A composite material defect detection method and system based on image processing

Abstract

The present invention provides a composite material defect detection method and system based on image processing, which relates to the field of composite material defect detection. The composite material defect detection method based on image processing comprises the following steps: obtaining an image of a composite material and performing preprocessing, extracting features of the image to be detected using a BTC algorithm, and converting the extracted features into a coordinate curve; performing a similarity comparison between the obtained characteristic coordinate curve and a preset standard curve to determine whether the composite material has defects; if it is determined that the composite material has defects, determining the location information of the defects according to the comparison result. The present invention has high detection accuracy and stability through feature extraction of the image to be detected, effective similarity evaluation and defect location determination, and is suitable for composite material defect detection.

Description

[0001] This application is a divisional application of the application filed on August 31, 2023, with application number 202311112028.X and invention name “A composite material defect detection method and system based on image processing”. Technical Field

[0002] The present invention relates to the field of composite material defect detection, and in particular to a composite material defect detection method and system based on image processing. Background Art

[0003] Composite materials are materials that are made of two or more different materials (called reinforcements and matrices). Reinforcements are usually materials with high strength and stiffness, such as fibers (such as carbon fiber, glass fiber, etc.) or particles. The matrix is ​​a continuous phase used to protect and support the reinforcement, usually a polymer (such as epoxy resin, polyimide, etc.). Composite materials have many excellent properties and are therefore widely used in various fields. However, due to the complex manufacturing process of composite materials and their susceptibility to damage from various environmental factors, the generation and development of defects are inevitable. In order to detect whether the material is qualified, non-destructive testing methods are widely used.

[0004] Nondestructive testing is a non-destructive and non-invasive quality inspection technology used to evaluate and inspect the production quality of industrial products. This technology uses physical means such as light, electricity, and magnetism to inspect, test, and evaluate the defects, chemical, and physical parameters of the object being tested without affecting the original physical functions and characteristics of the object being tested. Conventional nondestructive testing methods include ultrasonic and radiographic testing, but due to the complex structure of composite materials, various defects sometimes occur in the composite materials during the production, processing, and use of the composite materials. The above conventional testing methods cannot fully monitor the deep internal defects of the composite materials, resulting in inaccurate test results.

[0005] Currently, no effective solution has been proposed for the problems in the related technologies. Summary of the invention

[0006] In view of this, in response to the problems in the related art, the present invention provides a composite material defect detection method and system based on image processing to solve the above-mentioned problem that it is impossible to fully monitor the internal deep defects in the composite material, resulting in inaccurate detection results.

[0007] In order to solve the above problems, the specific technical solutions adopted by the present invention are as follows:

[0008] According to one aspect of the present invention, a composite material defect detection method based on image processing is provided, the method comprising the following steps:

[0009] S1. Using an infrared imager and an industrial camera to obtain an image of the composite material and preprocess it to obtain an image of the composite material to be tested;

[0010] S2, using BTC algorithm to extract features of the image to be detected, and converting the extracted features into coordinate curves;

[0011] S3, comparing the obtained characteristic coordinate curve with a preset standard curve for similarity to obtain a comparison result, and judging whether the composite material has defects according to the comparison result;

[0012] S4. If it is determined that the composite material has defects, the location information of the defects is determined according to the comparison result.

[0013] As an embodiment of this invention, the method of obtaining an image of a composite material by using an infrared imager and an industrial camera and performing preprocessing to obtain an image of the composite material to be tested includes the following steps:

[0014] S11, using a halogen lamp as a thermal excitation source to uniformly heat the composite material;

[0015] S12, using an infrared imager to record the heating process of the composite material and obtain an infrared thermal image of the composite material;

[0016] S13, using an industrial camera to obtain a visible light image of the composite material after it is cooled to room temperature;

[0017] S14, using a median filtering method to perform noise reduction processing on the obtained infrared thermal imaging image and visible light image;

[0018] S15, adjusting the infrared thermal image and the visible light image after noise reduction processing to the same size;

[0019] S16. Based on the intuitionistic fuzzy discount operator method, the adjusted infrared thermal imaging image and the visible light image are fused to obtain the image to be tested of the composite material.

