Monitoring method based on metal thermal insulation layer impact resistance detection process

Abstract

The invention relates to the field of image processing, in particular to a monitoring method based on a metal thermal insulation layer impact resistance detection process, and the method comprises the steps: obtaining a surface image, depth data and temperature data of a preset angle of a metal thermal insulation layer, carrying out the transformation and correction of a gray image of the surface image, and screening a target image; dividing the target image into regions according to the structural characteristics of the metal thermal insulation layer, calculating deformation scores of different regions based on depth data of the different regions, and performing weighted summation on the deformation scores of the different regions to obtain a comprehensive deformation score; and identifying a maximum deformation area and a temperature anomaly area according to the comprehensive deformation score, and performing temperature difference verification on the maximum deformation area of the target image to evaluate the impact resistance. According to the method, the deformation score and the temperature anomaly score are synthesized, the damage area is identified, and the accuracy of impact resistance detection is improved.

Description

Technical Field

[0001] The present invention relates to the field of image processing, and more specifically, to a monitoring method based on a metal insulation layer impact resistance detection process. Background Art

[0002] In industrial applications, metal insulation is used for thermal management of pipes, tanks and equipment. As the complexity of the operating environment increases, metal insulation is often exposed to external factors such as shock, vibration and temperature changes, which may cause damage to the insulation, thus affecting the energy efficiency and safety of the equipment.

[0003] The existing Chinese patent application document with publication number CN115359053A discloses a metal plate defect intelligent detection method and system, which relates to the field of computer vision technology. The method includes obtaining a grayscale image of the metal plate surface; obtaining a grayscale run-length matrix of the grayscale image after each grayscale dimensionality reduction; obtaining the dimensionality reduction credibility of the grayscale image after each grayscale dimensionality reduction; obtaining the maximum allowable dimensionality reduction degree of the grayscale image; obtaining the maximum value of the difference between each pixel point in the grayscale image and the average grayscale value of pixels in the neighborhood of the pixel point; obtaining the smoothing filter window size of each pixel point in the grayscale image according to the maximum value, and performing smoothing filtering on the grayscale image according to the smoothing filter window size of each pixel point to obtain a target grayscale image; performing edge detection on the target grayscale image to obtain a defect area in the target grayscale image; the present invention solves the technical problem in the related art that accurate detection of surface defects of metal plates cannot be achieved when performing defect detection on metal plates.

[0004] In the above application documents, the accuracy of detecting small defects is improved through grayscale dimension reduction and smoothing filtering. At present, the surface image of the metal insulation layer is subjected to impact test based on the defects on the surface of the metal insulation layer based on machine vision. However, due to the uneven surface and different reflectivity of the concave-convex structured metal insulation layer, it is impossible to accurately judge whether the metal insulation layer has defects and deformation. At the same time, subjective errors and plane image visual errors lead to inaccurate evaluation of the metal insulation layer's resistance to LOCA impact, causing potential risks during use. Summary of the invention

[0005] In order to solve the problem that the surface of the concave-convex structure metal insulation layer is not smooth and the reflectivity is different, which makes it impossible to accurately judge whether the metal insulation layer has defects and deformation, resulting in inaccurate impact resistance detection, the present invention provides solutions in the following aspects.

[0006] The monitoring method based on the impact resistance detection process of the metal insulation layer includes: obtaining a surface image, depth data and temperature data of the metal insulation layer at a preset angle, obtaining a grayscale image of the surface image at a preset angle, performing transformation correction, and screening a target image; dividing the target image into regions according to the structural characteristics of the metal insulation layer, calculating the deformation scores of different regions based on the depth data of different regions, taking the number and area of ​​deformation in different regions as weights for weighted summation, and obtaining a comprehensive deformation score; in response to the comprehensive deformation score being greater than a preset threshold, extracting the pixel point with the largest depth difference before and after detection in the target image as the center point of regional growth, performing regional growth, obtaining the maximum deformation region, and segmenting the temperature abnormal region according to the temperature data of the target image; matching the maximum deformation region with the temperature abnormal region to obtain the temperature abnormal score, and performing temperature difference verification on the maximum deformation region of the target image to evaluate the impact resistance; wherein the comprehensive deformation score Satisfies the following relationship: , where , They represent the number of concave and convex points in the concave and convex area and the area of ​​the smooth area before the test, , They represent the number of concave and convex points in the concave and convex area and the area of ​​deformation in the smooth area during the test. , Represent the deformation fractions of the concave and convex areas and the smooth area respectively.

