Hydraulic engineering concrete quality detection method and system based on image recognition technology

Abstract

The invention relates to a hydraulic engineering concrete quality detection method and system based on an image recognition technology, which realize real-time and continuous quality monitoring of hydraulic engineering concrete by fusing image sequence analysis and an ultrasonic detection technology, and effectively solve the inherent defects of the traditional regular detection means. By comparing the difference between the current image and the historical image, ultrasonic detection is automatically triggered, visual rough features and ultrasonic rough features are accurately extracted, visual roughness and actual roughness are calculated in combination with a preset fitting model, and early warning of the water seepage risk is achieved; meanwhile, the dynamic comparison mechanism of the crack characteristic area can track the crack expansion state in real time, and missing detection of sudden hidden dangers is avoided. The hidden danger response time is shortened from several weeks to the real-time level, scientific basis is provided for preventing engineering accidents and optimizing maintenance decisions, the safety and durability of a water conservancy project structure are guaranteed, the regular detection cost is reduced, and the method has remarkable engineering practical value and popularization potential.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of visual processing, in particular to a water conservancy project concrete quality detection method and system based on image recognition technology. BACKGROUND

[0002] Concrete detection is a key link to ensure the safety and durability of building engineering structures. Its importance lies in that concrete is a core component of modern buildings, and its quality is directly related to the safety, service life and overall stability of the building. Through scientific and systematic detection, potential problems such as cracks, insufficient strength, and steel corrosion in concrete can be found in time to ensure that the performance of engineering materials meets the design standards, providing reliable protection for the long-term safe operation of buildings, and providing scientific basis for construction quality control, structure evaluation and accident handling. It is an indispensable professional and technical means to prevent engineering quality problems, avoid safety hazards, and prolong the service life of buildings.

[0003] The existing concrete detection technologies in the field of water conservancy engineering, such as core drilling method, ultrasonic detection technology and rebound method, are all periodic detection means. However, during the long-term use of concrete, sudden problems such as dangerous cracks and water seepage may occur. These technologies are difficult to find hidden dangers in time due to long detection period, limited coverage and inability to achieve continuous monitoring, so it is necessary to introduce real-time detection technology based on vision to make up for the shortcomings of periodic detection and realize all-weather and dead-angle-free dynamic monitoring of the state of concrete structures. SUMMARY

[0004] Therefore, the purpose of the present application is to provide a water conservancy project concrete quality detection method and system based on image recognition technology to solve the above technical problems.

[0005] In order to achieve the above purpose, the present application adopts the following technical solutions:

[0006] The water conservancy project concrete quality detection method based on image recognition technology comprises the following steps:

[0007] Obtaining an image sequence of a to-be-detected area of a water conservancy project concrete, wherein the image sequence is obtained based on a pre-set image acquisition device;

[0008] Extracting an original image at a current time point and an original image at a previous time point from the image sequence, and calculating the difference between the original image at the current time point and the original image at the previous time point;

[0009] When the difference is greater than a preset difference threshold, controlling an ultrasonic detection device to emit an ultrasonic signal to the to-be-detected area and obtaining a return signal;

[0010] extracting a visual roughness feature and a crack feature region from the original image at the current time point, and extracting an ultrasonic roughness feature from the ultrasonic echo signal;

[0011] calculating a visual roughness based on a first fitting model pre-constructed and the visual roughness feature, and calculating an ultrasonic smoothness based on a second fitting model pre-constructed and the ultrasonic roughness feature, wherein the first fitting model represents a relationship between a visual roughness feature and a roughness, and the second fitting model represents a relationship between an ultrasonic roughness feature and a roughness;

[0012] comparing the visual roughness with the ultrasonic smoothness to obtain a roughness comparison result, and determining a water seepage condition based on the roughness comparison result; and comparing the crack feature region with a crack feature reference region of a previous detection period to obtain a crack comparison result;

[0013] real-time monitoring a water conservancy project concrete quality based on the water seepage condition and the crack comparison result.

[0014] In an embodiment of the present application, the difference between the original image at the current time point and the original image at a previous time point is calculated, including:

[0015] the original image at the current time point and the original image at the previous time point are respectively pre-processed to obtain a first pre-processed image and a second pre-processed image , wherein the pre-processing includes contrast enhancement, grayscale conversion and filtering processing;

[0016] the difference between the first pre-processed image and the second pre-processed image is calculated , wherein the mathematical expression of the difference is:

[0017]

[0018] In the formula, represents the image width, represents the image height, represents the pixel value of a pixel point with coordinates in the first pre-processed image, represents the pixel value of a pixel point with coordinates in the second pre-processed image.

[0019] In an embodiment of the present application, the visual roughness feature is extracted from the original image at the current time point, including:

[0020] The original image at the current time point is preprocessed to obtain a first preprocessed image. ;

[0021] Calculate the first preprocessed image Gray co-occurrence matrix ;

[0022] Based on the gray-level co-occurrence matrix Extract the first preprocessed image Contrast ,energy ,entropy and homogeneity characteristics This yields visually coarse features.

[0023] In one embodiment of this application, extracting the crack feature region from the original image at the current time point includes:

[0024] The original image at the current time point is preprocessed to obtain a first preprocessed image. ;

[0025] Calculate the first preprocessed image Each pixel Hessian matrix ;

[0026] According to the Hessian matrix Calculate pixels First eigenvalue Second eigenvalue ;

[0027] Based on the first feature value of the pixel Second eigenvalue Calculate the degree of anisotropy Wherein, the anisotropy degree The mathematical expression is:

[0028]

[0029] Will satisfy: The pixels are used as crack pixels, where, Indicates the heterogeneity threshold;

[0030] Density clustering is performed on the crack pixels to obtain multiple pixel clusters, and pixel clusters with a number of pixels greater than or equal to a preset threshold are taken as crack clusters; and crack regions are constructed based on the crack clusters.

