Insulation Crack Detection Method Based on Infrared Thermal Imaging Analysis

Abstract

This application relates to the field of image processing technology, specifically to a method for detecting thermal insulation cracks based on infrared thermal imaging analysis. The method includes: acquiring temperature, humidity, and wind speed data; obtaining a grayscale image; segmenting the grayscale image, determining texture similarity and positional similarity based on the texture features of image blocks, obtaining anomaly coefficients based on both, and thus identifying abnormal regions; obtaining crack feature coefficients based on the correlation between grayscale and temperature, humidity, and wind speed, the shape of the abnormal region, and the grayscale difference between the center and periphery, and thus screening crack regions; fitting a crack function and a midline function to the crack region, determining the objective function based on the integral area and the location of the extreme points of both, and then iteratively obtaining the set of optimal points to determine the crack location. This application improves the accuracy of crack detection and achieves crack localization.

Description

Technical Field

[0001] This application relates to the field of image processing technology, specifically to a method for detecting thermal insulation cracks based on infrared thermal imaging analysis. Background Technology

[0002] Building exterior wall insulation systems are crucial for reducing heat loss, enhancing insulation capacity, and mitigating the effects of thermal bridging. However, over time and due to external environmental factors, defects such as cracks and hollow areas may appear in the exterior insulation layer. These defects not only weaken the insulation layer's function but can also lead to insulation material detachment, causing hazards. Therefore, defect detection in exterior insulation layers is essential, with cracks being one of the most common defects. Because cracks occur inside the insulation layer, their characteristics are more concealed compared to defects like hollow areas, making crack detection a challenge. With the introduction of infrared thermal imaging into the field of building engineering, it has gained widespread attention due to its non-contact, non-destructive testing advantages and is gradually becoming an important tool.

[0003] While infrared thermal imaging technology has shown great potential in detecting defects in external wall insulation layers, it has also revealed some limitations in practical applications. Infrared thermal imaging detects defects based on differences in heat radiation at different locations on the external insulation layer. However, changes in heat radiation are highly susceptible to environmental conditions. Numerous studies have demonstrated the significant impact of environmental changes on crack detection using infrared thermal imaging. Specifically, changes in environmental factors can lead to complex temperature distribution patterns on the insulation layer surface, interfering with the local temperature difference signal caused by cracks and increasing the risk of misjudgment. Furthermore, heat conduction can cause the temperature field near cracks to diffuse, further obscuring the true boundaries of the cracks. These problems result in significant errors in crack detection based on infrared thermal imaging and make it difficult to pinpoint the location of crack defects. Therefore, there is an urgent need to develop a more accurate infrared thermal imaging crack detection method to overcome these challenges and improve the reliability and accuracy of detection. Summary of the Invention

[0004] To address the technical problem of low accuracy in crack detection, this application provides a method for detecting thermal insulation cracks based on infrared thermal imaging analysis. The specific technical solution adopted is as follows:

[0005] This application proposes a method for detecting thermal insulation cracks based on infrared thermal imaging analysis, which includes the following steps:

[0006] Infrared images, temperature, humidity, and wind speed are collected at each preset time; the infrared images are denoised and converted to grayscale to obtain grayscale images;

[0007] The grayscale image is segmented into several image blocks based on grayscale, and the texture features of the image blocks are obtained. The first grayscale image is recorded as the original image. The texture similarity between image blocks is calculated based on the texture features between the image blocks. The matching image block is determined based on the maximum value of the texture similarity between different images. The positional similarity is determined based on the coordinates of the pixels of the image block and the matching image block. The anomaly coefficient of the image block is determined based on the texture similarity between the image block and its neighboring image blocks and the positional similarity between the image block and the matching image block. The anomaly region is determined based on the anomaly coefficient.

[0008] The center set and edge set are screened based on the minimum bounding rectangle of the abnormal region; the correlation factor is determined based on the correlation between gray level and temperature, humidity and wind speed at all times; the crack feature coefficient is determined based on the gray level difference between the center and the periphery of the abnormal region, the aspect ratio of the minimum bounding rectangle of the abnormal region and the correlation factor, and the crack region is screened based on the crack feature coefficient.