[0020] As an embodiment of this invention, the method of fusing the adjusted infrared thermal imaging image and the visible light image based on the intuitionistic fuzzy discount operator method to obtain the image to be detected of the composite material includes the following steps:

[0021] S161, using SIFT algorithm to extract features from the adjusted infrared thermal imaging image and visible light image respectively to obtain feature vectors;

[0022] S162, forming a measurement matrix using the feature vector of the infrared thermal imaging image and the feature vector of the visible light image;

[0023] S163, calculating the Euclidean distance between the feature vector of the infrared thermal imaging image and the feature vector of the visible light image;

[0024] S164. Calculate the correlation coefficient between the feature vector of the infrared thermal imaging image and the feature vector of the visible light image according to the grey correlation analysis, and construct a correlation coefficient matrix;

[0025] S165, generating credibility for the obtained correlation coefficient matrix to obtain a credibility distribution matrix;

[0026] S166, converting the credibility distribution matrix into an intuitive fuzzy description matrix and performing a discount weighted operator operation to obtain a fused intuitive fuzzy decision vector, defuzzifying the intuitive fuzzy decision vector to obtain an image to be detected.

[0027] As an embodiment of this invention, the calculation formula for calculating the correlation coefficient between the feature vector of the infrared thermal imaging image and the feature vector of the visible light image is:

[0028]

[0029] In the formula, ξ(A i ,B j ) represents the correlation coefficient between the i-th element of the feature vector A of the infrared thermal image and the j-th element of the feature vector B of the visible light image;

[0030] D i,j (A i ,B j ) represents the Euclidean distance between the i-th element of the feature vector A of the infrared thermal image and the j-th element of the feature vector B of the visible light image;

[0031] ρ represents the resolution coefficient, and its value range is [0,1].

[0032] As an embodiment of this invention, the process of converting the credibility distribution matrix into an intuitive fuzzy description matrix and performing a discount weighted operator operation to obtain a fused intuitive fuzzy decision vector, defuzzifying the intuitive fuzzy decision vector, and obtaining an image to be detected includes the following steps:

[0033] S1661, defining a discount factor and multiplying each element in the intuitionistic fuzzy description matrix by the corresponding discount factor to obtain a new intuitionistic fuzzy description matrix;

[0034] S1662, using a weighted average method to fuse the new intuitionistic fuzzy description matrix to obtain an intuitionistic fuzzy decision vector;

[0035] S1663. Use the maximum membership method to defuzzify the obtained intuitive fuzzy decision vector to obtain the image to be detected.

[0036] As an embodiment of this invention, the method of extracting features of an image to be detected by using the BTC algorithm and converting the extracted features into a coordinate curve includes the following steps:

[0037] S21, converting the image to be detected into a grayscale image, and dividing the grayscale image into k×k non-overlapping blocks;

[0038] S22, calculating the grayscale mean of the pixels in each sub-block;

[0039] S23, according to the BTC algorithm, the pixel points whose grayscale values ​​in each sub-block are greater than the mean are assigned a value of 1, otherwise, the pixel points whose grayscale values ​​in each sub-block are less than or equal to the mean are assigned a value of 0, thereby obtaining a series of k×k binary feature blocks;

[0040] S24, marking each binary feature block with a feature number from left to right and from top to bottom;

[0041] S25. Use the marked feature serial number as the horizontal coordinate of the coordinate axis, and use the pixel value of the binary feature block as the vertical coordinate to obtain a feature curve graph.

[0042] As an embodiment of the present invention, the obtaining characteristic coordinate curve is compared with a preset standard curve for similarity to obtain a comparison result, and judging whether the composite material has defects according to the comparison result includes the following steps:

[0043] S31, obtaining a preset standard curve;

[0044] S32, aligning the obtained characteristic coordinate curve with the preset standard curve so that the characteristic coordinate curve and the preset standard curve have the same starting point and end point;

[0045] S33, using the Fréchet distance algorithm to evaluate the similarity between the characteristic coordinate curve and the preset standard curve;

[0046] S34. If the Fréchet distance is less than a preset threshold, it indicates that the composite material has no defects; otherwise, it indicates that the composite material has defects.

[0047] As an embodiment of this invention, the method of evaluating the similarity between the characteristic coordinate curve and the preset standard curve by using the Fréchet distance algorithm includes the following steps:

[0048] S331, calculating the distance between each data point on the characteristic coordinate curve and the preset standard curve to obtain a distance matrix;

[0049] S332. Find the maximum distance d in the distance matrix max and the minimum distance d min , and initialize the target distance f = d min , set the cycle interval r = (d max -d min ) / 100;

[0050] S333, setting the elements in the distance matrix that are less than or equal to the target distance f to 1, and setting the elements that are greater than the target distance to 0, to obtain a binary matrix;

[0051] S334, searching for an optimal path that meets preset conditions in the obtained binary matrix;

[0052] S335. If no path that meets the preset conditions is found, the target distance is updated, and the target distance f n =f+r, where n represents the number of updates, and steps S333-S334 are repeated until the optimal path that meets the preset conditions or the target distance f=d is found. max ;

[0053] S336, obtaining the Fréchet distance F=f between the characteristic coordinate curve and the preset standard curve n ;

[0054] S337. Calculate the similarity between the characteristic coordinate curve and the preset standard curve based on the obtained Fleche distance. The similarity calculation formula is: S=1 / F, where S represents the similarity between the characteristic coordinate curve and the preset standard curve, and F represents the Fleche distance between the characteristic coordinate curve and the preset standard curve.