[0031] In an embodiment of the present application, the ultrasonic roughness features include a peak amplitude of echo, a spectral center frequency and an attenuation coefficient, wherein the extracting the ultrasonic roughness features from the ultrasonic echo signal comprises:

[0032] The ultrasonic echo signal is preprocessed to obtain a preprocessed echo signal, wherein the preprocessing includes filtering, amplification and gain compensation;

[0033] A time window of surface reflection wave is extracted from the preprocessed echo signal, and an absolute maximum value of the signal is extracted from the time window to obtain a peak amplitude of echo ;

[0034] A spectral center frequency is extracted from the preprocessed echo signal ;

[0035] An echo signal envelope of the preprocessed echo signal is extracted , and an attenuation model in a selected time window is constructed, wherein the mathematical expression of the attenuation model is:

[0036]

[0037] In the formula, represents an initial amplitude, represents an attenuation coefficient, represents a propagation time;

[0038] A plurality of sampling time points and envelope amplitude values of the plurality of sampling time points are obtained by sampling from the selected time window, and the attenuation model is linearly fitted based on the plurality of sampling time points and the envelope amplitude values of the plurality of sampling time points to obtain an attenuation coefficient .

[0039] In an embodiment of the present application, the construction method of the first fitting model or the second fitting model comprises:

[0040] A plurality of concrete surface image samples and ultrasonic echo signal samples of the concrete surface are obtained;

[0041] Visual roughness feature samples are extracted from the concrete surface image samples, and ultrasonic roughness feature samples are extracted from the ultrasonic echo signal samples, wherein the visual roughness feature samples include contrast samples , energy samples , entropy samples and homogeneity feature samples , and the ultrasonic roughness feature samples include peak amplitude of echo samples , spectral center frequency samples and attenuation coefficient samples , wherein indexing a concrete surface image sample, indexing an ultrasound echo signal sample;

[0042] respectively normalizing the ultrasonic roughness feature sample and the visual roughness feature sample to obtain a normalized ultrasonic roughness feature sample and a normalized visual roughness feature sample;

[0043] constructing a first multivariate linear relationship and a second multivariate linear relationship, wherein a mathematical expression of the first multivariate linear relationship is:

[0044]

[0045] a mathematical expression of the second multivariate linear relationship is:

[0046]

[0047] wherein, denotes a labeled visual roughness, denotes a labeled ultrasonic smoothness, and both denote an intercept, denotes a contrast term coefficient, denotes a normalized contrast, denotes an energy term coefficient, denotes a normalized energy, denotes an entropy term coefficient, denotes a normalized entropy, denotes a homogeneity feature term coefficient, denotes a normalized homogeneity feature, denotes a peak amplitude term coefficient, denotes a normalized peak amplitude, denotes a spectral center frequency term coefficient, denotes a normalized spectral center frequency, denotes an attenuation coefficient term coefficient, denotes a normalized attenuation coefficient, and both are error terms;

[0048] substituting the normalized visual roughness feature sample into the first multivariate linear relationship and fitting by least squares to obtain a first fitting model, and substituting the normalized ultrasonic roughness feature sample into the second multivariate linear relationship and fitting by least squares to obtain a second fitting model.

[0049] In an embodiment of the present application, the visual roughness is calculated based on a first pre-constructed fitting model and the visual roughness features, and the ultrasonic smoothness is calculated based on a second pre-constructed fitting model and the ultrasonic roughness features, including:

[0050] The visual roughness features are normalized to obtain normalized visual roughness features, and the ultrasonic roughness features are normalized to obtain normalized ultrasonic roughness features;

[0051] The normalized visual roughness features are substituted into the first fitting model to obtain the visual roughness , and the normalized ultrasonic roughness features are substituted into the second fitting model to obtain the ultrasonic smoothness .

[0052] In an embodiment of the present application, the visual roughness and the ultrasonic smoothness are compared to obtain a roughness comparison result, and the water seepage condition is determined based on the roughness comparison result, including:

[0053] The difference rate of the visual roughness and the ultrasonic smoothness is calculated , wherein the mathematical expression of the difference rate is:

[0054]

[0055] When the difference rate is greater than a preset difference threshold, it is determined that there is water seepage in the concrete surface.

[0056] In an embodiment of the present application, the water conservancy concrete quality is detected in real time based on the water seepage condition and the crack comparison result, including:

[0057] When the concrete surface is seeping water, and the crack area of the current detection period is greater than the crack area of the last detection period, an alarm information is sent to a target object.

[0058] The present application also provides a water conservancy concrete quality detection system based on image recognition technology, including:

[0059] An acquisition module is configured to acquire an image sequence of a to-be-detected region of a water conservancy concrete, wherein the image sequence is acquired based on a pre-set image acquisition device;

[0060] A difference calculation module is configured to extract an original image at a current time point and an original image at a last time point from the image sequence, and calculate the difference between the original image at the current time point and the original image at the last time point;

[0061] an ultrasonic detection module, configured to control an ultrasonic detection device to emit an ultrasonic signal to the to-be-detected area and obtain an echo signal when the difference is greater than a preset difference threshold;

[0062] a feature extraction module, configured to extract visual roughness features and crack feature regions from the original image at the current time point, and extract ultrasonic roughness features from the ultrasonic echo signal;

[0063] a roughness calculation module, configured to calculate visual roughness based on a first fitting model pre-constructed and the visual roughness features, and calculate ultrasonic smoothness based on a second fitting model pre-constructed and the ultrasonic roughness features, wherein the first fitting model represents a relationship between visual roughness features and roughness, and the second fitting model represents a relationship between ultrasonic roughness features and roughness;

[0064] a comparison module, configured to compare the visual roughness with the ultrasonic smoothness to obtain a roughness comparison result, and determine a water seepage situation based on the roughness comparison result, and compare the crack feature regions with crack feature reference regions of a previous detection period to obtain a crack comparison result;

[0065] a detection module, configured to perform real-time detection on the quality of the water conservancy project concrete based on the water seepage situation and the crack comparison result.