[0009] A predetermined number of extreme points are selected along the width of the minimum bounding rectangle in the crack region as initial extreme points. The crack function is obtained by fitting the initial extreme points, and the function of the midline of the minimum bounding rectangle is denoted as the midline function. The target function is determined by the integral area of ​​the crack function and the midline function of all grayscale images and the difference in the position of the extreme points after adjustment. The set of optimal points of the original image is obtained by iterative adjustment based on the target function. The crack position is determined by fitting the set of optimal points.

[0010] In the above scheme, this application first calculates the texture change coefficient based on the grayscale and texture features in the infrared image. This index is used to initially screen texture anomaly areas, which helps to narrow the detection range and improve the efficiency and accuracy of subsequent detection. Then, based on the characteristics of the external insulation layer crack defects and the texture change of the anomaly area, the crack feature coefficient is calculated. This index further determines the crack area and enhances the accuracy of crack identification. Finally, the crack is located in the crack area through curve fitting and optimization algorithms. By using this method, the crack area is determined based on the temporal change of texture features in the infrared image and the relationship between the temporal change features of the infrared image and environmental conditions. This reduces the influence of environmental conditions, improves the accuracy of crack detection, and finally locates the crack in the crack area, thus realizing an accurate method for detecting insulation cracks based on infrared thermal imaging analysis and precise crack location.

[0011] In one embodiment, the method for obtaining the texture features is as follows:

[0012] The LBP value of each pixel is obtained through the LBP algorithm. For the segmented grayscale image, the LBP values ​​of all pixels in each segmented image block are used to form an LBP histogram. The LBP histogram of each image block represents the texture features of each image block.

[0013] In one embodiment, the texture similarity is the reciprocal of the Bach distance between the LBP histograms of the image patches. The positional similarity is the number of pixels with overlapping coordinates in the two image patches.

[0014] In one embodiment, the anomaly coefficient is negatively correlated with the texture similarity between an image patch and its neighboring image patches, and positively correlated with the positional similarity between an image patch and a matching image patch.

[0015] In one embodiment, the center set is the set of pixels within the eight neighborhoods of the center point, and the edge set is the set of pixels on the outermost periphery of the abnormal region.

[0016] In one embodiment, the method for obtaining the correlation metric is as follows:

[0017] For each anomalous region, the corresponding region is obtained from the remaining infrared images based on the coordinates of the anomalous region. The mean gray value of the corresponding region is calculated. The mean gray values ​​of the anomalous region and its corresponding region are sorted in time sequence to obtain a gray-scale time series. Then, temperature, humidity, and wind speed are sorted in time sequence to obtain temperature, humidity, and wind speed series, respectively. The correlation coefficients of the gray-scale time series with the temperature, humidity, and wind speed series are used as the correlation measure between gray-scale and temperature, humidity, and wind speed.

[0018] In one embodiment, the expression for the correlation factor is:

[0019] , This measures the correlation between grayscale and humidity. This measures the correlation between grayscale and temperature. This measures the correlation between grayscale and wind speed. Indicates the relevant factors.

[0020] In one embodiment, the crack characteristic coefficient is positively correlated with the gray-scale difference between the center and periphery of the crack region and related factors, and negatively correlated with the aspect ratio of the minimum bounding rectangle of the abnormal region.

[0021] In one embodiment, the method of taking a preset number of extreme points as initial extreme points along the width of the minimum circumscribed rectangle of the crack region is as follows:

[0022] Take the direction corresponding to the short side of the smallest bounding rectangle as the secondary direction; uniformly select a preset number of secondary direction edges in the crack region, and take the pixel point corresponding to the maximum and minimum gray values ​​on each edge as the initial maximum and minimum value point.

[0023] In one embodiment, the expression for the objective function is:

[0024] , Let represent the sum of the integral areas between all crack functions and midline functions in the t-th grayscale image. Let represent the sum of the distances before and after adjustment for all extreme points in the t-th grayscale image. The number of grayscale images, This represents the objective function.

