A concrete slab panel detection system and method based on image recognition

Abstract

The present invention relates to the field of image processing technology, and in particular to a concrete slab panel detection system and method based on image recognition, the method comprising: obtaining a grayscale image of a concrete panel; obtaining the inter-class variance of each grayscale level in the grayscale image; drawing a curve with the grayscale level as the abscissa and the inter-class variance as the ordinate to obtain a maximum point; calculating the weight corresponding to each maximum point, using each weight to perform weighted summation on the inter-class variance of all maximum points, and using the grayscale level corresponding to the weighted inter-class value as a segmentation threshold; using the segmentation threshold to segment the grayscale image to obtain a suspected crack area; identifying the suspected crack area to obtain a crack area. The scheme of the present invention can accurately identify crack areas in concrete panels.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and more specifically, to a concrete slab panel detection system and method based on image recognition. Background Art

[0002] Panel concrete structures play a key role in the field of water conservancy projects and are the main components of water conservancy facilities such as dams, channels, and energy dissipation buildings. Over time, cracks may appear in panel concrete structures. These cracks may be caused by a variety of reasons such as material aging, environmental impacts, load changes, etc. The presence of cracks may affect the safety, durability, and reliability of the structure, so it is crucial to regularly inspect the panel concrete structure and identify and measure cracks.

[0003] With the continuous development of computer image processing technology, concrete surface defect detection technology based on image recognition has emerged. Image recognition technology can automatically detect and identify concrete surface defects by processing and analyzing the captured concrete surface images, thereby improving detection efficiency and accuracy.

[0004] Among them, the existing image segmentation can realize the rapid and accurate detection of concrete surface defects. For example, the publication number is CN112785579A, and the name is "Concrete Crack Identification Method Based on Image Processing Technology", which discloses the use of a threshold segmentation method to segment the grayscale image of the collected concrete structure surface to distinguish between the crack area and the background area, and then further distinguish between the crack and the residual noise to accurately identify the crack and calculate the crack parameters.

[0005] The maximum inter-class variance method (Otsu method) used above determines the optimal threshold for image segmentation, where the maximum inter-class variance method (Otsu method) is a common image segmentation algorithm, which obtains the maximum inter-class variance, uses the corresponding gray level as the segmentation threshold, and binarizes the gray image to achieve image segmentation. However, in practical applications, due to the complexity of image content, such as concrete panel images often contain complex backgrounds, such as uneven lighting, surface contamination, texture changes, etc., these factors may lead to uneven gray distribution in the crack area, making it difficult to accurately segment the cracks by directly selecting the threshold corresponding to the maximum inter-class variance.

[0006] Therefore, it is particularly important to perform accurate image segmentation and improve the accuracy of crack identification. Summary of the invention

[0007] The purpose of the present invention is to propose a concrete slab panel detection system and method based on image recognition, so as to solve the problem that it is difficult to accurately perform image segmentation in the prior art; to this end, the present invention provides solutions in the following two aspects.

[0008] In a first aspect, the present invention provides a concrete slab panel detection method based on image recognition, comprising:

[0009] Obtain a grayscale image of a concrete panel; obtain the inter-class variance of each grayscale level in the grayscale image; draw a curve with the grayscale level as the abscissa and the inter-class variance as the ordinate to obtain a maximum point;

[0010] Calculate the weight corresponding to each maximum point, use each weight to perform weighted summation on the inter-class variance of all maximum points, and use the gray level corresponding to the weighted inter-class value as the segmentation threshold; use the segmentation threshold to segment the gray image to obtain suspected crack areas; identify the suspected crack areas to obtain crack areas;

[0011] The weight is the normalized value of the product of the prominence and the expression of any maximum point;

[0012] The prominence degree is the ratio of the difference between any maximum point and its adjacent points on both sides to the sum of the differences corresponding to all maximum points; the performance degree is negatively correlated with the directional characteristics and linear characteristics of any maximum point, and the directional characteristics characterize the gradient change of the grayscale corresponding to any maximum point; the linear characteristics are positively correlated with the proportion of feature points in the grayscale corresponding to any maximum point, and negatively correlated with the mean of the feature degree; the feature degree reflects the change in the distance and gradient between each feature point and the corresponding feature pixel point, and the feature pixel point is located in the vertical direction of the gradient direction of the corresponding feature point.