[0007] The effect is: by transforming and correcting the preset angle and screening the target image, it is beneficial to select the image with the richest texture information and avoid the deformation caused by image shooting. At the same time, the target image is divided into concave and convex areas and smooth areas, allowing targeted analysis of deformation in different areas. By calculating the deformation scores of different areas and combining the number and area of ​​deformation as weights, the maximum deformation area is obtained according to the structural characteristics of the metal insulation layer, and the degree of impact resistance detection in different areas is analyzed. Combined with the temperature abnormality area, the accuracy of impact resistance detection is verified.

[0008] Preferably, the step of obtaining a grayscale image of a surface image at a preset angle and performing transformation correction includes: The angle perpendicular to the surface image is taken as the standard viewing angle, and the perspective transformation matrix of the surface images taken at different preset angles relative to the standard viewing angle is calculated. Based on the perspective transformation matrix, the coordinates of each pixel point are mapped from the surface image at the original preset angle to the standard viewing angle to obtain a mapped image, and bilinear interpolation is used to calculate the pixel values ​​of the pixel points of the mapped image to obtain corrected grayscale images of images at different angles.

[0009] The effect is: by taking the angle perpendicular to the surface image as the standard viewing angle, it ensures that all images taken from different angles are converted to the same reference frame, thereby standardizing the image data, and correcting the perspective distortion caused by the shooting angle by calculating the perspective transformation matrix, so that the objects in the image are presented in their actual size and shape. The corrected grayscale image can more clearly show the texture and features of the surface for subsequent image analysis.

[0010] Preferably, the screening target image includes: The rectified grayscale image is segmented using the Otsu method to obtain the foreground image and the background image. The connected domain of the foreground image is extracted to obtain the horizontal and vertical gradients of each pixel in the connected domain. The standard deviation of the gradient amplitude of each pixel is used as the clarity of the connected domain. The absolute value of the difference between the standard deviations of the pixel values ​​in the connected domain and the background area is taken as the brightness difference, the ratio between the average values ​​of the gradient amplitudes in the connected domain and the background area is taken as the amplitude difference, the ratio between the number of pixels in the connected domain and the background area is taken as the area ratio, and the product of the brightness difference, the amplitude difference and the area ratio is taken as the contrast of the connected domain; The surface image with the maximum product of the contrast and the clarity is taken as the target image.

[0011] The effect is that through Otsu's segmentation, the foreground and background can be effectively separated, and the connected domain of the foreground area can be extracted. The standard deviation of the gradient amplitude of each pixel in the connected domain is calculated as the clarity, which helps to identify and select images with clear texture and rich details.

[0012] Preferably, the calculating deformation scores of different regions includes: The different regions include: a concave-convex region and a smooth region; A depth co-occurrence matrix is ​​constructed based on the depth data of different regions. According to the depth co-occurrence matrix, all depth data pairs are traversed, the average gradient amplitude of each depth data pair in different regions is calculated, the probability of occurrence of each depth data pair in different regions, and the ratio of the absolute difference of the depth values ​​of the depth data pairs in different regions to the average gradient amplitude of all pixel points are obtained, and the product of the average value, the probability and the ratio is summed as the deformation score of the different regions.