[0025] The beneficial effects of this application are as follows:

[0026] This application first calculates the texture change coefficient based on the grayscale and texture features in infrared images. This index is used for initial screening of texture anomaly areas, helping to narrow down the detection range and improve the efficiency and accuracy of subsequent detection. Then, based on the characteristics of crack defects in the external insulation layer and the texture changes in the anomaly areas, a crack feature coefficient is calculated. This index further clarifies the crack region, enhancing the accuracy of crack identification. Finally, crack localization is achieved in the crack region through curve fitting and optimization algorithms. This method, by determining the crack region based on the temporal changes of texture features in infrared images and the relationship between these temporal changes and environmental conditions, reduces the influence of environmental conditions, improves the accuracy of crack detection, and finally achieves crack localization within the crack region. This results in an accurate method for detecting insulation cracks based on infrared thermal imaging analysis, and precise crack localization. Attached Figure Description

[0027] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0028] Figure 1 This is a flowchart of a thermal insulation crack detection method based on infrared thermal imaging analysis provided in one embodiment of this application. Detailed Implementation

[0029] To further illustrate the technical means and effects adopted by this application to achieve the intended purpose of the invention, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the insulation crack detection method based on infrared thermal imaging analysis proposed in this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0030] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0031] Example of a method for detecting thermal insulation cracks based on infrared thermal imaging analysis:

[0032] The following description, in conjunction with the accompanying drawings, details the specific scheme of the insulation crack detection method based on infrared thermal imaging analysis provided in this application.

[0033] Please see Figure 1 The diagram illustrates a flowchart of a method for detecting thermal insulation cracks based on infrared thermal imaging analysis according to an embodiment of this application. The method includes the following steps:

[0034] Step S001: Collect temperature, humidity, and wind speed data; and acquire a grayscale image.

[0035] Infrared images of the building's external insulation layer were acquired using an infrared thermal imager, where the temperature measurement range of the infrared thermal imager was [missing information]. The resolution is Temperature measurement accuracy is The acquired infrared image is used as input, and image denoising algorithms and color space conversion are applied sequentially to obtain a denoised grayscale image. Image denoising algorithms and color space conversion are well-known techniques and will not be described in detail here. In different embodiments, image denoising may employ, but is not limited to, median filtering, mean filtering, and non-local mean filtering. In this embodiment, median filtering is used for image denoising. Subsequently, temperature, humidity, and wind speed at the infrared thermal imaging detection site are collected using temperature sensors, humidity sensors, and an anemometer. These three data are recorded as environmental data.

[0036] In order to accurately capture temporal variation characteristics and improve detection accuracy when performing infrared thermal imaging crack detection of external insulation layers, it is necessary to... Infrared images were acquired at different times. In this embodiment... To ensure sufficient difference between different infrared images, the time interval between two adjacent acquisitions is more than 1 hour.

[0037] At this point, grayscale images and data of different data types at each time point have been obtained.

[0038] Step S002: After segmenting the grayscale image, determine the texture similarity and positional similarity based on the texture features of the image blocks, obtain the anomaly coefficient based on both, and then determine the abnormal region.

[0039] External insulation layers, as a crucial structural component of buildings, play a vital role in thermal insulation, waterproofing, and moisture protection. Therefore, defect detection in external insulation layers has become a hot topic in the construction engineering field. Among numerous external insulation layer defects, cracks pose a significant challenge due to their internal location and relatively concealed external features. While infrared thermal imaging technology can theoretically detect cracks based on temperature changes, practical applications still face difficulties. Infrared thermal imaging is significantly affected by environmental factors, including temperature variations, leading to misjudgments and missed detections of cracks. Furthermore, heat conduction causes temperature fluctuations around cracks, further complicating accurate crack location. Therefore, further optimization of infrared thermal imaging-based crack detection is needed to improve detection accuracy and localization precision.

[0040] When defects exist on the surface of a building's external insulation layer, the thermal conductivity at the defect location differs significantly from that of the normal area, resulting in noticeable changes in texture in infrared thermal imaging images. However, besides various defects causing significant texture changes, dirt and shadows on the surface of the external insulation layer can also alter the thermal conductivity, leading to abnormal texture features. Therefore, to detect crack defects, the first step is to identify areas with abnormal texture changes in the infrared thermal imaging image.

[0041] Under normal circumstances, since the material of the outer insulation layer is consistent, the texture features of the outer insulation layer should be basically consistent when performing thermal infrared detection. However, when there are defects, dirt, or shadows in the outer insulation layer, the texture features will change. Among them, the texture feature abnormalities caused by defects and dirt and the areas where they appear are relatively stable. That is, when multiple tests are performed, there will be texture feature abnormalities in some fixed areas. However, the abnormalities caused by shadows are variable.