[0066] The water conservancy project concrete quality detection method and system based on image recognition technology has the advantages that by fusing image sequence analysis and ultrasonic detection technology, real-time and continuous quality monitoring of the water conservancy project concrete is realized, and the inherent defects of the traditional periodic detection means are effectively solved. By comparing the difference between the current image and the historical image, ultrasonic detection is automatically triggered, visual roughness features and ultrasonic roughness features are accurately extracted, visual roughness and actual roughness are calculated by combining the preset fitting model, early warning of water seepage risk is realized, and the dynamic comparison mechanism of the crack feature regions can track the crack expansion state in real time, so that the missed detection of sudden hidden dangers such as dangerous cracks and water seepage is avoided. The technology greatly shortens the detection period, shortens the hidden danger response time from several weeks to real-time level, provides a scientific basis for preventing engineering accidents and optimizing maintenance decisions, guarantees the safety and durability of the water conservancy project structure, reduces the periodic detection cost, and has significant engineering practical value and popularization potential. BRIEF DESCRIPTION OF DRAWINGS

[0067] The application will be further described below in combination with the drawings and embodiments:

[0068] Figure 1 is a water conservancy project concrete quality detection method based on image recognition technology in an embodiment of the application.

[0069] Figure 2 is a flowchart of a water conservancy concrete quality detection method based on image recognition technology shown in an embodiment of the present application;

[0070] Figure 3 is a data processing flow relationship diagram in an embodiment of the present application;

[0071] Figure 4 is a structural diagram of a water conservancy concrete quality detection system based on image recognition technology shown in an embodiment of the present application;

[0072] Figure 5 shows a structural diagram of a computer system of an electronic device suitable for implementing the embodiments of the present application. DETAILED DESCRIPTION

[0073] The embodiments of the present application will be described in detail hereinafter with specific reference to the drawings. Other advantages and effects of the present application can be easily understood by those skilled in the art from the contents disclosed in the present specification. The present application can also be implemented or applied in other different embodiments, and the details in the present specification can be modified or changed in various ways based on different views and applications without departing from the spirit of the present application. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict.

[0074] It should be noted that the diagrams provided in the following embodiments only schematically illustrate the basic concept of the present application, and only the layers related to the present application are shown in the diagrams, not the number of layers, shapes and size ratios when actually implemented. The actual implementation of each layer may be changed in various ways, and the layer layout pattern may also be more complex.

[0075] In the following description, a large number of details are discussed to provide a more thorough explanation of the embodiments of the present application, however, it is obvious to those skilled in the art that the embodiments of the present application can be implemented without these specific details.

[0076] Figure 1 is a flowchart of a water conservancy concrete quality detection method based on image recognition technology shown in an embodiment of the present application; Figure 1As shown, the concrete quality detection method in the present application is applied in the field of concrete facade or slope surface detection, and the detection system includes an edge controller 110, a camera 120, an ultrasonic detection device 130, and a remote server 140. The edge controller 110, the camera 120, and the ultrasonic detection device 130 are arranged at the detection site. The edge controller 110 controls the camera 120 to collect images of the concrete surface at regular time intervals to form an image sequence. The edge controller 110 extracts dynamic frames with changes by using a simple comparison algorithm. Then, the dynamic frames are sent to the remote server 140 through a network, and the remote server 140 performs analysis on the dynamic frames to obtain features such as whether there is water seepage and whether the cracks are spreading. If the alarm is triggered, the alarm information is sent to the administrator 150.

[0077] The above edge monitoring system can be arranged in a key monitoring area with existing cracks to timely monitor the further spreading of cracks and the water seepage of cracks.

[0078] Figure 2 is a flowchart of the concrete quality detection method based on image recognition technology in the water conservancy project in an embodiment of the present application, as shown in Figure 2 As shown, the concrete quality detection method based on image recognition technology in the water conservancy project in the present embodiment can include steps S210 to S270:

[0079] S210, obtaining an image sequence of a to-be-detected area of concrete in a water conservancy project, wherein the image sequence is obtained based on a pre-set image collection device;

[0080] S220, extracting an original image at a current time point and an original image at a previous time point from the image sequence, and calculating the difference between the original image at the current time point and the original image at the previous time point;

[0081] S230, when the difference is greater than a pre-set difference threshold, controlling an ultrasonic detection device to emit an ultrasonic signal to the to-be-detected area and obtaining an echo signal;

[0082] S240, extracting a visual roughness feature and a crack feature area from the original image at the current time point, and extracting an ultrasonic roughness feature from the ultrasonic echo signal;

[0083] S250, calculating a visual roughness based on a pre-constructed first fitting model and the visual roughness feature, and calculating an ultrasonic smoothness based on a pre-constructed second fitting model and the ultrasonic roughness feature, wherein the first fitting model represents the relationship between the visual roughness feature and the roughness, and the second fitting model represents the relationship between the ultrasonic roughness feature and the roughness;

[0084] S260, compare the visual roughness with the ultrasonic smoothness to obtain a roughness comparison result, and determine the water seepage situation based on the roughness comparison result; and compare the crack feature area with the crack feature reference area of ​​the previous detection cycle to obtain a crack comparison result;

[0085] S270, Real-time monitoring of concrete quality in water conservancy projects based on the comparison results of water seepage and cracks.

[0086] Figure 3 This is a schematic diagram of the data processing flow in one embodiment of this application. The following will combine... Figure 3 Each step of the above scheme is explained, specifically including:

[0087] S210, acquire an image sequence of the concrete area to be inspected in the hydraulic engineering project, wherein the image sequence is acquired based on a pre-set image acquisition device;

[0088] Among them, image sequence Multiple frames of images acquired at regular intervals constitute.

[0089] S220, extract the original image at the current time point and the original image at the previous time point from the image sequence, and calculate the difference between the original image at the current time point and the original image at the previous time point, specifically including:

[0090] S221, the original image at the current time point. The original image from the previous time point Preprocessing is performed separately to obtain the first preprocessed image. Second preprocessed image The preprocessing includes contrast enhancement, grayscale conversion, and filtering.

[0091] Contrast enhancement can improve the visibility of details in an image; grayscale conversion can convert a color image to a grayscale image, reducing computational complexity; filtering can remove image noise and smooth the image.