[0013] In the above scheme, the inter-class variance of each gray level in the grayscale image is obtained, and the gray level corresponding to the inter-class variance maximum is screened out. Through multi-angle analysis (inter-class variance difference, gradient direction, and the characteristics of the feature points in the corresponding grayscale), the influence of the inter-class variance maximum corresponding to different grayscale levels is comprehensively considered, and the inter-class value is obtained by weighting according to the importance of the inter-class variance maximum. It can take into account the cracks in different situations in the concrete panel, so that it can accurately segment the grayscale image to obtain the final crack area.

[0014] In one embodiment, the step of judging the segmentation threshold to determine the optimal segmentation threshold is further included, specifically: when the inter-class value is not on the curve, obtaining the current gray level corresponding to the inter-class variance whose difference with the inter-class value is less than the set threshold; if the number of current gray levels is one, the current gray level is used as the optimal segmentation threshold; if the number of current gray levels is at least two, selecting the minimum gray level among the current gray levels as the optimal segmentation threshold.

[0015] In the above scheme, the optimal segmentation threshold is determined by judging whether the weighted inter-class value is on the curve, so as to achieve accurate segmentation of the grayscale image.

[0016] In one embodiment, the weight for: ; In the formula, is the prominence of the ith maximum point, is the gray level expression of the i-th maximum point, I is the number of maximum points in the curve, is the normalization function.

[0017] In one embodiment, the performance level for: ; In the formula, is the directional feature of the gray level corresponding to the i-th maximum point, is the linear feature of the gray level corresponding to the i-th maximum point.

[0018] In one embodiment, the linear feature for: ; is the number of feature points in the gray level corresponding to the i-th maximum point, is the total number of pixels in the gray level corresponding to the i-th maximum point, is the mean value of the characteristic degree of the gray level corresponding to the i-th maximum point; , is the distance between the feature point t and the pixel point p in the gray level corresponding to the i-th maximum point, is the gradient direction of feature point t, is the gradient direction of pixel p, is an exponential function with the natural constant e as the base, and m is the total number of pixels in the vertical direction of the gradient direction of the feature point t.

[0019] In the above scheme, the pixel points in the grayscale corresponding to any maximum point are screened for feature points, and then the linear characteristics of the grayscale corresponding to any maximum point are determined by analyzing the feature points, which can reflect the situation of the cracks and provide data support for the subsequent identification of the crack area.

[0020] In one embodiment, the directional feature is the average difference of all gradient directions in the grayscale corresponding to any maximum point.

[0021] In the above scheme, by analyzing the gradient direction of all pixel points in the grayscale corresponding to any maximum point, the characteristics characterizing the crack can be obtained, providing data support for the subsequent identification of the crack area.

[0022] In one embodiment, the specific process of segmenting the grayscale image using the segmentation threshold to obtain the crack area of ​​the concrete panel is as follows: marking the pixel points greater than the segmentation threshold as the background area, and adjusting the corresponding grayscale value to 0, marking the pixel points less than the segmentation threshold as the foreground area, and adjusting the corresponding grayscale value to 255, to obtain a binary image, and then performing connected domain analysis on the binary image to obtain the crack area.

[0023] In one embodiment, the grayscale level is the grayscale value of a pixel in a grayscale image, and the inter-class variance of each grayscale level is obtained by an Otsu threshold method.

[0024] In a second aspect, the present invention provides a concrete slab panel detection system based on image recognition, comprising:

[0025] processor;

[0026] The memory stores computer instructions for concrete slab panel detection based on image recognition. When the computer instructions are executed by the processor, the system executes the above-mentioned concrete slab panel detection method based on image recognition.

[0027] The beneficial effects of the present invention are:

[0028] The scheme of the present invention first analyzes the inter-class variance in each gray level that has been screened out to determine the prominence of different gray levels; then, the characteristics of all pixel points in each gray level are analyzed to determine the index of crack possibility, that is, the degree of expression; then, the degree of prominence and the degree of expression are combined to determine the importance of the gray level corresponding to the maximum value of the inter-class variance, and based on the different importance levels and the corresponding maximum values ​​of the inter-class variance, the combined inter-class value is obtained to obtain the segmentation threshold; compared with the maximum inter-class variance value in the prior art, it takes into account the possibility of cracks in different situations, determines the segmentation threshold from a global perspective, can accurately perform image segmentation, and ensures the accuracy of subsequent crack area identification in different situations. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] 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:

[0030] Figure 1 A flowchart of the steps of a concrete slab panel detection method based on image recognition in this embodiment is schematically shown;

[0031] Figure 2The structural block diagram of a concrete slab panel detection system based on image recognition in this embodiment is schematically shown. DETAILED DESCRIPTION

[0032] 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.