[0013] The effect is that by analyzing the gradient amplitude and depth changes, the tiny damage and deformation of the metal insulation layer can be detected more sensitively, thereby improving the accuracy of impact resistance detection.

[0014] Preferably, obtaining the maximum deformation area includes: The array whose center point completely contains the growth area is selected as the initial array, wherein the initial array is 2 2. Calculate the mean and standard deviation of the depth values ​​of all pixels in the neighborhood of the initial array 8, use the standard deviation as the growth threshold of regional growth, perform regional growth from the peak point, add points that meet the growth threshold to the growth area, and stop growing until there are no points with the standard deviation that meet the growth threshold in the area, and obtain the maximum deformation area.

[0015] The effect is that by calculating the average and standard deviation of the depth values ​​of the pixels in the neighborhood of the initial array 8 and using the standard deviation as the growth threshold, it helps to more accurately obtain the degree and distribution of deformation; and it can dynamically adapt to different degrees of damage. In areas with less damage, the growth threshold may be higher, and a smaller deformation area is obtained; while in areas with more serious damage, the growth threshold may be lower, and a larger deformation area is obtained.

[0016] Preferably, segmenting the temperature abnormality area according to the temperature data of the target image includes: The pixel point with the smallest temperature difference between the cold surface and the hot surface in different area tests in the target image is selected as the center point of region growing, and the temperature anomaly areas in different areas are obtained by using adaptive region growing.

[0017] Preferably, the temperature anomaly fraction satisfies the following relationship: ; In the formula, represents the temperature anomaly fraction, , Respectively represent the ratio of the number of pixels in the same position to the number of pixels in different positions in the maximum deformation area and the temperature anomaly area in the concave and convex area and the smooth area, , Represents the standard deviation of the temperature values ​​of the pixels at the same position in the maximum deformation area and the temperature anomaly area in the concave and convex area and the smooth area respectively.

[0018] Preferably, the verification and evaluation of impact resistance includes: In response to the temperature anomaly score being greater than or equal to a preset threshold, it indicates that the metal insulation layer has a poor impact resistance, otherwise, it indicates that the metal insulation layer has a good impact resistance.

[0019] The present invention has the following effects: 1. The present invention can comprehensively capture the deformation and temperature anomaly of the metal insulation layer by comprehensively utilizing surface images, depth data and temperature data; through transformation correction and regional growing technology, the maximum deformation area can be accurately located, thereby improving the accuracy of impact resistance detection; by calculating the comprehensive deformation score and temperature anomaly score, the damaged area can be identified, and the severity of the damage can also be evaluated.

[0020] 2. The present invention can more reliably identify damaged areas by combining deformation and temperature data. Temperature difference verification not only solves the damage that is not obvious when evaluated by deformation score alone, thereby improving the reliability of the overall detection, but also analyzes the situation where deformation occurs but is not damaged due to high temperature shock, and can further evaluate the impact resistance of the maximum deformation area, thereby improving the comprehensiveness of the evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein: Figure 1 It is a method flow chart of steps S1 to S4 in the monitoring method based on the metal insulation layer impact resistance detection process in an embodiment of the present invention.

[0022] Figure 2 It is a surface image of the metal thermal insulation layer in the monitoring method based on the metal thermal insulation layer impact resistance detection process in an embodiment of the present invention.

[0023] Figure 3 It is a schematic diagram of impact resistance detection in a monitoring method based on a metal insulation layer impact resistance detection process according to an embodiment of the present invention. DETAILED DESCRIPTION

[0024] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0025] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0026] Specific implementation scenario: refer to Figure 3Under different lighting conditions, five high-definition cameras a, b, c, d, and e are set according to different positions. The specific layout scene can be adjusted according to the detection environment. A high-resolution camera is used to shoot the metal insulation layer L at multiple angles. The preset angle is 15° to obtain surface images of the metal insulation layer at multiple angles. An infrared thermal imaging camera C` is used to detect the thermal distribution of the metal insulation layer to obtain the temperature data of the metal insulation layer. The position of the depth camera is consistent with the position of the camera for shooting the surface image of the metal insulation layer at multiple angles. The depth data of the metal insulation layer L obtained during shooting is obtained by the depth camera vertically. It should be noted that the surface image and depth data are obtained on the front side of the impact detection, and the temperature data is obtained on the back side of the impact detection.