[0092] S222, Calculate the first preprocessed image Second preprocessed image Differences , wherein the difference The mathematical expression is:

[0093]

[0094] In the formula, Indicates the image width. Indicates the image height. This indicates that the coordinates in the first preprocessed image are... The pixel value of the pixel, This indicates that the coordinates in the second preprocessed image are... The pixel value of the pixel.

[0095] It can accurately quantify the degree of change of concrete surface between two time points and effectively detect minute changes in concrete surface (such as crack propagation, surface spalling, etc.).

[0096] S230, when the difference is greater than a preset difference threshold, control the ultrasonic detection device to emit an ultrasonic signal to the area to be detected and obtain an echo signal;

[0097] When image differences When the threshold is exceeded, it indicates that a significant change may have occurred on the concrete surface. At this point, the ultrasonic detection device is controlled to emit ultrasonic signals towards the area to be detected and receive the echo signals. This achieves intelligent detection triggered on demand, avoiding unnecessary ultrasonic detection and saving equipment resources.

[0098] S240, extract visual roughness features and crack feature regions from the original image at the current time point, and extract ultrasonic roughness features from the ultrasonic echo signal. The specific extraction process is as follows:

[0099] (1) Visual coarseness features

[0100] Visual roughness features refer to the assessment of the roughness of a concrete surface from a visual perspective. In the case of water seepage, visual roughness features are minimally affected by avoiding water stains or water flow. The extraction process is as follows:

[0101] (1-1) Preprocess the original image at the current time point to obtain the first preprocessed image. ;

[0102] The preprocessing process includes grayscale conversion, filtering, and grayscale compression (compressing the grayscale level from 255 to 16 or 32).

[0103] (1-2) Calculate the first preprocessed image Gray-level co-occurrence matrix ;

[0104] The gray-level co-occurrence matrix (GLCM) is a core tool for describing the second-order statistical properties of image texture. It statistically analyzes a pair of gray values ​​under specific spatial relationships (distance d, direction θ, such as 0°, 45°, 90°, 135°). The probability of co-occurrence in an image. For rough surfaces, the gray-level changes between pixels are drastic and irregular, which will affect the gray-level co-occurrence matrix. This is reflected in the text.

[0105] (1-3) Based on the gray level co-occurrence matrix extract the first pre-processed image contrast , energy , entropy and homogeneity features , get the visual roughness features.

[0106] contrast measures the intensity of local gray level variation in the image. The rougher the concrete surface, the stronger the contrast between shadows and highlights produced by light, the greater the gray level difference between pixels and the higher the probability of occurrence, so the higher the contrast value.

[0107] energy reflects the uniformity or regularity of the image texture. When the element distribution is concentrated near the main diagonal (i.e. the pixel pair has similar gray level), the energy value is high. The smoother the concrete surface, the more uniform the texture, the higher the energy value. Conversely, the rougher the surface, the more dispersed the element distribution, the lower the energy value.

[0108] entropy measures the complexity or randomness of the image texture. The larger the probability distribution in , the greater the entropy value. The rougher the concrete surface, the more cracks and spalling, the more complex and disordered the texture information, the higher the entropy value. The entropy value of a smooth surface is low.

[0109] homogeneity feature measures the consistency of local gray levels in the texture. It gives higher weight to elements on the main diagonal (pixel pairs with the same gray level). The smoother the surface, the closer the pixel gray levels, the higher the homogeneity value. The rougher the surface, the lower the homogeneity value.

[0110] Mathematical expressions of contrast

[0111]

[0112]

[0113]

[0114]

[0115] In the formula, denotes the number of gray levels, denotes the first gray level,​​​ represents a second gray level, represents a gray co-occurrence probability.

[0116] (2) Crack feature region

[0117] The crack feature region extraction is used to describe all suspected crack regions of the concrete, so as to determine whether there is crack propagation and enlargement phenomenon. Since the crack is a linear structure, and the Hessian matrix has a good description effect on the linear structure, the Hessian matrix is used to screen the crack pixel points in the present application, so as to form a crack region. The specific process includes:

[0118] (2-1) preprocessing the original image of the current time point to obtain a first preprocessed image ;

[0119] (2-2) calculating the Hessian matrix of each pixel point in the first preprocessed image , wherein the mathematical expression of the Hessian matrix is:

[0120]

[0121] In the formula, is the second-order partial derivative in the x direction, is the second-order partial derivative in the y direction, is the second-order partial derivative in the x direction, and is the second-order partial derivative in the y direction; (2-3) calculating the first eigenvalue

[0122] and the second eigenvalue of the pixel point according to the Hessian matrix , wherein the mathematical expressions of the first eigenvalue and the second eigenvalue are respectively:

[0123]

[0124]

[0125] (2-4) calculating the anisotropy degree according to the first eigenvalue and the second eigenvalue of the pixel point , wherein the mathematical expression of the anisotropy degree is:

[0126] ​​​​

[0127] (2-5) taking the pixel point satisfying as a crack pixel point, wherein, represents a heterogeneity threshold value;

[0128] (2-6) performing density clustering on the crack pixel points to obtain a plurality of pixel point clusters, and taking a pixel point cluster with a number of pixel points greater than or equal to a preset number threshold as a crack cluster; and constructing a crack region based on the crack cluster.

[0129] (3) ultrasonic roughness feature

[0130] When the ultrasonic wave propagates to the surface or internal defects of the concrete, the amplitude, frequency component and energy attenuation of the echo signal will change significantly due to the surface roughness (caused by scattering) and internal structure changes.