[0055] As an embodiment of this invention, if it is determined that the composite material has defects, determining the location information of the defects according to the comparison result includes the following steps:

[0056] S41, if it is determined that the composite material has defects, find out the point where the distance between each data point on the characteristic coordinate curve and the preset standard curve is greater than a preset threshold;

[0057] S42, determining the corresponding feature serial number according to the points whose distance is greater than the preset threshold;

[0058] S43, determining the image position corresponding to the feature serial number as the defect area, and obtaining the defect position of the composite material.

[0059] According to another aspect of the present invention, a composite material defect detection system based on image processing is provided, the system comprising: an image acquisition processing module, a feature extraction and conversion module, a similarity comparison module and a defect position judgment module, and the image acquisition processing module, the feature extraction and conversion module, the similarity comparison module and the defect position judgment module are connected in sequence;

[0060] The image acquisition and processing module is used to acquire the image of the composite material using an infrared imager and an industrial camera and perform preprocessing to obtain the image to be tested of the composite material;

[0061] The feature extraction and conversion module is used to extract features of the image to be detected using the BTC algorithm and convert the extracted features into a coordinate curve;

[0062] The similarity comparison module is used to compare the obtained characteristic coordinate curve with the preset standard curve to obtain a comparison result, and judge whether the composite material has defects according to the comparison result;

[0063] The defect position judgment module is used to determine the position information of the defect according to the comparison result if it is judged that the composite material has a defect.

[0064] Compared with the prior art, the present invention provides a composite material defect detection method and system based on image processing, which has the following beneficial effects:

[0065] (1) The present invention utilizes an infrared imager to obtain infrared thermal images of composite materials, while an industrial camera can obtain visible light images. The combined use of these two imaging methods can provide richer information and help detect defects more accurately. By extracting features from the image to be detected and effectively evaluating similarity and determining the defect location, the present invention has high detection accuracy and stability and is suitable for composite material defect detection.

[0066] (2) The present invention uses the SIFT algorithm to extract features from infrared thermal images and visible light images, and forms a measurement matrix with feature vectors, thereby effectively extracting the key features of the image, which is beneficial to subsequent fusion and similarity comparison. By generating a credibility distribution matrix and performing a discount weighted operator operation, the fused intuitive fuzzy decision vector is obtained, making the fusion result more reasonable and accurate, so that the image to be tested of the composite material can more comprehensively reveal the defects of the material.

[0067] (3) The present invention converts the image to be detected into a grayscale image, which helps to simplify the image processing and feature extraction process. The grayscale image is divided into non-overlapping blocks, the image can be divided into small blocks for processing, and features can be extracted in a local range, reducing the complexity of the overall image processing. The BTC algorithm is used to extract features from the image to be detected and convert it into a feature coordinate curve, which can simplify the image processing process and help improve the accuracy and efficiency of composite material defect detection.

[0068] (4) The present invention ensures the accuracy of comparison by aligning the characteristic coordinate curve and the preset standard curve so that they have the same starting point and end point. By calculating the distance matrix between the characteristic coordinate curve and the preset standard curve and searching for the optimal path according to preset conditions, the Fréchet distance between the characteristic coordinate curve and the preset standard curve can be obtained. Based on the obtained Fréchet distance, the similarity between the characteristic coordinate curve and the preset standard curve can be calculated, thereby determining whether the composite material has defects. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. 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 creative work. In the drawings:

[0070] Figure 1 is a flow chart of a composite material defect detection method based on image processing according to an embodiment of the present invention;

[0071] Figure 2 4 is a principle block diagram of a composite material defect detection system based on image processing according to an embodiment of the present invention.

[0072] In the figure:

[0073] 1. Image acquisition and processing module; 2. Feature extraction and conversion module; 3. Similarity comparison module; 4. Defect location judgment module. DETAILED DESCRIPTION

[0074] In order to enable those skilled in the art to better understand the technical solutions in this application, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this application.

[0075] According to an embodiment of the present invention, a composite material defect detection method and system based on image processing are provided.

[0076] The present invention is further described with reference to the accompanying drawings and specific embodiments. Figure 1 As shown, according to one embodiment of the present invention, a composite material defect detection method based on image processing is provided, and the method comprises the following steps:

[0077] S1. Using an infrared imager and an industrial camera to obtain an image of the composite material and preprocess it to obtain an image of the composite material to be tested;

[0078] Preferably, the method of obtaining an image of the composite material by using an infrared imager and an industrial camera and performing preprocessing to obtain an image of the composite material to be tested comprises the following steps:

[0079] S11, using a halogen lamp as a thermal excitation source to uniformly heat the composite material;

[0080] Specifically, two halogen lamps with a power of 800 W are selected as thermal excitation sources to uniformly heat the composite material.