[0042] Using a grayscale image as input, the natural break method and region growing algorithm are employed to obtain segmented grayscale images. These two algorithms are well-known techniques and will not be elaborated upon further. The natural break method classifies grayscale values ​​based on their intensity, while the region growing algorithm uses a similarity criterion that grayscale values ​​of the same class can be added to the same region. Texture features of each segmented image block are then obtained from the grayscale image.

[0043] Preferably, in this embodiment, the LBP value of each pixel is obtained by the LBP algorithm. For the segmented grayscale image, the LBP values ​​of all pixels in each segmented image block are used to form an LBP histogram. The LBP histogram of each image block represents the texture features of each image block.

[0044] The grayscale image of the first infrared image taken during each detection is recorded as the original image.

[0045] Each image patch in the original image is designated as the target image patch. The texture similarity between image patches is calculated based on their texture features. The texture similarity between the target image patch and the remaining grayscale image patches (excluding the original image) is calculated, and the image patch with the highest texture similarity to the target image patch in each grayscale image is designated as the matching image patch. The positional similarity is determined based on the coordinates of pixels in the target image patch and the matching image patch.

[0046] Preferably, in this embodiment, the texture similarity is the reciprocal of the Bach distance between the LBP histograms of the image blocks. The positional similarity is the number of pixels with overlapping coordinates in two image blocks.

[0047] The anomaly coefficient of the target image patch is determined based on the texture similarity between the target image patch and its neighboring image patches, as well as the positional similarity between the target image patch and the matching image patch. The neighboring image patches refer to image patches where any two image patches have any adjacent pixels.

[0048] The anomaly coefficient is negatively correlated with the texture similarity between the target image block and its neighboring image blocks, and positively correlated with the positional similarity between the target image block and the matching image block.

[0049] It should be noted that positive correlation means that when one variable increases, the other variable also increases, and the two variables change in the same direction. When one variable changes from large to small or from small to large, the other variable also changes from large to small or from small to large. The specific relationship is determined by the actual application, and this application does not impose any special restrictions.

[0050] It should be noted that negative correlation means that when one variable increases, the other variable decreases accordingly, and the two variables change in opposite directions. When one variable changes from large to small or from small to large, the other variable also changes from small to large or from large to small. The specific relationship is determined by practical application, and this application does not impose any special restrictions.

[0051] Preferably, in this embodiment, the expression for the anomaly coefficient is:

[0052] , This indicates the number of adjacent image patches of the target image patch. This represents the texture similarity between the i-th image patch and the n-th image patch of the original image. This represents the positional similarity between the i-th image patch and the matching image patch of the t-th grayscale image; This indicates the number of grayscale images detected each time. This represents the anomaly coefficient of the i-th image block; the i-th image block is the target image block.

[0053] It is understandable that, for defects in the outer insulation layer, the texture changes in the defect area are significantly different from the surrounding area. That is, the greater the difference between the image block and the surrounding image blocks, the smaller the texture similarity, the larger the anomaly coefficient, and the more it indicates that the image block is an abnormal area. Since the location of the defect cannot be changed, by comparing the pixel positions of the image block with the highest texture similarity among different images with the target image block, the more overlapping the pixel positions, the greater the positional similarity, the larger the anomaly coefficient, and the more it indicates that the image block is an abnormal area.

[0054] Using the above method, the texture variation coefficient of each image patch in the original image is calculated. All anomaly coefficients are used as input, and Otsu thresholding is applied to output the anomaly threshold. Otsu thresholding is a well-known technique and will not be elaborated further. Image patches with anomaly coefficients greater than the anomaly threshold are marked as anomaly regions.

[0055] At this point, the abnormal regions of the original image have been obtained.

[0056] Step S003: Based on the correlation between grayscale and temperature, humidity and wind speed, the shape of the abnormal area and the grayscale difference between the center and the periphery, the crack feature coefficient is obtained, and then the crack area is screened.

[0057] After screening out abnormal areas through the above steps, since abnormal areas include more than just cracks and defects, further screening is needed in order to further determine the crack areas, taking into account the characteristics of cracks and defects in the external insulation layer.