[0131] The ultrasonic roughness feature can reflect the actual roughness. When water seeps into the concrete surface, the flowing water will cover the surface cracks and uneven areas, resulting in a significant decrease in the actual roughness. At this time, the scattering of ultrasonic waves on the concrete surface decreases, and the peak amplitude of the echo increases, so the ultrasonic smoothness will increase significantly. The visual roughness is less affected by the reflection characteristics of water and changes little. Based on this principle, accurate identification of water seepage can be achieved, and this identification method can avoid the interference of various noises on the concrete surface. The extraction process of the ultrasonic roughness feature includes:

[0132] (3-1) preprocessing the ultrasonic echo signal to obtain a preprocessed echo signal, wherein the preprocessing includes filtering, amplification and gain compensation;

[0133] (3-2) extracting a time window of surface reflection wave from the preprocessed echo signal, and extracting an absolute maximum value of the signal from the time window to obtain a peak amplitude of the echo ;

[0134] When the ultrasonic wave is vertically incident on the concrete surface, part of the energy is reflected to form an echo. The rougher the surface, the more scattering of the sound wave occurs at the micro protrusions, resulting in a decrease in the energy reflected back to the transducer, thereby reducing the peak amplitude of the echo.

[0135] (3-3) extracting a frequency spectrum center frequency from the preprocessed echo signal; the mathematical expression of the frequency spectrum center frequency is:

[0136]

[0137] In the formula, represents the number of points of the frequency spectrum obtained by fast Fourier transform, represents the frequency component index, represents the first​ a frequency component, represents a discrete frequency spectrum of the echo signal after fast Fourier transform.

[0138] (3-4) extracting an echo signal envelope of the pre-processed echo signal , constructing an attenuation model in a selected time window, wherein the mathematical expression of the attenuation model is:

[0139]

[0140] wherein, represents an initial amplitude, represents an attenuation coefficient, represents a propagation time;

[0141] sampling from the selected time window to obtain a plurality of sampling time points and envelope amplitude values of the plurality of sampling time points, and linearly fitting the attenuation model based on the plurality of sampling time points and the envelope amplitude values of the plurality of sampling time points to obtain an attenuation coefficient . The larger the attenuation coefficient , the faster the attenuation, and the more severe the roughness or internal damage.

[0142] S250, calculating a visual roughness based on a pre-constructed first fitting model and the visual roughness feature, and calculating an ultrasonic smoothness based on a pre-constructed second fitting model and the ultrasonic roughness feature, wherein the first fitting model represents the relationship between the visual roughness feature and the roughness, and the second fitting model represents the relationship between the ultrasonic roughness feature and the roughness;

[0143] In this application, a multiple linear relationship is used to fit the relationship between roughness and visual roughness feature, and the relationship between roughness and ultrasonic roughness feature, which specifically includes:

[0144] (1) obtaining a plurality of concrete surface image samples and ultrasonic echo signal samples of the concrete surface;

[0145] The plurality of concrete surface image samples and the ultrasonic echo signal samples of the concrete surface are manually labeled with roughness, i.e. visual roughness and ultrasonic smoothness . When labeling, the visual roughness and ultrasonic smoothness of the corresponding concrete surface image sample and the ultrasonic echo signal sample of the same concrete wall surface should be equal to ensure that the subsequent obtained fitting model is consistent with the label,

[0146] (2) extracting visual roughness feature samples from the concrete surface image samples and extracting ultrasonic roughness feature samples from the ultrasonic echo signal samples, wherein the visual roughness feature samples include contrast samples , energy samples , entropy samples , and homogeneity feature samples , and the ultrasonic roughness feature samples include echo peak amplitude samples , spectral center frequency samples , and attenuation coefficient samples , wherein, is a concrete surface image sample index, is an ultrasonic echo signal sample index;

[0147] The feature extraction process is described above and will not be repeated here.

[0148] (3) respectively normalizing the ultrasonic roughness feature samples and the visual roughness feature samples to obtain normalized ultrasonic roughness feature samples and normalized visual roughness feature samples;

[0149] constructing a first multivariate linear relationship and a second multivariate linear relationship, wherein the mathematical expression of the first multivariate linear relationship is:

[0150]

[0151] The mathematical expression of the second multivariate linear relationship is:

[0152]

[0153] In the formula, denotes the labeled visual roughness, denotes the labeled ultrasonic smoothness, and both denote the intercept, denotes the contrast term coefficient, denotes the normalized contrast, denotes the energy term coefficient, denotes the normalized energy, denotes the entropy term coefficient, denotes the normalized entropy, denotes the homogeneity feature term coefficient, denotes the normalized homogeneity feature, denotes the echo peak amplitude term coefficient, denotes the normalized echo peak amplitude, denotes the spectral center frequency term coefficient, denotes the normalized spectral center frequency, This represents the coefficient of the attenuation coefficient term. Represents the normalized attenuation coefficient. and All are error terms;

[0154] (4) Substitute the normalized visual roughness feature samples into the first multivariate linear relationship and fit them using the least squares method to obtain the first fitting model; and substitute the normalized ultrasonic roughness feature samples into the second multivariate linear relationship and fit them using the least squares method to obtain the second fitting model.

[0155] The least squares fitting of the multivariate linear model is an existing technique and will not be elaborated here. The aforementioned model transforms complex feature parameters into quantifiable roughness indices, improving the accuracy and robustness of roughness assessment through multi-feature fusion. The fitted model is based on actual sample data, ensuring the objectivity and reliability of the results. Furthermore, the coefficients of each feature obtained using the multivariate linear model fitting can explain the most significant contribution of each parameter to roughness, exhibiting good interpretability and facilitating subsequent algorithm debugging.

[0156] After obtaining the first and second fitting models mentioned above, the calculation process for visual roughness and hypervelocity roughness includes:

[0157] S251, normalize the visual roughness feature to obtain normalized visual roughness feature; and normalize the ultrasonic roughness feature to obtain normalized ultrasonic roughness feature.

[0158] Normalization can be achieved using max-min normalization.

[0159] S252, Substitute the normalized visual roughness features into the first fitting model to obtain the visual roughness. The normalized ultrasonic roughness features are then substituted into the second fitting model to obtain the ultrasonic smoothness. .