[0033] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0034] Among them, image segmentation is often used in computer pattern recognition, in which operations such as target detection, feature screening, and target recognition are all achieved through image segmentation technology. Common image segmentation methods include threshold method, edge detection method, and region tracking method. Among the threshold methods, the Otsu threshold method is one of the typical threshold segmentation methods. The main idea of ​​the Otsu threshold method is to obtain the inter-class variance of each gray level in the image, and use the gray level corresponding to the maximum inter-class variance as the segmentation threshold to perform binary segmentation on the image. However, there are cracks in different situations on the concrete panel. At this time, directly using the maximum inter-class variance for image segmentation cannot guarantee that it can be applied to the concrete panel crack segmentation in each situation, and then obtain accurate cracks.

[0035] Therefore, based on the above problems, the present invention provides a concrete slab panel detection method and system based on image recognition, which starts from the image itself to find an optimal threshold to improve the traditional threshold obtained by the maximum inter-class variance, which can improve the robustness of the detection model to a certain extent, and thus can accurately segment the image.

[0036] like Figure 1 As shown, a concrete slab panel detection method based on image recognition in this embodiment includes the following steps:

[0037] In step S1, a grayscale image of a concrete panel is obtained. Specifically, an image of a concrete panel in an outdoor beam bridge or a building is collected by using a drone, or an image of a concrete panel is captured from a surveillance video, and the collected image is grayscaled to obtain a grayscale image.

[0038] In step S2, the inter-class variance of each gray level in the gray image is obtained; a curve is drawn with the gray level as the abscissa and the inter-class variance as the ordinate to obtain a maximum point.

[0039] Specifically, the grayscale values ​​of the pixels in the grayscale image are divided into grayscale levels to obtain all grayscale levels, all grayscale levels in the image are traversed, the inter-class variance corresponding to each grayscale level is obtained, and the inter-class variance curve of the collected image is drawn to obtain the maximum inter-class variance in the curve.

[0040] The above inter-class variance is obtained by using the Otsu threshold method, which is a prior art and will not be described in detail here.

[0041] The grayscale division of this embodiment is to correspond the grayscale to the grayscale value one by one, and the specific grayscale is determined according to the grayscale value of the pixel in the grayscale image. For example, if the grayscale value is 183, the grayscale is also 183; therefore, if the grayscale value is 0-255, the grayscale is also 0-255, and the total number of grayscales is 256.

[0042] The method for obtaining the inter-class variance is to use the Otsu threshold method. Since it is an existing technology, it will not be described in detail here.

[0043] In step S3, the weight corresponding to each maximum point is calculated, and the inter-class variance of all the maximum points is weightedly summed using the weight to obtain the inter-class value, and the gray level corresponding to the inter-class value is used as the segmentation threshold.

[0044] In this embodiment, the weight corresponding to each maximum point is obtained according to the prominence degree and the expression degree.

[0045] The prominence degree is the ratio of the difference between any maximum point and its two adjacent points to the sum of the differences corresponding to all maximum points. It is expressed as:

[0046] ; In the formula, is the inter-class variance of the i-th maximum point, is the inter-class variance of the adjacent points on the left side of the i-th maximum point, is the inter-class variance of the adjacent points on the right side of the i-th maximum point, is the number of maximum points in the curve.

[0047] In one embodiment, the process of obtaining the performance level is:

[0048] First, the directional feature of any maximum point is calculated. Specifically, the directional feature in this embodiment is the average difference of all gradient directions in the grayscale corresponding to any maximum point. Specifically, the gradients of all pixels in the grayscale corresponding to any maximum point are traversed and their gradient directions are obtained, the mean of the gradient directions of all pixels is obtained, and then the difference between the gradient of the pixel corresponding to the grayscale of the inter-class variance maximum value and the mean of the gradient direction is obtained, and the mean is obtained, and the mean is used as the directional feature of the pixel corresponding to the grayscale of the inter-class variance maximum value. The smaller the mean, the more obvious the directional feature, and the higher the degree of expression of the pixel corresponding to the grayscale of the inter-class variance maximum value for the crack edge.