[0058] First, the external insulation layer material is a thermal insulation material with relatively low thermal conductivity. Therefore, under normal circumstances, external heat radiation accumulates on the surface and quickly dissipates with airflow. However, when cracks appear, the medium within the crack is air, and air has a higher thermal conductivity than the insulation material. As heat is transferred into the crack, it accumulates continuously due to the lower thermal conductivity of the surrounding insulation layer, leading to a higher temperature. Second, although the insulation material around the crack has lower thermal conductivity, the continuous focusing and outward transfer of heat from the crack results in a higher temperature for the surrounding insulation layer, with the heat decreasing from the crack outwards. Third, when ambient humidity increases, moisture seeps into the insulation material through the cracks. Because water has a high specific heat capacity, it needs to absorb more heat to raise its temperature, causing cold spots to appear at the cracks, with the cold spots becoming more severe closer to the crack. Finally, changes in wind speed affect airflow over the surface of the external insulation layer, carrying away some heat. However, airflow inside the cracks is relatively limited, so wind speed has a relatively smaller impact on the cracks. Meanwhile, the area where the crack is located usually appears as a narrow strip.

[0059] For each abnormal region, the smallest bounding rectangle of the abnormal region is obtained, and the intersection of the diagonals is taken as the center point of the abnormal region. The pixels in the neighborhood around the center point constitute the center set of the abnormal region, and the pixels on the outermost edge of the abnormal region constitute the edge set of the abnormal region. In this embodiment, the center set is the set of pixels in the eight neighborhoods of the center point.

[0060] Based on the coordinates of the pixels in the abnormal region, the corresponding regions are determined in each infrared image, and the mean gray value of each region is calculated. After normalizing all the mean values, a gray-scale time sequence of the abnormal region is constructed according to the order of infrared image acquisition. Then, the temperature, humidity, and wind speed are normalized and constructed according to the order of infrared image acquisition to form temperature, humidity, and wind speed sequences.

[0061] The correlation between grayscale and temperature, humidity, and wind speed in the abnormal area is calculated, and a correlation factor is calculated based on the correlation measure. Because the surface of the external insulation layer accumulates less heat, it is highly susceptible to changes in temperature and wind speed, while the crack area is relatively less affected, resulting in a lower correlation between grayscale changes and temperature and wind speed in the crack area. Furthermore, moisture at the crack is difficult to evaporate, making the crack area more susceptible to changes in humidity. Therefore, the correlation factor is positively correlated with the correlation measures of grayscale and humidity, and negatively correlated with the correlation measures of grayscale and temperature and wind speed. The correlation measure is represented by the correlation coefficient between sequences. For example, the correlation measure between grayscale and temperature is the correlation coefficient between the grayscale time series and the temperature series. In this embodiment, the correlation coefficient is calculated using the Pearson correlation coefficient.

[0062] Preferably, in this embodiment, the expression for the relevant factor is:

[0063] , This measures the correlation between grayscale and humidity. This measures the correlation between grayscale and temperature. This measures the correlation between grayscale and wind speed. Indicates the relevant factors.

[0064] Because there is a significant heat change trend between the center and the periphery of the crack area, the grayscale difference between the center and the periphery is large; and the narrowness of the crack makes the area it is located in also present a certain strip-shaped region; therefore, by comparing the grayscale difference between the center and the periphery of the crack area, the shape of the crack, and combining the grayscale of the crack area with the correlation of environmental data, the crack characteristic coefficient can be determined, thereby determining whether the abnormal area is a crack area.

[0065] Therefore, the crack characteristic coefficients are determined based on the gray-scale difference between the center and periphery of the abnormal region, the aspect ratio of the minimum bounding rectangle of the abnormal region, and related factors.

[0066] The crack characteristic coefficient is positively correlated with the gray level difference between the center and periphery of the crack area and related factors, and negatively correlated with the aspect ratio of the minimum bounding rectangle of the abnormal area.

[0067] Preferably, in this embodiment, the expression for the crack characteristic coefficient is:

[0068] , This represents the average grayscale value of the pixels at the center of the abnormal region. This represents the average grayscale value of the pixels at the edge of the abnormal region. This represents the aspect ratio of the smallest bounding rectangle of the abnormal region. Indicates the relevant factors, and Indicates the first weighting factor and the second weighting factor. This represents an exponential function with the natural constant as its base. This represents the crack characteristic coefficient of the j-th anomalous region.

[0069] Due to the diversity of cracks and the extent of their influence on the surrounding area, the aspect ratio of the crack region may not be significantly different. Therefore, in this embodiment, the first weighting factor is 0.8 and the second weighting factor is 0.2.