[0027] Reference Figure 1 The monitoring method based on the metal insulation layer impact resistance detection process includes steps S1 to S4, which are as follows: It should be noted that when the metal insulation layer is subjected to the LOCA impact test, different high-temperature and high-pressure steam shocks and impact times will cause the shape of the metal insulation layer to undergo various changes. The traditional single light intensity and shooting angle cannot avoid the accidental errors caused by the reduced visibility or invisibility of the surface changes of the metal insulation layer. To solve this problem, a high-resolution camera is used to capture multi-angle surface images of the metal insulation layer under different lighting conditions. An infrared thermal imaging camera is used to obtain the temperature data of the cold surface, hot surface and side of the metal insulation layer, and a depth camera is used to capture the depth data of the cold surface, hot surface and side of the metal insulation layer at different positions. For each high-temperature and high-pressure steam test, the surface image, depth data and temperature data of the metal insulation layer from before the start of the test to the end of the test are collected at a fixed acquisition frequency.

[0028] S1: Obtain a surface image, depth data and temperature data of a metal insulation layer at a preset angle, obtain a grayscale image of the surface image at a preset angle, perform transformation correction, and screen a target image.

[0029] Further explanation: the metal insulation layer is different from the previous metal insulation layer structure. Figure 2 The metal insulation layer structure of the present invention is to evenly arrange semicircular protrusions on the metal surface, and regularly arrange them in a row of protrusions and a row of depressions, the purpose of which is to enhance the mechanical strength of the metal insulation layer so that it can better resist external impact and pressure; the alternating arrangement of protrusions and depressions can serve as an energy absorption area. When impacted, this structure can more effectively absorb and disperse the impact energy and reduce damage to the insulation layer; the regular protrusion and depression structure helps to alleviate thermal stress caused by temperature changes, because this structure can distribute heat more evenly and reduce the impact of thermal expansion and contraction.

[0030] In this embodiment, the surface image of the metal insulation layer is converted into a grayscale image, and the grayscale image is denoised using Gaussian filtering, and then the following operation steps are performed.

[0031] The angle perpendicular to the surface image is taken as the standard viewing angle, and the perspective transformation matrix of the surface images taken at different preset angles relative to the standard viewing angle is calculated. Based on the perspective transformation matrix, the coordinates of each pixel point are mapped from the surface image at the original preset angle to the standard viewing angle to obtain a mapped image, and bilinear interpolation is used to calculate the pixel values ​​of the pixel points of the mapped image to obtain corrected grayscale images of images at different angles.

[0032] The rectified grayscale image is segmented using the Otsu method to obtain the foreground image and the background image. The connected domain of the foreground image is extracted to obtain the horizontal and vertical gradients of each pixel in the connected domain. The standard deviation of the gradient amplitude of each pixel is used as the clarity of the connected domain. Exemplarily, the gradients of the target pixel in the horizontal and vertical directions are obtained, and the specific steps are as follows: Take any connected domain in the surface image of the standard view as the target connected domain, take any pixel point in the target connected domain as the target pixel point, calculate the average difference between the pixel values ​​of the target pixel point and the adjacent pixels in the horizontal direction of the target connected domain, and get the average difference between the target pixel point and the adjacent pixels in the horizontal direction. The gradient in the vertical direction is calculated by calculating the average difference between the pixel values ​​of the target pixel and the upper and lower adjacent pixels in the vertical direction of the target connected domain, and the vertical gradient of the target pixel is obtained. Directional gradient; Specifically, the gradient of the target pixel in the horizontal direction satisfies the following relationship: ; In the formula, Indicates Pixels in The gradient in direction, , Respectively represent Pixels in The pixel values ​​of the pixels before and after in the direction.