[0160] S260, compare the visual roughness with the ultrasonic smoothness to obtain a roughness comparison result, and determine the water seepage situation based on the roughness comparison result; and compare the crack feature area with the crack feature reference area of ​​the previous detection cycle to obtain a crack comparison result;

[0161] Specifically, it includes:

[0162] Calculate the visual roughness With the aforementioned ultrasonic smoothness Difference rate , wherein the difference rate The mathematical expression is:

[0163]

[0164] In the difference rate When the difference rate is greater than a preset difference threshold, it is determined that there is water seepage in the concrete surface.

[0165] If the visual roughness And the ultrasonic smoothness Come from the same wall surface, so theoretically the visual roughness And the ultrasonic smoothness Equal. If the difference is large, after excluding the rainfall factor (judged by the staff), it can be judged that the wall surface is seeped.

[0166] S270, based on the water seepage condition and the crack comparison result, real-time monitoring of the quality of the hydraulic engineering concrete.

[0167] When the concrete surface is seeped, and the crack area of the current detection period is greater than the crack area of the last detection period, the alarm information is sent to the target object.

[0168] If the concrete surface is seeped, and the crack area of the current detection period is greater than the crack area of the last detection period, it indicates that the concrete has quality problems and needs to be repaired, so the alarm information is sent to the staff.

[0169] The water conservancy engineering concrete quality detection method based on image recognition technology of the application realizes real-time and continuous quality monitoring of the water conservancy engineering concrete by fusing image sequence analysis and ultrasonic detection technology, effectively solving the inherent defects of traditional periodic detection means. By comparing the difference between the current and historical images, the ultrasonic detection is automatically triggered, the visual roughness features and ultrasonic roughness features are accurately extracted, the visual roughness and actual roughness are calculated by combining the preset fitting model, and the early warning of water seepage risk is realized. At the same time, the dynamic comparison mechanism of the crack feature area can track the crack expansion state in real time, avoiding the missed detection of sudden hidden dangers (such as dangerous cracks and water seepage). The technology greatly shortens the detection period, shortens the hidden danger response time from several weeks to real-time level, provides scientific basis for preventing engineering accidents and optimizing maintenance decisions, not only guarantees the safety and durability of the water conservancy engineering structure, but also reduces the periodic detection cost, has obvious engineering practical value and popularization potential.

[0170] As Figure 4 shown, the application also provides a water conservancy engineering concrete quality detection system based on image recognition technology, comprising:

[0171] An acquisition module is configured to acquire an image sequence of a to-be-detected region of the hydraulic engineering concrete, wherein the image sequence is acquired based on a pre-set image acquisition device;

[0172] A difference calculation module is configured to extract an original image at a current time point and an original image at a previous time point from the image sequence, and calculate a difference between the original image at the current time point and the original image at the previous time point;

[0173] An ultrasonic detection module is configured to, when the difference is greater than a pre-set difference threshold, control an ultrasonic detection device to emit an ultrasonic signal to the to-be-detected region, and obtain an echo signal;

[0174] A feature extraction module is configured to extract a visual roughness feature and a crack feature region from the original image at the current time point, and extract an ultrasonic roughness feature from the ultrasonic echo signal;

[0175] A roughness calculation module is configured to calculate a visual roughness based on a pre-constructed first fitting model and the visual roughness feature, and calculate an ultrasonic smoothness based on a pre-constructed second fitting model and the ultrasonic roughness feature, wherein the first fitting model represents a relationship between the visual roughness feature and the roughness, and the second fitting model represents a relationship between the ultrasonic roughness feature and the roughness;

[0176] A comparison module is configured to compare the visual roughness with the ultrasonic smoothness to obtain a roughness comparison result, and determine a water seepage condition based on the roughness comparison result; and compare the crack feature region with a crack feature reference region of a previous detection period to obtain a crack comparison result;

[0177] A detection module is configured to perform real-time detection on the quality of the hydraulic engineering concrete based on the water seepage condition and the crack comparison result.

[0178] The image recognition technology-based water conservancy concrete quality detection system of the application realizes real-time and continuous quality monitoring of water conservancy concrete by fusing image sequence analysis and ultrasonic detection technology, effectively solving the inherent defects of traditional periodic detection methods. By comparing the differences between the current and historical images, the ultrasonic detection is automatically triggered, the visual roughness features and ultrasonic roughness features are accurately extracted, the visual roughness and actual roughness are calculated combined with the preset fitting model, and the early warning of water seepage risk is realized. At the same time, the dynamic comparison mechanism of the crack feature area can track the crack expansion state in real time, avoiding the missed detection of sudden hazards (such as dangerous cracks and water seepage). The technology greatly shortens the detection cycle, shortens the hidden danger response time from several weeks to real-time level, provides a scientific basis for preventing engineering accidents and optimizing maintenance decisions, ensures the safety and durability of water conservancy structures, reduces the cost of periodic detection, and has significant engineering practical value and popularization potential.

[0179] Figure 5 The structural diagram of the computer system of the electronic device suitable for realizing the embodiments of the application is shown. It should be noted that, Figure 5 The computer system of the electronic device shown is only an example and should not limit the functions and use range of the embodiments of the application.

[0180] As Figure 5 shown, the computer system includes a central processing unit (CPU) 501, which can perform various appropriate actions and processes according to programs stored in a read-only memory (ROM) 502 or programs loaded from a storage portion 508 to a random access memory (RAM) 503, such as performing the methods in the above embodiments. In the RAM 503, various programs and data required for system operation are also stored. The CPU 501, the ROM 502, and the RAM 503 are connected to each other through a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0181] The following components are connected to the I / O interface 505: an input part 506 including a keyboard, a mouse, etc.; an output part 507 including a display such as a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), etc., and a speaker, etc.; a storage part 508 including a hard disk, etc.; and a communication part 509 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication part 509 performs communication processing via a network such as the Internet. A drive 510 is also connected to the I / O interface 505 as necessary. A removable medium 511 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc. is attached to the drive 510 as necessary, so that a computer program read out therefrom is installed in the storage part 508 as necessary.

[0182] In particular, according to embodiments of the present application, the processes described above with reference to the flowcharts can be implemented as a computer software program. For example, embodiments of the present application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing a computer program for executing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network by the communication part 509, and / or installed from the removable medium 511. When the computer program is executed by the central processing unit (CPU) 501, various functions defined in the system of the present application are executed.