[0070] Similarly, using the crack feature coefficients of all abnormal regions as input, Otsu thresholding is employed to output the crack threshold. Otsu thresholding is a well-known technique and will not be elaborated upon further. Abnormal regions with crack feature coefficients greater than the crack threshold are marked as crack regions.

[0071] At this point, the crack region has been obtained.

[0072] Step S004: Fit the crack function and the midline function to the crack region, determine the objective function based on the integral area and the location of the extreme points of the two, and then iteratively obtain the set of optimal points to determine the crack location.

[0073] Following the steps described above, each grayscale image is analyzed to obtain the crack region. For each crack region, based on the minimum bounding rectangle of the crack region, the direction corresponding to the long side of the rectangle is taken as the principal direction, and the direction perpendicular to the principal direction is taken as the secondary direction (the direction of the wide side of the rectangle). Uniformly sampled values ​​are taken within the crack region. The image extracts the edges in each of the following directions and takes the maximum and minimum grayscale values ​​for each edge. It should be noted that if the crack region is a hot spot, the minimum value is taken; if the crack region is a cold spot, the maximum value is taken. The pixel corresponding to the extracted maximum and minimum values ​​is recorded as the initial maximum and minimum value point. If the average grayscale value of the crack region is less than the average grayscale value of the entire grayscale image, the crack region is a cold spot; otherwise, it is a hot spot. In this embodiment... Take 10.

[0074] Polynomial fitting is performed on all initial extreme points, and the fitted curve function is output, denoted as the initial crack function. Polynomial fitting is a well-known technique and will not be elaborated further. Then, the median of the widest side of the smallest bounding rectangle of the crack region is taken, and the function of the line containing the median is calculated, denoted as the median function. In the same way, the initial crack function and median function of the crack region are determined in each grayscale image.

[0075] The initial extreme points, objective function, iteration step size, and maximum number of iterations in the crack regions of all grayscale images are taken as input. An optimization algorithm is used to iteratively output the adjusted coordinates of the extreme points. Finally, the algorithm stops when the optimal value is reached or the maximum number of iterations is reached, outputting the optimal combination of points. The optimization algorithm is a well-known technique and will not be described in detail here. In different implementations, the optimization algorithm can include, but is not limited to, gradient descent, Newton's method, and quasi-Newton methods. The iteration step size can be obtained using, but is not limited to, exact line search, Wolfe criterion, and Armijo criterion. In this embodiment, the maximum number of iterations is 100.

[0076] Since the goal of the optimization algorithm is to minimize the objective function, it can be understood that, theoretically, since the crack is close to the center of the entire region, the integral area between the crack function and the centerline function should be small. At the same time, since the gray value at the center of the crack should theoretically be the smallest, and the initial corresponding gray value is the smallest, when adjusting the extreme point, the distance of the extreme point should be moved as little as possible. Therefore, the smaller the corresponding objective function, the better.

[0077] Therefore, the objective function is determined based on the integral area between the crack function and the median function in all grayscale images, as well as the sum of the distances before and after the adjustment of the extremum point. The position of the extremum point before adjustment is the position of the initial extremum point.

[0078] Preferably, in this embodiment, the expression for the objective function is:

[0079] , Let represent the sum of the integral areas between all crack functions and midline functions in the t-th grayscale image. Let represent the sum of the distances before and after adjustment for all extreme points in the t-th grayscale image. The number of grayscale images, This represents the objective function.

[0080] The set of points in the original image corresponding to the minimum value of the objective function is denoted as the optimal point set. The points in the optimal point set are subjected to polynomial fitting, and the position corresponding to the fitted curve is the crack location. Thus, the crack detection and localization are realized.