[0033] The gradient amplitude of the target pixel is further calculated to satisfy the following relationship: ; In the formula, Indicates The gradient amplitude of each pixel is Indicates Pixels in The gradient in direction, Indicates Pixels in Directional gradient.

[0034] The standard deviation of the gradient magnitude is taken as the clarity of the target connected domain.

[0035] That is to say, the greater the clarity of the connected domain, the clearer the image, and vice versa, the smaller the clarity, the blurrier the image.

[0036] The absolute value of the difference between the standard deviations of the pixel values ​​in the connected domain and the background area is taken as the brightness difference, the ratio between the average values ​​of the gradient amplitudes in the connected domain and the background area is taken as the amplitude difference, the ratio between the number of pixels in the connected domain and the background area is taken as the area ratio, and the product of the brightness difference, the amplitude difference and the area ratio is taken as the contrast of the connected domain; Specifically, the contrast satisfies the following relationship: ; In the formula, Indicates The contrast of the connected domains, Indicates The standard deviation of the pixel values ​​of the connected components, Represents the standard deviation of pixel values ​​in the background area, Indicates The average gradient amplitude of the connected domains, represents the average gradient amplitude of the background area, Indicates The number of pixels in a connected domain, Indicates the number of pixels in the background area.

[0037] That is to say, The larger the value is, the greater the brightness difference between the connected domain and the background area, and the more obvious the contrast is. The larger the contrast is, the more obvious the defects in the connected domain and the background area are. Conversely, the smaller the contrast is, the more blurred the defects are.

[0038] The surface image with the maximum product of contrast and clarity is taken as the target image.

[0039] In addition, in another embodiment, any pixel in the grayscale image is counted The pixel value is , the preset distances in the directions of 0 degrees, 45 degrees, 90 degrees, and 135 degrees are point The pixel value is Probability, statistics The frequency of occurrence is obtained to obtain the gray-level co-occurrence matrix; the maximum product of entropy and energy in the gray-level co-occurrence matrix is ​​calculated as the target image.

[0040] By selecting the image with the largest product of entropy and energy as the target image, it can be ensured that the selected image has the richest texture information, thereby improving the reliability of target detection and providing a strong guarantee for subsequent analysis.

[0041] S2: Divide the target image into regions according to the structural characteristics of the metal insulation layer, calculate the deformation scores of different regions based on the depth data of different regions, and take the number and area of ​​deformation in different regions as weights for weighted summation to obtain the comprehensive deformation score.

[0042] The different regions include: concave and convex regions and smooth regions; A depth co-occurrence matrix is ​​constructed based on the depth data of different regions. According to the depth co-occurrence matrix, all depth data pairs are traversed, the average gradient amplitude of each depth data pair in different regions is calculated, the probability of occurrence of each depth data pair in different regions, and the ratio of the absolute difference of the depth values ​​of the depth data pairs in different regions to the average gradient amplitude of all pixels are obtained, and the product of the average value, probability and ratio is summed as the deformation score of different regions.

[0043] Specifically, the average value of the gradient amplitude satisfies the following relationship: ; In the formula, Indicates depth value and The average value of the gradient amplitude is Indicates that all depth values ​​are The average gradient amplitude of the pixel points is Indicates that all depth values ​​are The average gradient amplitude of the pixel points.

[0044] Specifically, the deformation fraction of the concave-convex area satisfies the following relationship: ; In the formula, represents the deformation fraction of the concave and convex area, Indicates depth value and The probability of appearing adjacent to each other is, Indicates depth value and The average value of the gradient amplitude is Represents the average value of the gradient amplitude of all pixels in the concave-convex area. Indicates the maximum depth value within the bump area.

