board to be detected, that is, enhances the characteristic information of each pixel point in the crack area, and avoids the influence of the background area in the glass fiber board on crack detection. The present invention improves the detection accuracy of crack detection in glass fiber boards, and can be used for new material detection, measurement, related standardization, certification and accreditation services, etc.

[0100] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for detecting cracks in a glass fiber board, characterized in that: The following steps are involved: Acquire a visible light surface image of the glass fiber board to be inspected, and preprocess the visible light surface image to obtain a first eigenvalue and a second eigenvalue corresponding to each pixel point in the preprocessed visible light surface image; Determine, according to the first eigenvalue and the second eigenvalue corresponding to each pixel point, a linear eigenvalue corresponding to each pixel point in the preprocessed visible light surface image; Determine an optimization factor corresponding to each pixel point in the preprocessed visible light surface image according to the first eigenvalue, the second eigenvalue and the gray value corresponding to each pixel point; Determine the linearity determination index value corresponding to each pixel point in the preprocessed visible light surface image according to the first eigenvalue, the second eigenvalue, the linearity eigenvalue and the optimization factor corresponding to each pixel point; According to the linear judgment index value corresponding to each pixel point in the preprocessed visible light surface image, a feature enhanced significant image of the glass fiber board to be detected is determined, and then crack information of the glass fiber board to be detected is determined, wherein the crack information includes whether there is a crack area in the glass fiber board to be detected and each initial crack area of ​​the glass fiber board to be detected, and the pixel value of each pixel point of the feature enhanced significant image is a linear judgment index value; The calculation formula for determining the linear feature value corresponding to each pixel point in the preprocessed visible light surface image is: in, is the linear eigenvalue corresponding to the i-th pixel in the preprocessed visible light surface image, λ1(i) is the first eigenvalue corresponding to the i-th pixel in the preprocessed visible light surface image, and λ2(i) is the second eigenvalue corresponding to the i-th pixel in the preprocessed visible light surface image; The calculation formula for determining the optimization factor corresponding to each pixel in the preprocessed visible light surface image is: Among them, τ i is the optimization factor corresponding to the i-th pixel in the preprocessed visible light surface image, c i is the grayscale variance value of the i-th pixel and its eight neighboring pixels in the preprocessed visible light surface image, λ1(i) is the first eigenvalue corresponding to the i-th pixel in the preprocessed visible light surface image, λ2(i) is the second eigenvalue corresponding to the i-th pixel in the preprocessed visible light surface image, and k is a hyperparameter; The calculation formula for determining the linear determination index value corresponding to each pixel point in the preprocessed visible light surface image is: Among them, γ i is the linearity judgment index value corresponding to the i-th pixel point in the preprocessed visible light surface image, λ1(i) is the first eigenvalue corresponding to the i-th pixel point in the preprocessed visible light surface image, λ2(i) is the second eigenvalue corresponding to the i-th pixel point in the preprocessed visible light surface image, τ i is the optimization factor corresponding to the i-th pixel in the preprocessed visible light surface image, is the linear feature value corresponding to the i-th pixel in the preprocessed visible light surface image, is the preset linear feature value threshold.

2. A glass fiber board crack detection method according to claim 1, characterized in that: Also includes: Obtaining first related pixel points and second related pixel points corresponding to each crack edge pixel point in each initial crack region of the glass fiber board to be detected, and determining a crack determination coefficient corresponding to each crack edge pixel point in each initial crack region according to the linear determination index values ​​corresponding to each crack edge pixel point and the first related pixel point and the second related pixel point corresponding to each crack edge pixel point in each initial crack region; Determine each final crack region of the glass fiber board to be detected according to the crack determination coefficient corresponding to each crack edge pixel point in each initial crack region; The calculation formula for determining the crack determination coefficient corresponding to each crack edge pixel point in each initial crack area is: Among them, ω i is the crack determination coefficient corresponding to the i-th crack edge pixel point in each initial crack area, γ i,a is the linear judgment index value corresponding to the first related pixel point corresponding to the i-th crack edge pixel point in each initial crack area, γ i,b is the linear determination index value corresponding to the second related pixel point corresponding to the i-th crack edge pixel point in each initial crack area, γ i is the linear judgment index value corresponding to the i-th crack edge pixel point in each initial crack area.

3. A glass fiber board crack detection method according to claim 1, characterized in that: The steps of determining the crack information of the glass fiber board to be tested include: According to the position of each pixel point in the feature enhanced significant image of the glass fiber board to be detected and the linear determination index value corresponding to each pixel point, each pixel point in the feature enhanced significant image is clustered to obtain M clustering regions of the feature enhanced significant image; If the number M of clustered regions of the feature enhanced significant image is equal to 1, it is determined that there is no crack region in the glass fiber board to be detected. If the number M of clustered regions of the feature enhanced significant image is greater than 1, it is determined that there is a crack region in the glass fiber board to be detected, and according to the mean of the linear judgment index values ​​corresponding to the M clustered regions of the feature enhanced significant image, the clustered region corresponding to the mean of the larger M-1 linear judgment index values ​​is selected from the M clustered regions, and the M-1 clustered regions are used as the initial crack regions of the glass fiber board to be detected.

4. A glass fiber board crack detection method according to claim 2, characterized in that: The steps for determining each final crack area of ​​the fiberglass panel to be inspected include: Determining each first true crack edge pixel point and each pseudo crack edge pixel point in each initial crack region according to a crack determination coefficient corresponding to each crack edge pixel point in each initial crack region; Obtain N first related pixel points and N second related pixel points corresponding to each pseudo crack edge pixel point in each initial crack region, and according to the crack determination coefficients corresponding to each pseudo crack edge pixel point and its corresponding N first related pixel points and N second related pixel points in each initial crack region, select the pixel point corresponding to the maximum crack determination coefficient from each pseudo crack edge pixel point and the N first related pixel points and N second related pixel points corresponding to each pseudo crack edge pixel point, and use the pixel point corresponding to the maximum crack determination coefficient as the second true crack edge pixel point; According to each first true crack edge pixel point and each second true crack edge pixel point in each initial crack region, each final crack region of the glass fiber board to be detected is determined.

5. A glass fiber board crack detection method according to claim 2, characterized in that: The step of obtaining first related pixel points and second related pixel points corresponding to each crack edge pixel point of each initial crack area includes: Determine the tangent line of each crack edge pixel point of each initial crack region according to each crack edge pixel point of each initial crack region; According to the tangent lines of each crack edge pixel point of each initial crack area, one adjacent pixel point is taken from each side of the tangent line of each crack edge pixel point in the vertical direction, and one of the two adjacent pixel points is taken as the first related pixel point, and the other adjacent pixel point of the two adjacent pixel points is taken as the second related pixel point.

6. A fiberglass board crack detection device, characterized in that: It comprises a processor and a memory, wherein the processor is used to process instructions stored in the memory to implement a glass fiber board crack detection method as claimed in any one of claims 1 to 5.

Images