[0081] It should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

[0082] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A heat-retention crack detection method based on infrared thermography analysis, characterized by, The method comprises the following steps: Collecting infrared images and temperature, humidity and wind speed at each time point in a preset time; denoising and gray scaling the infrared images to obtain gray images; Obtaining a plurality of image blocks based on gray segmentation of the gray images, and obtaining texture features of the image blocks; taking the first gray image as an original image, calculating texture similarity between the image blocks based on the texture features between the image blocks; determining a matching image block of the image block based on a maximum value of the texture similarity between different image blocks; determining position similarity based on coordinates of pixel points of the image block and the matching image block; determining an anomaly coefficient of the image block based on the texture similarity between the image block and its adjacent image blocks and the position similarity between the image block and the matching image block; determining an abnormal area based on the anomaly coefficient; Filtering a center set and an edge set based on a minimum circumscribed rectangle of the abnormal area; determining a correlation factor based on correlation between the gray values and the temperature, humidity and wind speed at all time points; determining a crack feature coefficient based on gray difference between the center and the periphery of the abnormal area, an aspect ratio of the minimum circumscribed rectangle of the abnormal area and the correlation factor, and filtering a crack area based on the crack feature coefficient; Taking a preset number of extreme points in a wide direction of a minimum circumscribed rectangle of the crack area as initial extreme points; fitting the initial extreme points to obtain a crack function, and taking a function of a center line of the minimum circumscribed rectangle as a center line function; determining a target function based on integral areas of the crack functions of all the gray images and the center line function and position difference after adjustment of the extreme points, and iteratively adjusting the target function to obtain an optimal point set of the original image; fitting the optimal point set to determine a crack position.

2. The heat-preservation crack detection method based on infrared thermography analysis according to claim 1, characterized by, The method for obtaining the texture features comprises the following steps: Obtaining an LBP value of each pixel point by an LBP algorithm; for the segmented gray image, constructing an LBP histogram of all the pixel points in each image block after segmentation, and taking the LBP histogram of each image block as the texture features of each image block.

3. The heat-preservation crack detection method based on infrared thermography analysis according to claim 2, characterized in that, The texture similarity is a reciprocal of a Bhattacharyya distance of the LBP histograms between the image blocks; and the position similarity is a number of pixel points with coinciding coordinates in the two image blocks.

4. The infrared thermography analysis based heat retention crack detection method according to claim 1, wherein, The anomaly coefficient is negatively correlated with the texture similarity between the image block and its adjacent image blocks, and is positively correlated with the position similarity between the image block and the matching image block.

5. The infrared thermography-based analysis method for curing crack detection according to claim 1, wherein, The center set is a set of pixel points in an eight-neighborhood of a center point, and the edge set is a set of pixel points at an outermost periphery of the abnormal area.

6. The infrared thermography-based analysis of thermal crack detection method as claimed in claim 1, wherein, The method for obtaining the correlation comprises the following steps: For each abnormal area, obtaining a corresponding area in the remaining infrared images according to coordinates of the abnormal area, calculating a mean value of gray values of the corresponding area, obtaining a gray time sequence by sorting mean values of the abnormal area and the corresponding area according to time sequences, and obtaining a temperature sequence, a humidity sequence and a wind speed sequence by sorting the temperature, humidity and wind speed according to time sequences; taking correlation coefficients of the gray time sequence with the temperature sequence, the humidity sequence and the wind speed sequence as correlation between the gray values and the temperature, humidity and wind speed.

7. The infrared thermography-based analysis method for curing crack detection according to claim 1, wherein, The expression of the correlation factor is , denotes a measure of correlation of the gray scale with the humidity, denotes a measure of correlation of the gray scale with the temperature, denotes a measure of correlation of the gray scale with the wind speed, denotes a correlation factor.

8. The infrared thermography-based analysis of curing crack detection method according to claim 1, characterized in that, The crack feature coefficient is positively correlated with the gray difference and the correlation factor of the center and the periphery of the crack area, and is negatively correlated with the length-width ratio of the minimum circumscribed rectangle of the abnormal area.

9. The infrared thermography-based analysis of curing crack detection method according to claim 1, characterized in that, The method of taking the preset number of extreme points in the direction of the long side of the minimum circumscribed rectangle of the crack area as initial extreme points is: Taking the direction corresponding to the short side of the minimum circumscribed rectangle as a sub-direction; evenly taking a preset number of edges of the sub-direction in the crack area, and taking the pixel point corresponding to the extreme value of the gray value on each edge as an initial extreme point.

10. The infrared thermography-based analysis of thermal crack detection method as claimed in claim 1, wherein, The expression of the objective function is: , denotes the sum of integral areas between all crack functions and the midline function in the t-th gray-scale image, denotes the sum of distances of all local maximum points before and after adjustment in the t-th gray-scale image, is the number of gray-scale images, denotes the objective function.

Images