[0183] It should be noted that the computer-readable medium in the embodiments of the present application can be a computer-readable signal medium or a computer-readable storage medium or any combination of the two. The computer-readable storage medium may, for example, be an electrical, magnetic, optical, electromagnetic, infrared or semiconductor system, device or apparatus, or any combination of the above. More specific examples of the computer-readable storage medium can include, but are not limited to, an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disk read-only memory (Compact Disc Read-Only Memory, CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this application, the computer-readable signal medium can include a data signal carried in a baseband or as a part of a carrier wave, which carries computer-readable computer programs. Such a propagated data signal can take on various forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. The computer-readable signal medium can also be any computer-readable medium other than the computer-readable storage medium, which can send, propagate or transmit programs for use by or in connection with an instruction execution system, apparatus or device. The computer programs contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to wireless, wired, or the like, or any suitable combination of the above.

[0184] The flowcharts and block diagrams in the drawings illustrate the possible implementation architectures, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. Each block in the flowcharts or block diagrams can represent a module, a program segment or a part of code containing one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions noted in the blocks can occur in different orders than that shown in the drawings. For example, two blocks that are shown in succession can actually be executed substantially in parallel, and sometimes they can be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams or flowcharts, and the combination of blocks in the block diagrams or flowcharts, can be implemented by a dedicated hardware-based system that performs the specified functions or operations, or can be implemented by a combination of special-purpose hardware and computer instructions.

[0185] The units described in the embodiments of the present application can be implemented by software, or by hardware, or by a combination of software and hardware. The units described can also be located in a single processor. In some cases, the names of the units do not constitute a limitation on the units themselves.

[0186] Another aspect of the present application provides a computer readable storage medium, which stores a computer program. The computer program is executed by a processor of a computer, and causes the computer to perform the method described above. The computer readable storage medium can be included in the electronic device described in the embodiments above, or can exist separately from the electronic device.

[0187] Another aspect of the present application provides a computer program product or a computer program. The computer program product or the computer program includes computer instructions. The computer instructions are stored in a computer readable storage medium. A processor of a computer reads the computer instructions from the computer readable storage medium. The processor executes the computer instructions, and causes the computer to perform the method described in the embodiments above.

[0188] The above embodiments are merely preferred embodiments of the present application and are not intended to limit the protection scope of the present application. Equivalent replacements or transformations of the present application made by those skilled in the art based on the embodiments of the present application are within the protection scope of the present application.

Claims

1. A method for detecting the quality of concrete in hydraulic engineering based on image recognition technology, characterized in that, Including the following steps: An image sequence of the concrete area to be inspected in a hydraulic engineering project is acquired, wherein the image sequence is acquired based on a pre-set image acquisition device; Extract the original image at the current time point and the original image at the previous time point from the image sequence, and calculate the difference between the original image at the current time point and the original image at the previous time point; When the difference is greater than a preset difference threshold, the ultrasonic detection device is controlled to emit an ultrasonic signal toward the area to be detected and an echo signal is obtained. Visual roughness features and crack feature regions are extracted from the original image at the current time point, and ultrasonic roughness features are extracted from the ultrasonic echo signal; Visual roughness is calculated based on a pre-built first fitting model and the visual roughness feature, and ultrasonic smoothness is calculated based on a pre-built second fitting model and the ultrasonic roughness feature, wherein the first fitting model characterizes the relationship between visual roughness feature and roughness, and the second fitting model characterizes the relationship between ultrasonic roughness feature and roughness. The visual roughness is compared with the ultrasonic smoothness to obtain a roughness comparison result, and the water seepage situation is determined based on the roughness comparison result; and the crack feature area is compared with the crack feature reference area of ​​the previous detection cycle to obtain a crack comparison result. The concrete quality of water conservancy projects is monitored in real time based on the comparison results of the water seepage and cracks.

2. The method for detecting the quality of concrete in water conservancy projects based on image recognition technology according to claim 1, characterized in that, Calculating the difference between the original image at the current time point and the original image at the previous time point includes: The original image at the current time point And the original image from the previous time point Preprocessing is performed separately to obtain the first preprocessed image. Second preprocessed image The preprocessing includes contrast enhancement, grayscale conversion, and filtering. Calculate the first preprocessed image Second preprocessed image Differences , wherein the difference The mathematical expression is: In the formula, Indicates the image width. Indicates the image height. This indicates that the coordinates in the first preprocessed image are... The pixel value of the pixel. This indicates that the coordinates in the second preprocessed image are... The pixel value of the pixel.

3. The method for detecting the quality of concrete in water conservancy projects based on image recognition technology according to claim 1, characterized in that, Extracting visual coarse features from the original image at the current time point includes: The original image at the current time point is preprocessed to obtain a first preprocessed image. ; Calculate the first preprocessed image Gray co-occurrence matrix ; Based on the gray-level co-occurrence matrix Extract the first preprocessed image Contrast ,energy ,entropy and homogeneity characteristics This yields visually coarse features.

4. The method for detecting the quality of concrete in water conservancy projects based on image recognition technology according to claim 1, characterized in that, Extracting crack feature regions from the original image at the current time point includes: The original image at the current time point is preprocessed to obtain a first preprocessed image. ; Calculate the first preprocessed image Each pixel Hessian matrix ; According to the Hessian matrix Calculate pixels First eigenvalue Second eigenvalue ; Based on the first feature value of the pixel Second eigenvalue Calculate the degree of anisotropy Wherein, the anisotropy degree The mathematical expression is: Will satisfy: The pixels are used as crack pixels, where, Indicates the heterogeneity threshold; Density clustering is performed on the crack pixels to obtain multiple pixel clusters, and pixel clusters with a number of pixels greater than or equal to a preset threshold are taken as crack clusters; and crack regions are constructed based on the crack clusters.