[0045] That is to say, The larger the value, the greater the deformation of the concave and convex area. The larger the depth value, the larger the gradient magnitude. The greater the probability of occurrence, the greater the degree of deformation.

[0046] Specifically, the deformation score of the smooth area satisfies the following relationship: ; In the formula, represents the deformation fraction of the smooth area, Indicates depth value and The probability of appearing adjacent to each other is, Indicates depth value and The average value of the gradient amplitude is Represents the average gradient amplitude of all pixels in the smooth area. Indicates the maximum depth value within the smoothing area.

[0047] In addition, in another embodiment, the deformation scores of the concave-convex area and the smooth area are calculated using the local binary pattern (LBP) and the gradient amplitude, and the specific steps are as follows: Taking any pixel point in the target image as the center, a preset window is constructed, and each pixel point is compared with each pixel in the neighborhood. In response to the neighborhood pixel value being greater than or equal to the value of the central pixel, a binary value of 1 is assigned; in response to the neighborhood pixel value being less than the value of the central pixel, a binary value of 0 is assigned, and the LBP value of each pixel point is obtained, and the gradient amplitude in the vertical and horizontal directions of each pixel point is calculated; Specifically, the deformation fraction of the concave-convex area satisfies the following relationship: ; In the formula, represents the deformation fraction of the concave and convex area, Indicates the number of pixels in the horizontal direction in the concave-convex area. Indicates the number of pixels in the vertical direction in the concave-convex area. Indicates Line The mean value of the gradient amplitude between the pixel point and the neighborhood. Indicates Line Column pixels value, Represents all pixels in the concave and convex area The average value of Indicates Line The gradient magnitude of the pixel column Similarly, the deformation score of the smooth area is obtained.

[0048] Get the comprehensive deformation score. The specific steps are as follows: According to the difference in the experimental depths before and after the concave-convex area and the smooth area, the number of concave-convex points deformed in the concave-convex area and the area deformed in the smooth area are obtained; Specifically, the number of the concave and convex points satisfies the following relationship: ; In the formula, Indicates the number of concave and convex points where deformation occurs in the concave and convex area during the test, Indicates the initial depth of a concave or convex point. Indicates the depth of a concave and convex point test, Represents a sum function.

[0049] That is to say, when a concave or convex point Then this concave-convex point is a deformation concave-convex point.

[0050] Specifically, the area of ​​the smooth region where the deformation occurs satisfies the following relationship: ; In the formula, It represents the area of ​​deformation of the smooth area during the test. Indicates the initial depth of a pixel in the smooth area, Indicates the depth of the pixel during the test. Represents a sum function.

[0051] That is, when a pixel , then this pixel is deformed during the test, and the total number of deformed pixels represents the deformed area of ​​the smooth region.

[0052] The ratio of the concave and convex points that are deformed in the concave and convex area during the test to the total number of concave and convex points before the test is taken as the weight of the deformation score of the concave and convex area, and the ratio of the area of ​​the smooth area that is deformed during the calculation test to the total area of ​​the smooth area before the test is taken as the weight of the deformation score of the smooth area. The weighted summation is performed to obtain the comprehensive deformation score.

[0053] Specifically, the comprehensive deformation score satisfies the following relationship: ; In the formula, represents the comprehensive deformation fraction of the metal insulation layer, Indicates the number of concave and convex points in the concave and convex area before the test, Indicates the number of concave and convex areas that deform during the test, represents the deformation fraction of the concave and convex area, represents the area of ​​the smooth region before the test, It represents the area of ​​the smooth area deformed during the test. Represents the deformation fraction of the smooth area.

[0054] In this embodiment, the comprehensive deformation score is normalized to ensure that the value range of the comprehensive deformation score is within between.

[0055] S3: In response to the comprehensive deformation score being greater than a preset threshold, the pixel point with the largest depth difference before and after detection in the target image is extracted as the center point of region growth, and region growth is performed to obtain the maximum deformation region, and the temperature abnormality region is segmented according to the temperature data of the target image.