[0049] Therefore, the ith maximum point corresponds to the directional feature of the gray level It is expressed as:

[0050] ; In the formula, is the gray level corresponding to the i-th maximum point The gradient direction of each pixel is is the mean value of the gradient direction of all pixels in the gray level corresponding to the i-th maximum point, is the total number of pixels in the grayscale corresponding to the i-th maximum point.

[0051] The above gradient direction is obtained by using the Sobel operator. Since the Sobel operator is a prior art, it will not be described in detail here.

[0052] Secondly, calculate the performance degree of any maximum point. Specifically, first determine whether any pixel point in the grayscale corresponding to any maximum point exists in the vertical direction of its gradient direction, and if so, mark any pixel point as a feature point, and obtain the distance between the feature point and other pixels in the vertical direction of its gradient direction, as well as the difference in the gradient direction between the two, and obtain the feature degree, and then obtain the mean of the feature degree of all feature points; based on the mean of the feature degree and the proportion of the feature point in the grayscale, obtain the corresponding linear feature.

[0053] Specifically, the mean value of the characteristic degree of the gray level corresponding to the i-th maximum point is:

[0054] ; is the distance between the feature point t and the pixel point p in the gray level corresponding to the i-th maximum point, is the gradient direction of feature point t, is the gradient direction of pixel p, is an exponential function with the natural constant e as the base, and m is the total number of pixels in the vertical direction of the gradient direction of the feature point t.

[0055] in, Indicates the characteristic degree of feature point t. The closer the distance and the smaller the difference in gradient direction, the greater the characteristic degree of feature point t. The larger the mean value of the above characteristic degree, the more obvious the linear feature, and the higher the degree of expression of the pixel point in the gray level corresponding to the i-th maximum point on the crack edge.

[0056] Among them, the linear feature for: ; is the number of feature points in the gray level corresponding to the i-th maximum point, is the total number of pixels in the gray level corresponding to the i-th maximum point, is the mean value of the characteristic degree of the gray level corresponding to the i-th maximum point.

[0057] in, It represents the credibility factor of the linear feature of the grayscale corresponding to the i-th maximum point, that is, the greater the proportion of the number of feature points in the total number of pixel points in the corresponding grayscale in the i-th maximum point, the higher the credibility of the linear feature.

[0058] Then, according to the directional characteristics and linear characteristics of any maximum point, the degree of expression of the gray level corresponding to any maximum point on the crack is obtained. Among them, the degree of expression of the gray level corresponding to the maximum value of the inter-class variance on the crack It is expressed as: ; In the formula, is the directional feature of the gray level corresponding to the i-th maximum point, is the linear feature of the gray level corresponding to the i-th maximum point. Used to constrain the result between 0 and 1.

[0059] Among them, the more obvious the directional characteristics and linear characteristics are, the greater the degree to which the gray level of the maximum variance between classes expresses the cracks.

[0060] The weight in this embodiment is the normalized value of the product of the prominence and the performance of any maximum point. In one embodiment, the weight is expressed as:

[0061] , is the prominence of the ith maximum point, is the gray level expression of the i-th maximum point, I is the number of maximum points in the curve, is a normalization function; the normalization function specifically adopts the maximum and minimum normalization method.

[0062] In summary, the greater the prominence of the i-th maximum point, and the greater the grayscale corresponding to the prominence of the i-th maximum point in terms of the performance of the crack, the better the performance of the crack when the grayscale corresponding to the i-th maximum point is used as the threshold for segmentation, the greater the corresponding weight, and the greater the possibility that the corresponding grayscale will be the optimal threshold.