[0081] S12, using an infrared imager to record the heating process of the composite material and obtain an infrared thermal image of the composite material;

[0082] S13, using an industrial camera to obtain a visible light image of the composite material after it is cooled to room temperature;

[0083] S14, using a median filtering method to perform noise reduction processing on the obtained infrared thermal imaging image and visible light image;

[0084] It should be noted that median filtering is a common nonlinear filtering method used for image noise reduction. Its principle is to replace the gray value of each pixel with the median of the gray values ​​of all pixels in a certain area around the pixel. This method can effectively remove salt and pepper noise or other random noise in the image while maintaining the edge information of the image.

[0085] S15, adjusting the infrared thermal image and the visible light image after noise reduction processing to the same size;

[0086] S16. Based on the intuitionistic fuzzy discount operator method, the adjusted infrared thermal imaging image and the visible light image are fused to obtain the image to be tested of the composite material.

[0087] Preferably, the step of fusing the adjusted infrared thermal imaging image and the visible light image based on the intuitionistic fuzzy discount operator method to obtain the image to be detected of the composite material comprises the following steps:

[0088] S161, using SIFT algorithm to extract features from the adjusted infrared thermal imaging image and visible light image respectively to obtain feature vectors;

[0089] It should be noted that the SIFT algorithm is a classic algorithm for image feature extraction and matching. It can extract key points and feature descriptors in images and has the advantages of scale invariance and rotation invariance. The feature extraction using the SIFT algorithm includes the following steps:

[0090] Perform key point detection of SIFT algorithm on the adjusted infrared thermal image and visible light image respectively;

[0091] For each detected key point, the SIFT algorithm generates a feature descriptor. The feature descriptor is a 128-dimensional vector that describes the characteristics of the image area around the key point. It can represent information such as image texture and gradient direction, and has good invariance;

[0092] The feature descriptors of all key points are combined together to obtain a feature vector.

[0093] S162, forming a measurement matrix using the feature vector of the infrared thermal imaging image and the feature vector of the visible light image;

[0094] S163, calculating the Euclidean distance between the feature vector of the infrared thermal imaging image and the feature vector of the visible light image;

[0095] It should be noted that Euclidean distance is a basic distance measurement method, which is widely used in data analysis, pattern recognition, image processing and other fields to measure the similarity or distance between samples.

[0096] S164. Calculate the correlation coefficient between the feature vector of the infrared thermal imaging image and the feature vector of the visible light image according to the grey correlation analysis, and construct a correlation coefficient matrix;

[0097] Preferably, the calculation formula for calculating the correlation coefficient between the feature vector of the infrared thermal imaging image and the feature vector of the visible light image is:

[0098]

[0099] In the formula, ξ(A i ,B j ) represents the correlation coefficient between the i-th element of the feature vector A of the infrared thermal image and the j-th element of the feature vector B of the visible light image;

[0100] D i,j (A i ,B j) represents the Euclidean distance between the i-th element of the feature vector A of the infrared thermal image and the j-th element of the feature vector B of the visible light image;

[0101] ρ represents the resolution coefficient, and its value range is [0,1].

[0102] S165, generating credibility for the obtained correlation coefficient matrix to obtain a credibility distribution matrix;

[0103] It should be noted that generating a credibility distribution matrix is ​​the process of converting each element in the correlation coefficient matrix into an intuitive fuzzy description matrix. The credibility distribution matrix reflects the degree of correlation between features and is used to consider the weights between features in the fusion process. Specifically, it includes the following steps:

[0104] For each element in the correlation coefficient matrix, its value is mapped to a membership value by linear mapping;

[0105] Determine the form of the credibility distribution function. The credibility distribution function can be a curve used to describe the membership values ​​between features. Usually, the credibility distribution function is a standard fuzzy membership function, such as an S-curve or a Z-curve.

[0106] According to the credibility allocation function, the membership values ​​are obtained in the form of an intuitive fuzzy description matrix, which is the credibility allocation matrix.

[0107] S166, converting the credibility distribution matrix into an intuitive fuzzy description matrix and performing a discount weighted operator operation to obtain a fused intuitive fuzzy decision vector, defuzzifying the intuitive fuzzy decision vector to obtain an image to be detected.