5. The method for detecting the quality of concrete in water conservancy projects based on image recognition technology according to claim 1, characterized in that, The ultrasonic roughness features include echo peak amplitude, spectral center frequency, and attenuation coefficient. Extracting the ultrasonic roughness features from the ultrasonic echo signal includes: The ultrasonic echo signal is preprocessed to obtain a preprocessed echo signal, wherein the preprocessing includes filtering, amplification and gain compensation. The time window for surface reflected waves is extracted from the preprocessed echo signal, and the absolute maximum value of the signal is extracted from the time window to obtain the echo peak amplitude. ; Extract the spectral center frequency from the preprocessed echo signal ; Extract the echo envelope of the preprocessed echo signal. Construct a decay model within a selected time window, wherein the mathematical expression of the decay model is: In the formula, Indicates the initial amplitude. Indicates the attenuation coefficient. Indicates the time of transmission; Sampling is performed within the selected time window to obtain multiple sampling time points and their envelope amplitude values. The attenuation model is then linearly fitted based on these multiple sampling time points and their envelope amplitude values ​​to obtain the attenuation coefficient. .

6. The method for detecting the quality of concrete in hydraulic engineering based on image recognition technology according to claim 1, characterized in that, The method for constructing the first fitting model or the second fitting model includes: Acquire multiple concrete surface image samples and ultrasonic echo signal samples of the concrete surface; Visual roughness feature samples are extracted from the concrete surface image samples, and ultrasonic roughness feature samples are extracted from the ultrasonic echo signal samples, wherein the visual roughness feature samples include contrast samples. Energy sample Entropy samples and homogeneous feature samples The ultrasonic roughness feature samples include echo peak amplitude samples. Spectrum center frequency sample and attenuation coefficient samples ,in, For concrete surface image sample index, For the ultrasonic echo signal sample index; The ultrasonic coarse feature samples and the visual coarse feature samples are normalized respectively to obtain normalized ultrasonic coarse feature samples and normalized visual coarse feature samples. Construct a first multivariate linear relationship and a second multivariate linear relationship, wherein the mathematical expression of the first multivariate linear relationship is: The mathematical expression for the second multivariate linear relationship is: In the formula, Indicates the visual roughness of the annotation. This indicates the labeled ultrasonic smoothness. and Both represent intercepts. Represents the contrast coefficient. Indicates normalized contrast. Represents the coefficient of the energy term. Represents normalized energy. Represents the coefficient of the entropy term. Represents the normalized entropy. Represents the coefficient of the homogeneity characteristic term. Indicates normalized homogeneity characteristics. This represents the coefficient of the peak echo amplitude term. Indicates the normalized echo peak amplitude. Represents the coefficient of the center frequency term in the spectrum. Indicates the center frequency of the normalized spectrum. This represents the coefficient of the attenuation coefficient term. Represents the normalized attenuation coefficient. and All are error terms; Substituting the normalized visual roughness feature samples into the first multivariate linear relationship and combining it with the least squares method for fitting, a first fitting model is obtained; and substituting the normalized ultrasonic roughness feature samples into the second multivariate linear relationship and combining it with the least squares method for fitting, a second fitting model is obtained.

7. The method for detecting the quality of concrete in water conservancy projects based on image recognition technology according to claim 6, characterized in that, Visual roughness is calculated based on a pre-built first fitting model and the visual roughness features, and ultrasonic smoothness is calculated based on a pre-built second fitting model and the ultrasonic roughness features, including: The visual roughness features are normalized to obtain normalized visual roughness features; and the ultrasonic roughness features are normalized to obtain normalized ultrasonic roughness features. Substituting the normalized visual roughness features into the first fitting model, we obtain the visual roughness. The normalized ultrasonic roughness features are then substituted into the second fitting model to obtain the ultrasonic smoothness. .

8. The method for detecting the quality of concrete in water conservancy projects based on image recognition technology according to claim 1, characterized in that, The visual roughness is compared with the ultrasonic smoothness to obtain a roughness comparison result, and the water seepage situation is determined based on the roughness comparison result, including: Calculate the visual roughness With the aforementioned ultrasonic smoothness Difference rate , wherein the difference rate The mathematical expression is: The difference rate If the difference exceeds the preset threshold, it is determined that there is water seepage on the concrete surface.

9. The method for detecting the quality of concrete in water conservancy projects based on image recognition technology according to claim 1, characterized in that, Real-time monitoring of concrete quality in hydraulic engineering projects is conducted based on the comparison results of the water seepage and cracks, including: When water seeps into the concrete surface, and the area of ​​the cracked area in the current inspection cycle is larger than the area of ​​the cracked area in the previous inspection cycle, an alarm message will be sent to the target object.

10. A concrete quality inspection system for hydraulic engineering based on image recognition technology, characterized in that, include: The acquisition module is used to acquire an image sequence of the concrete area to be inspected in a hydraulic engineering project, wherein the image sequence is acquired based on a pre-set image acquisition device; The difference calculation module is used to extract the original image at the current time point and the original image at the previous time point from the image sequence, and to calculate the difference between the original image at the current time point and the original image at the previous time point; An ultrasonic detection module is used to control an ultrasonic detection device to emit ultrasonic signals toward the area to be detected and obtain echo signals when the difference is greater than a preset difference threshold. The feature extraction module is used to extract visual roughness features and crack feature regions from the original image at the current time point, and to extract ultrasonic roughness features from the ultrasonic echo signal; The roughness calculation module is used to calculate visual roughness based on a pre-built first fitting model and the visual roughness feature, and to calculate ultrasonic smoothness based on a pre-built second fitting model and the ultrasonic roughness feature, wherein the first fitting model characterizes the relationship between visual roughness feature and roughness, and the second fitting model characterizes the relationship between ultrasonic roughness feature and roughness. The comparison module is used to compare the visual roughness with the ultrasonic smoothness to obtain a roughness comparison result, and to determine the water seepage situation based on the roughness comparison result; and to compare the crack feature area with the crack feature reference area of ​​the previous detection cycle to obtain a crack comparison result. The detection module is used to perform real-time detection of the concrete quality of water conservancy projects based on the comparison results of the water seepage and the cracks.

Images