[0056] In this embodiment, the preset threshold is 0.6, which can be adjusted according to actual conditions.

[0057] Get the maximum deformation area, including: The array whose center point completely contains the growth area is selected as the initial array, where the initial array is 2 2. Calculate the mean and standard deviation of the depth values ​​of all pixels in the neighborhood of the initial array 8, use the standard deviation as the growth threshold of regional growth, perform regional growth from the peak point, add points that meet the growth threshold to the growth area, and stop growing until there are no points with the standard deviation that meet the growth threshold in the area, and obtain the maximum deformation area.

[0058] The pixel point with the smallest temperature difference between the cold surface and the hot surface in different area tests in the target image is selected as the center point of region growing, and the temperature anomaly areas in different areas are obtained by using adaptive region growing.

[0059] That is to say, the smaller the temperature difference between the cold surface and the hot surface, the greater the probability of the metal insulation layer breaking. When it is equal to 0, it means that the metal insulation layer is perforated.

[0060] S4: Match the maximum deformation area with the temperature anomaly area to obtain the temperature anomaly score, and perform temperature difference verification on the maximum deformation area of ​​the target image to evaluate the impact resistance.

[0061] Specifically, the temperature anomaly fraction satisfies the following relationship: ; In the formula, represents the temperature anomaly fraction, , Respectively represent the ratio of the number of pixels in the same position to the number of pixels in different positions in the maximum deformation area and the temperature anomaly area in the concave and convex area and the smooth area, , Represents the standard deviation of the temperature values ​​of the pixels at the same position in the maximum deformation area and the temperature anomaly area in the concave and convex area and the smooth area respectively.

[0062] That is to say, The larger the value, the closer the area with the largest deformation in the concave-convex area matches the area with the largest temperature anomaly in the concave-convex area, and the better the impact resistance of the insulation layer. Same as above.

[0063] In addition, in another embodiment, based on the depth data of the target image, the target image is binarized and segmented using the OTSU method to obtain the connected domain with the maximum deformation degree. Similarly, based on the temperature data of the target image, the connected domain with the maximum temperature anomaly is obtained. The two connected domains are positionally matched, and the temperature standard deviation of the pixels at the same position is calculated to evaluate the impact resistance.

[0064] In response to the temperature anomaly score being greater than or equal to a preset threshold, it indicates that the metal insulation layer has a poor impact resistance, otherwise, it indicates that the metal insulation layer has a good impact resistance.

[0065] In this embodiment, the temperature anomaly score is normalized, wherein the preset threshold is 0.8, which can be adjusted according to actual conditions.

[0066] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0067] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. Monitoring method based on metal insulation layer impact resistance detection process, It is characterized in that include: Obtaining a surface image, depth data and temperature data of a metal insulation layer at a preset angle, obtaining a grayscale image of the surface image at a preset angle, performing transformation correction, and screening a target image; The target image is divided into regions according to the structural characteristics of the metal insulation layer, and the deformation scores of different regions are calculated based on the depth data of different regions. The number and area of ​​deformation in different regions are used as weights for weighted summation to obtain a comprehensive deformation score. In response to the comprehensive deformation score being greater than a preset threshold, the pixel point with the largest depth difference before and after detection in the target image is extracted as the center point of region growth, and region growth is performed to obtain the maximum deformation region, and the temperature abnormality region is segmented according to the temperature data of the target image; Match the maximum deformation area with the temperature anomaly area to obtain the temperature anomaly score , and verify the impact resistance by performing temperature difference verification on the maximum deformation area of ​​the target image; , where , Respectively represent the ratio of the number of pixels in the same position to the number of pixels in different positions in the maximum deformation area and the temperature anomaly area in the concave and convex area and the smooth area, , The standard deviations of the temperature values ​​of the pixels at the same position in the maximum deformation area and the temperature anomaly area in the concave-convex area and the smooth area, respectively; Among them, the comprehensive deformation score Satisfies the following relationship: , where , They represent the number of concave and convex points in the concave and convex area and the area of ​​the smooth area before the test, , They represent the number of concave and convex points in the concave and convex area and the area of ​​deformation in the smooth area during the test. , Represent the deformation fractions of the concave and convex areas and the smooth area respectively.