[0108] Preferably, the step of converting the credibility distribution matrix into an intuitionistic fuzzy description matrix and performing a discount weighted operator operation to obtain a fused intuitionistic fuzzy decision vector, defuzzifying the obtained intuitionistic fuzzy decision vector, and obtaining the image to be detected comprises the following steps:

[0109] S1661, defining a discount factor and multiplying each element in the intuitionistic fuzzy description matrix by the corresponding discount factor to obtain a new intuitionistic fuzzy description matrix;

[0110] It should be noted that the discount factor is a parameter used to adjust the weights between different features, and it can be used to reflect the importance of the feature. In the intuitionistic fuzzy description matrix, each element represents the relationship or similarity between features. The discount factor is usually a real number between 0 and 1, which represents the weight of the feature. If the discount factor is 1, it means that the weight of the feature remains unchanged and no weighting is performed. If the discount factor is less than 1, it means that the weight of the feature is reduced, and accordingly, the influence of the feature on the final result is also reduced. If the discount factor is 0, it means that the weight of the feature is 0, and the feature does not participate in the decision-making process, that is, it has no effect.

[0111] S1662, using a weighted average method to fuse the new intuitionistic fuzzy description matrix to obtain an intuitionistic fuzzy decision vector;

[0112] It should be noted that the basic idea of ​​the weighted average method is to perform weighted averaging on the fuzzy description matrices of different features to obtain a comprehensive intuitive fuzzy decision vector, and to reasonably integrate the information of different features to obtain a more comprehensive and accurate decision result.

[0113] S1663. Use the maximum membership method to defuzzify the obtained intuitive fuzzy decision vector to obtain the image to be detected.

[0114] It should be noted that the maximum membership method is a commonly used defuzzification method, which is used to convert the intuitive fuzzy decision vector into a specific numerical result. For each element in the intuitive fuzzy decision vector, the category index with the maximum membership value is found, and the category index corresponding to the maximum membership value is used as the pixel value of the image to be detected to obtain the final image to be detected.

[0115] S2, using BTC algorithm to extract features of the image to be detected, and converting the extracted features into coordinate curves;

[0116] It should be noted that the BTC (Block Truncation Coding) algorithm is a method for image feature extraction, which can convert an image into a series of feature blocks and encode the features based on the comparison between the pixel gray value and the mean.

[0117] Preferably, the method of extracting features from the image to be detected by using the BTC algorithm and converting the extracted features into a coordinate curve comprises the following steps:

[0118] S21, converting the image to be detected into a grayscale image, and dividing the grayscale image into k×k non-overlapping blocks;

[0119] It should be noted that converting the image to be detected into a grayscale image is the process of converting a color image into a single-channel grayscale image. The pixel value of the RGB color image can be converted into the pixel value of the grayscale image by a conversion formula. The conversion formula is:

[0120] H=0.3R+0.59G+0.11B

[0121] In the formula, H represents the converted grayscale image;

[0122] R, G, and B represent the red channel value, green channel value, and blue channel value, respectively.

[0123] S22, calculating the grayscale mean of the pixels in each sub-block;

[0124] It should be noted that the calculation formula for the grayscale mean of pixels in each sub-block is:

[0125]

[0126] In the formula, μ represents the grayscale mean of the pixels in the sub-block;

[0127] w(o,p) represents the gray value of the pixel at the coordinate (o,p) in the sub-block;

[0128] k×k represents the size of each word block.

[0129] S23, according to the BTC algorithm, the pixel points whose grayscale values ​​in each sub-block are greater than the mean are assigned a value of 1, otherwise, the pixel points whose grayscale values ​​in each sub-block are less than or equal to the mean are assigned a value of 0, thereby obtaining a series of k×k binary feature blocks;

[0130] S24, marking each binary feature block with a feature number from left to right and from top to bottom;

[0131] S25. Use the marked feature serial number as the horizontal coordinate of the coordinate axis, and use the pixel value of the binary feature block as the vertical coordinate to obtain a feature curve graph.

[0132] S3, comparing the obtained characteristic coordinate curve with a preset standard curve for similarity to obtain a comparison result, and judging whether the composite material has defects according to the comparison result;

[0133] Preferably, the process of comparing the obtained characteristic coordinate curve with a preset standard curve for similarity to obtain a comparison result, and judging whether the composite material has defects according to the comparison result comprises the following steps:

[0134] S31, obtaining a preset standard curve;

[0135] It should be noted that the preset standard curve can be formed by selecting a composite material without defects and performing the processing of steps S1-S2 to form a standard characteristic curve.