[0063] It should be noted that, in this embodiment, when obtaining the weight corresponding to the maximum value of the inter-class variance, the distribution characteristics of the grayscale corresponding to the maximum value of the inter-class variance and the pixel points corresponding to the grayscale are analyzed. Specifically, on the one hand, the inter-class variance reflects the difference between the two classes. The larger the inter-class variance, the more obvious the difference between the two classes, and the better the effect of using the corresponding grayscale as the threshold for segmentation. In other words, if the difference between the maximum value of the inter-class variance and the variance between adjacent classes is greater, the more prominent the maximum value of the inter-class variance is compared with other inter-class variance maxima, the better the corresponding segmentation is, and the greater the weight is. On the other hand, the image segmentation of the concrete panel is to detect whether there are cracks, and the cracks have certain directional and linear characteristics. In other words, if there are cracks in the concrete panel image, the optimal segmentation threshold selected here needs to better show the cracks. The maximum value of the inter-class variance often corresponds to the edge of the crack, because at this time the difference between the inside and outside of the crack is the largest. In other words, the distribution of pixels corresponding to the grayscale of the maximum inter-class variance is mostly at the edge of the crack, that is, the distribution of pixels corresponding to the grayscale of the maximum inter-class variance, the better the characteristic performance of the crack, the better the segmentation effect of the threshold corresponding to the maximum inter-class variance, and the greater the weight of the maximum inter-class variance.

[0064] Furthermore, since the obtained inter-class value may not be on the curve and has no corresponding grayscale level, after obtaining the segmentation threshold, it is also necessary to judge the segmentation threshold to determine the optimal segmentation threshold step, specifically: when the inter-class value is not on the curve, obtain the grayscale corresponding to the inter-class variance whose difference with the inter-class value is less than the set threshold; if the number of grayscale levels is one, use the current grayscale level as the optimal segmentation threshold; if the number of grayscale levels is at least two, select the smallest grayscale level as the optimal segmentation threshold, and use the optimal segmentation threshold to segment the grayscale image.

[0065] At step S4, the grayscale image is segmented using a segmentation threshold to obtain the crack area of ​​the concrete panel. Specifically, the image is binarized and segmented according to the segmentation threshold. Since the color of the concrete panel is grayish white and the crack is black, the pixels greater than the segmentation threshold are marked as the background area and their grayscale is adjusted to 0, and the pixels less than the segmentation threshold are marked as the foreground area and their grayscale is adjusted to 255. In this way, the binary segmentation of the collected image is completed to obtain a binary image. After that, the binary image is identified to obtain the crack area. Specifically, the method for identifying the binary image is to perform a connected domain analysis on the binarized image, mark it, and calculate the area of ​​the suspected crack area after the mark and the major-minor axis ratio of the minimum circumscribed ellipse to determine whether it is a crack area.

[0066] As another implementation method, the method for identifying a binary image can also be performed by inputting the binary image into a pre-trained network model to identify whether the foreground area is a crack. If the foreground area is a crack, an early warning is issued to remind relevant staff to handle it in time to avoid the collapse or breakage of the concrete panel and cause a safety accident.

[0067] The above network model can adopt a CNN network or a RNN network.

[0068] The training process of the above network model is:

[0069] Obtain a training data set; specifically, obtain historical concrete images, and according to the above segmentation threshold method, obtain the segmentation threshold of each historical concrete image, and perform threshold segmentation on the historical image to obtain a binarized historical image. At the same time, annotate the binarized historical image to obtain label data, in which cracks are annotated as 1 and others are annotated as 0.

[0070] The training data set is input into the network model for training to obtain a trained network model. The loss function is the mean square error loss. Since the training process of the network model is an existing technology, it will not be described in detail here.

[0071] The scheme of the present invention can determine the weight of the corresponding gray level by introducing the prominence and expression of the gray level corresponding to the maximum value of the inter-class variance, thereby determining the importance of the gray level corresponding to the maximum value of the inter-class variance, and performing weighted summation based on different importance levels and the corresponding maximum values ​​of the inter-class variance to obtain the comprehensive inter-class value, and then obtain the segmentation threshold; that is, the acquisition of the segmentation threshold of the present invention takes into account the possibility of cracks in different situations, determines the segmentation threshold from a global perspective, can accurately perform image segmentation, and ensures the accuracy of subsequent crack area identification under different situations.

[0072] The present invention also provides a concrete slab panel detection system based on image recognition. Figure 2 As shown, the system includes a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned concrete slab panel detection method based on image recognition according to the present invention is implemented.

[0073] The system also includes other components familiar to those skilled in the art, such as a communication bus and a communication interface, whose configuration and functions are known in the art and thus will not be described in detail here.