2. The monitoring method based on the metal insulation layer impact resistance detection process according to claim 1, It is characterized in that The step of obtaining a grayscale image of a surface image at a preset angle and performing transformation correction includes: The angle perpendicular to the surface image is taken as the standard viewing angle, and the perspective transformation matrix of the surface images taken at different preset angles relative to the standard viewing angle is calculated. Based on the perspective transformation matrix, the coordinates of each pixel point are mapped from the surface image at the original preset angle to the standard viewing angle to obtain a mapped image, and bilinear interpolation is used to calculate the pixel values ​​of the pixel points of the mapped image to obtain corrected grayscale images of images at different angles.

3. The monitoring method based on the metal insulation layer impact resistance detection process according to claim 1, It is characterized in that The screening target image comprises: The rectified grayscale image is segmented using the Otsu method to obtain the foreground image and the background image. The connected domain of the foreground image is extracted to obtain the horizontal and vertical gradients of each pixel in the connected domain. The standard deviation of the gradient amplitude of each pixel is used as the clarity of the connected domain. The absolute value of the difference between the standard deviations of the pixel values ​​in the connected domain and the background area is taken as the brightness difference, the ratio between the average values ​​of the gradient amplitudes in the connected domain and the background area is taken as the amplitude difference, the ratio between the number of pixels in the connected domain and the background area is taken as the area ratio, and the product of the brightness difference, the amplitude difference and the area ratio is taken as the contrast of the connected domain; The surface image with the maximum product of the contrast and the clarity is taken as the target image.

4. The monitoring method based on the metal insulation layer impact resistance detection process according to claim 1, It is characterized in that The calculation of deformation scores of different regions includes: The different regions include: a concave-convex region and a smooth region; A depth co-occurrence matrix is ​​constructed based on the depth data of different regions. According to the depth co-occurrence matrix, all depth data pairs are traversed, the average gradient amplitude of each depth data pair in different regions is calculated, the probability of occurrence of each depth data pair in different regions, and the ratio of the absolute difference of the depth values ​​of the depth data pairs in different regions to the average gradient amplitude of all pixel points are obtained, and the product of the average value, the probability and the ratio is summed as the deformation score of the different regions.

5. The monitoring method based on the metal insulation layer impact resistance detection process according to claim 1, It is characterized in that Obtaining the maximum deformation area includes: The array whose center point completely contains the growth area is selected as the initial array, wherein the initial array is 2 2. Calculate the mean and standard deviation of the depth values ​​of all pixels in the neighborhood of the initial array 8, use the standard deviation as the growth threshold of regional growth, perform regional growth from the peak point, add points that meet the growth threshold to the growth area, and stop growing until there are no points with the standard deviation that meet the growth threshold in the area, and obtain the maximum deformation area.

6. The monitoring method based on the metal insulation layer impact resistance detection process according to claim 1, It is characterized in that The step of segmenting the temperature abnormality area according to the temperature data of the target image includes: The pixel point with the smallest temperature difference between the cold surface and the hot surface in different area tests in the target image is selected as the center point of region growing, and the temperature anomaly areas in different areas are obtained by using adaptive region growing.

7. The monitoring method based on the metal insulation layer impact resistance detection process according to claim 1, It is characterized in that The verification evaluates the impact resistance, including: In response to the temperature anomaly score being greater than or equal to a preset threshold, it indicates that the metal insulation layer has a poor impact resistance, otherwise, it indicates that the metal insulation layer has a good impact resistance.

Images