[0136] S32, aligning the obtained characteristic coordinate curve with the preset standard curve so that the characteristic coordinate curve and the preset standard curve have the same starting point and end point;

[0137] S33, using the Fréchet distance algorithm to evaluate the similarity between the characteristic coordinate curve and the preset standard curve;

[0138] Preferably, the method of evaluating the similarity between the characteristic coordinate curve and the preset standard curve using the Fréchet distance algorithm comprises the following steps:

[0139] S331, calculating the distance between each data point on the characteristic coordinate curve and the preset standard curve to obtain a distance matrix;

[0140] It should be noted that after the characteristic coordinate curve is aligned with the preset standard curve, the distance between each corresponding data point is calculated, and the obtained distances are arranged in the order of the data points to form a distance matrix, and the formed distance matrix expression is:

[0141]

[0142] Where G represents the distance matrix;

[0143] d mn It represents the distance from the mth data point on the characteristic coordinate curve to the nth data point on the preset standard curve, and 1≤m≤M, 1≤n≤N.

[0144] S332. Find the maximum distance d in the distance matrix max and the minimum distance d min , and initialize the target distance f = d min , set the cycle interval r = (d max -d min ) / 100;

[0145] S333, set the elements in the distance matrix that are less than or equal to the target distance f to 1, and set the elements that are greater than the target distance to 0, to obtain a binary matrix; the expression of the binary matrix is:

[0146]

[0147] In the formula, G′ represents a binary matrix;

[0148] d ′ mn Represents the element value of the mth row and nth column of a binary matrix.

[0149] S334, searching for an optimal path T that meets preset conditions in the obtained binary matrix;

[0150] It should be noted that the preset condition is: the starting point of T is d ′ 11 , the end point is d ′ MN , path T passes through point d ′ mn After that, the next point it passes through is d ′ (m+1)n d ′ m(n+1) d ′ (m+1)n+1) , and the value of all points in path T must be 1.

[0151] S335. If no path that meets the preset conditions is found, the target distance is updated, and the target distance f n =f+r, where n represents the number of updates, and steps S333-S334 are repeated until the optimal path that meets the preset conditions or the target distance f=d is found. max ;

[0152] S336, obtaining the Fréchet distance F=f between the characteristic coordinate curve and the preset standard curve n ;

[0153] S337. Calculate the similarity between the characteristic coordinate curve and the preset standard curve based on the obtained Fleche distance. The similarity calculation formula is: S=1 / F, where S represents the similarity between the characteristic coordinate curve and the preset standard curve, and F represents the Fleche distance between the characteristic coordinate curve and the preset standard curve.

[0154] S34. If the Fréchet distance is less than a preset threshold, it indicates that the composite material has no defects; otherwise, it indicates that the composite material has defects.

[0155] S4. If it is determined that the composite material has defects, the location information of the defects is determined according to the comparison result.

[0156] Preferably, if it is determined that the composite material has defects, determining the location information of the defects according to the comparison result comprises the following steps:

[0157] S41, if it is determined that the composite material has defects, find out the point where the distance between each data point on the characteristic coordinate curve and the preset standard curve is greater than a preset threshold;

[0158] S42, determining the corresponding feature serial number according to the points whose distance is greater than the preset threshold;

[0159] S43, determining the image position corresponding to the feature serial number as the defect area, and obtaining the defect position of the composite material.

[0160] like Figure 2 As shown, according to another embodiment of the present invention, a composite material defect detection system based on image processing is provided, the system comprising: an image acquisition processing module 1, a feature extraction and conversion module 2, a similarity comparison module 3 and a defect position judgment module 4, and the image acquisition processing module 1, the feature extraction and conversion module 2, the similarity comparison module 3 and the defect position judgment module 4 are connected in sequence;

[0161] The image acquisition and processing module 1 is used to acquire an image of the composite material using an infrared imager and an industrial camera and perform preprocessing to obtain an image of the composite material to be tested;

[0162] The feature extraction and conversion module 2 is used to extract features of the image to be detected using the BTC algorithm and convert the extracted features into a coordinate curve;

[0163] The similarity comparison module 3 is used to compare the obtained characteristic coordinate curve with the preset standard curve to obtain a comparison result, and judge whether the composite material has defects according to the comparison result;

[0164] The defect position determination module 4 is used to determine the position information of the defect according to the comparison result if it is determined that the composite material has a defect.