[0074] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, apparatus or device. For example, a computer-readable storage medium may be any appropriate magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory RRAM (Resistive Random Access Memory), a dynamic random access memory DRAM (Dynamic Random Access Memory), a static random access memory SRAM (Static Random-Access Memory), an enhanced dynamic random access memory EDRAM (Enhanced Dynamic Random Access Memory), a high-bandwidth memory HBM (High-Bandwidth Memory), a hybrid memory cube HMC (Hybrid Memory Cube), etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of a device or accessible or connectable to a device. Any application or module described in the present invention may be implemented using computer-readable / executable instructions that may be stored or otherwise maintained by such a computer-readable medium.

[0075] In the description of this specification, “plurality” means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0076] 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. A concrete slab panel detection method based on image recognition, characterized in that: include: Obtain a grayscale image of the concrete panel; Obtaining the inter-class variance of each gray level in the grayscale image; Draw a curve with grayscale as the horizontal axis and inter-class variance as the vertical axis to obtain the maximum point; Calculate the weights corresponding to each maximum point, use each weight to perform weighted summation on the inter-class variances of all maximum points, and use the grayscale level corresponding to the weighted inter-class value as the segmentation threshold; use the segmentation threshold to segment the grayscale image and obtain the crack area of ​​the concrete panel; The weight is the normalized value of the product of the prominence and the expression of any maximum point; The prominence degree is the ratio of the difference between any maximum point and its two adjacent points to the sum of the differences corresponding to all maximum points; the performance degree is inversely correlated with the directional characteristics and linear characteristics of any maximum point, and the directional characteristics represent the gradient change of the grayscale corresponding to any maximum point; the linear characteristics are positively correlated with the proportion of feature points in the grayscale corresponding to any maximum point, and inversely correlated with the mean of the feature degree; the feature degree reflects the change in the distance and gradient between each feature point and the corresponding feature pixel point, and the feature pixel point is located in the vertical direction of the gradient direction of the corresponding feature point; Performance level for: ; In the formula, is the directional feature of the gray level corresponding to the i-th maximum point, is the linear feature of the gray level corresponding to the i-th maximum point; Linear Features for: ; In the formula, is the number of feature points in the gray level corresponding to the i-th maximum point, is the total number of pixels in the gray level corresponding to the i-th maximum point, is the mean value of the characteristic degree of the gray level corresponding to the i-th maximum point; , is the distance between the feature point t and the feature pixel p in the gray level corresponding to the i-th maximum point, is the gradient direction of feature point t, is the gradient direction of the feature pixel point p, is an exponential function with the natural constant e as the base, and m is the total number of feature pixels in the vertical direction of the gradient direction of the feature point t.

2. The method for detecting a concrete slab panel based on image recognition according to claim 1, characterized in that: The method also includes judging the segmentation threshold to determine the optimal segmentation threshold, specifically: when the inter-class value is not on the curve, obtaining the current gray level corresponding to the inter-class variance whose difference with the inter-class value is less than the set threshold; if the number of current gray levels is one, the current gray level is used as the optimal segmentation threshold; if the number of current gray levels is at least two, the minimum gray level among the current gray levels is selected as the optimal segmentation threshold.

3. The concrete slab panel detection method based on image recognition according to claim 1, characterized in that: The weight for: ; In the formula, is the prominence of the ith maximum point, is the gray level expression of the i-th maximum point, I is the number of maximum points in the curve, is the normalization function.

4. The method for detecting a concrete slab panel based on image recognition according to claim 3, characterized in that: The directional feature is the average difference of all gradient directions in the grayscale corresponding to any maximum point.

5. The method for detecting concrete slab panels based on image recognition according to claim 1, characterized in that: The specific process of segmenting the grayscale image using the segmentation threshold to obtain the crack area of ​​the concrete panel is as follows: marking the pixel points greater than the segmentation threshold as the background area, and adjusting the corresponding grayscale value to 0, marking the pixel points less than the segmentation threshold as the foreground area, and adjusting the corresponding grayscale value to 255, to obtain a binary image, and then performing connected domain analysis on the binary image to obtain the crack area.

6. The method for detecting concrete slab panels based on image recognition according to claim 1, characterized in that: The grayscale level is the grayscale value of a pixel in a grayscale image.

7. A concrete slab panel detection system based on image recognition, characterized in that: include: processor; A memory storing computer instructions for concrete slab panel detection based on image recognition, wherein when the computer instructions are executed by the processor, the system executes a concrete slab panel detection method based on image recognition according to any one of claims 1 to 6.

Images