[0165] In summary, with the aid of the above technical scheme of the present invention, the present invention can obtain infrared thermal imaging images of composite materials by using an infrared imager, and an industrial camera can obtain visible light images. The combined use of these two imaging methods can provide richer information, which is helpful for more accurate defect detection. By extracting features from the image to be detected, as well as effective similarity evaluation and defect location determination, the present invention has high detection accuracy and stability, and is suitable for composite material defect detection. The present invention uses the SIFT algorithm to extract features from infrared thermal imaging images and visible light images, and forms a measurement matrix with feature vectors, thereby effectively extracting key features of the image, which is beneficial to subsequent fusion and similarity comparison. By generating a credibility distribution matrix and performing a discount weighted operator operation, a fused intuitive fuzzy decision vector is obtained, which makes the fusion result more reasonable and accurate, so that the image to be detected of the composite material can more comprehensively reveal the defects of the material. The present invention will Converting the image to be detected into a grayscale image helps to simplify the image processing and feature extraction process. By dividing the grayscale image into non-overlapping blocks, the image can be divided into small blocks for processing, and features can be extracted in a local range, reducing the complexity of the overall image processing. Using the BTC algorithm to extract features from the image to be detected and convert it into a characteristic coordinate curve can simplify the image processing process and help improve the accuracy and efficiency of composite material defect detection. The present invention ensures the accuracy of comparison by aligning the characteristic coordinate curve and the preset standard curve so that they have the same starting point and end point. By calculating the distance matrix between the characteristic coordinate curve and the preset standard curve and searching for the optimal path according to preset conditions, the Fréchet distance between the characteristic coordinate curve and the preset standard curve can be obtained. According to the obtained Fréchet distance, the similarity between the characteristic coordinate curve and the preset standard curve can be calculated, so as to determine whether the composite material has defects.

[0166] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0167] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A composite material defect detection method based on image processing, characterized in that: The method comprises the following steps: S1. Using an infrared imager and an industrial camera to obtain an image of the composite material and preprocess it to obtain an image of the composite material to be tested; S2, using BTC algorithm to extract features of the image to be detected, and converting the extracted features into coordinate curves; S3, comparing the obtained characteristic coordinate curve with a preset standard curve for similarity to obtain a comparison result, and judging whether the composite material has defects according to the comparison result; The method of comparing the obtained characteristic coordinate curve with a preset standard curve for similarity to obtain a comparison result, and judging whether the composite material has defects according to the comparison result comprises the following steps: S31, obtaining a preset standard curve; S32, aligning the obtained characteristic coordinate curve with the preset standard curve so that the characteristic coordinate curve and the preset standard curve have the same starting point and end point; S33, using the Fréchet distance algorithm to evaluate the similarity between the characteristic coordinate curve and the preset standard curve; S34, if the Fréchet distance is less than a preset threshold, it indicates that the composite material has no defects, otherwise, it indicates that the composite material has defects; S4. If it is determined that the composite material has defects, determine the location information of the defects according to the comparison result; If it is determined that the composite material has defects, determining the location information of the defects according to the comparison result includes the following steps: S41, if it is determined that the composite material has defects, find out the point where the distance between each data point on the characteristic coordinate curve and the preset standard curve is greater than a preset threshold; S42, determining the corresponding feature serial number according to the points whose distance is greater than the preset threshold; S43, determining the image position corresponding to the feature number as the defect area, and obtaining the defect position of the composite material; The method of obtaining an image of the composite material by using an infrared imager and an industrial camera and performing preprocessing to obtain an image of the composite material to be tested includes the following steps: S11, using a halogen lamp as a thermal excitation source to uniformly heat the composite material; S12, using an infrared imager to record the heating process of the composite material and obtain an infrared thermal image of the composite material; S13, using an industrial camera to obtain a visible light image of the composite material after it is cooled to room temperature; S14, using a median filtering method to perform noise reduction processing on the obtained infrared thermal imaging image and visible light image; S15, adjusting the infrared thermal image and the visible light image after noise reduction processing to the same size; S16. Based on the intuitionistic fuzzy discount operator method, the adjusted infrared thermal imaging image and the visible light image are fused to obtain an image of the composite material to be tested; The method of fusing the adjusted infrared thermal imaging image and the visible light image based on the intuitionistic fuzzy discount operator method to obtain the image to be detected of the composite material includes the following steps: S161, using SIFT algorithm to extract features from the adjusted infrared thermal imaging image and visible light image respectively to obtain feature vectors; S162, forming a measurement matrix using the feature vector of the infrared thermal imaging image and the feature vector of the visible light image; S163, calculating the Euclidean distance between the feature vector of the infrared thermal imaging image and the feature vector of the visible light image; S164. Calculate the correlation coefficient between the feature vector of the infrared thermal imaging image and the feature vector of the visible light image according to the grey correlation analysis, and construct a correlation coefficient matrix; S165, generating credibility for the obtained correlation coefficient matrix to obtain a credibility distribution matrix; S166, converting the credibility distribution matrix into an intuitionistic fuzzy description matrix and performing a discount weighted operator operation to obtain a fused intuitionistic fuzzy decision vector, defuzzifying the obtained intuitionistic fuzzy decision vector to obtain an image to be detected; The process of converting the credibility distribution matrix into an intuitionistic fuzzy description matrix and performing a discount weighted operator operation to obtain a fused intuitionistic fuzzy decision vector, defuzzifying the obtained intuitionistic fuzzy decision vector, and obtaining an image to be detected includes the following steps: S1661, defining a discount factor and multiplying each element in the intuitionistic fuzzy description matrix by the corresponding discount factor to obtain a new intuitionistic fuzzy description matrix; S1662, using a weighted average method to fuse the new intuitionistic fuzzy description matrix to obtain an intuitionistic fuzzy decision vector; S1663. Use the maximum membership method to defuzzify the obtained intuitive fuzzy decision vector to obtain the image to be detected.

2. The composite material defect detection method based on image processing according to claim 1, characterized in that: The calculation formula for calculating the correlation coefficient between the feature vector of the infrared thermal imaging image and the feature vector of the visible light image is: In the formula, ξ(A i ,B j ) represents the correlation coefficient between the i-th element of the feature vector A of the infrared thermal image and the j-th element of the feature vector B of the visible light image; D i,j (A i ,B j ) represents the Euclidean distance between the i-th element of the feature vector A of the infrared thermal image and the j-th element of the feature vector B of the visible light image; ρ represents the resolution coefficient, and its value range is [0,1].

3. The composite material defect detection method based on image processing according to claim 1, characterized in that: The method of extracting features from the image to be detected by using the BTC algorithm and converting the extracted features into a coordinate curve comprises the following steps: S21, converting the image to be detected into a grayscale image, and dividing the grayscale image into k×k non-overlapping blocks; S22, calculating the grayscale mean of the pixels in each sub-block; S23, according to the BTC algorithm, the pixel points whose grayscale values ​​in each sub-block are greater than the mean are assigned a value of 1, otherwise, the pixel points whose grayscale values ​​in each sub-block are less than or equal to the mean are assigned a value of 0, thereby obtaining a series of k×k binary feature blocks; S24, marking each binary feature block with a feature number from left to right and from top to bottom; S25. Use the marked feature serial number as the horizontal coordinate of the coordinate axis, and use the pixel value of the binary feature block as the vertical coordinate to obtain a feature curve graph.

4. The composite material defect detection method based on image processing according to claim 1, characterized in that: The method of evaluating the similarity between the characteristic coordinate curve and the preset standard curve by using the Fréchet distance algorithm comprises the following steps: S331, calculating the distance between each data point on the characteristic coordinate curve and the preset standard curve to obtain a distance matrix; S332. Find the maximum distance d in the distance matrix max and the minimum distance d min , and initialize the target distance f = d min , set the cycle interval r = (d max -d min ) / 100; S333, setting the elements in the distance matrix that are less than or equal to the target distance f to 1, and setting the elements that are greater than the target distance to 0, to obtain a binary matrix; S334, searching for an optimal path that meets preset conditions in the obtained binary matrix; S335. If no path that meets the preset conditions is found, the target distance is updated, and the target distance f n =f+r, where n represents the number of updates, and steps S333-S334 are repeated until the optimal path that meets the preset conditions or the target distance f=d is found. max ; S336, obtaining the Fréchet distance F=f between the characteristic coordinate curve and the preset standard curve n ; S337. Calculate the similarity between the characteristic coordinate curve and the preset standard curve based on the obtained Fleche distance. The similarity calculation formula is: S=1 / F, where S represents the similarity between the characteristic coordinate curve and the preset standard curve, and F represents the Fleche distance between the characteristic coordinate curve and the preset standard curve.

5. A composite material defect detection system based on image processing, used to implement the composite material defect detection method based on image processing according to any one of claims 1 to 4, characterized in that: The system comprises: an image acquisition processing module, a feature extraction and conversion module, a similarity comparison module and a defect position judgment module, and the image acquisition module, the feature extraction and conversion module, the similarity comparison module and the defect position judgment module are connected in sequence; The image acquisition and processing module is used to acquire the image of the composite material using an infrared imager and an industrial camera and perform preprocessing to obtain the image to be tested of the composite material; The feature extraction and conversion module is used to extract features of the image to be detected using the BTC algorithm and convert the extracted features into a coordinate curve; The similarity comparison module is used to compare the obtained characteristic coordinate curve with the preset standard curve to obtain a comparison result, and judge whether the composite material has defects according to the comparison result; The defect position determination module is used to determine the position information of the defect according to the comparison result if it is determined that the composite material has a defect, and comprises the following steps: S41, if it is determined that the composite material has defects, find out the point where the distance between each data point on the characteristic coordinate curve and the preset standard curve is greater than a preset threshold; S42, determining the corresponding feature serial number according to the points whose distance is greater than the preset threshold; S43, determining the image position corresponding to the feature serial number as the defect area, and obtaining the defect position of the composite material.

Images