Dangerous rock mass shear zone quality evaluation method based on machine vision

Abstract

The invention discloses a dangerous rock mass shear zone quality evaluation method based on machine vision, and relates to the technical field of dangerous rock mass monitoring and evaluation, and the method comprises the following steps: S1, arranging a high-resolution camera, adjusting the angle and focal length, and regularly and automatically collecting and transmitting image data to a server; s2, de-noising by using a filter, adjusting a contrast ratio, equalizing a histogram, and enhancing image definition; s3, extracting key features and covering key areas by applying edge detection and feature point detection algorithms; canny edge detection and an SIFT algorithm are applied to perform feature extraction, a convolutional neural network (CNN) is adopted to perform model training and cross validation, shear band changes are monitored in real time, early warning is triggered, analysis results are displayed through visual tools such as a thermodynamic diagram, a standardized report is generated, normalization and consistency of the monitoring results are ensured, and the monitoring accuracy is improved. Therefore, the monitoring efficiency and accuracy of the dangerous rock mass shear zone are improved, and the geological disaster risk is effectively prevented.

Description

Technical Field

[0001] The invention relates to the technical field of dangerous rock mass monitoring and evaluation, and in particular to a dangerous rock mass shear zone quality evaluation method based on machine vision. Background Art

[0002] In the field of geological disaster prevention and control, dangerous rock masses refer to those rock masses whose structures are loose, broken, or about to collapse due to natural factors or human activities. The existence of dangerous rock masses poses a huge threat to human life and property safety, so the monitoring and evaluation of dangerous rock masses has always been one of the focuses of geological disaster prevention and control work. Traditional dangerous rock mass monitoring methods mainly rely on manual observation and on-site detection, which not only has a large workload and low efficiency, but also has strong subjectivity, making it difficult to achieve continuous and real-time monitoring.

[0003] In recent years, with the rapid development of image processing technology and artificial intelligence technology, monitoring methods based on machine vision have been widely used in various fields. Machine vision technology obtains image data through camera equipment and uses image processing algorithms to extract and analyze features, which can achieve efficient and accurate monitoring and evaluation. In the monitoring of dangerous rock masses, machine vision technology can make up for the shortcomings of traditional methods. Through high-resolution cameras and intelligent analysis systems, continuous and real-time monitoring of dangerous rock masses can be achieved. The shear zone of dangerous rock mass is the most vulnerable part of the dangerous rock mass, and the quality of the shear zone is directly related to the stability of the dangerous rock mass. The shear zone refers to the cross-section or fracture zone formed inside the rock mass due to shear stress. Its characteristics are usually dense cracks, scattered rock blocks, loose structure, etc. Accurate quality assessment of the shear zone of dangerous rock mass can timely discover potential dangers, take effective prevention and control measures, and prevent disasters.

[0004] Traditional methods for assessing the quality of shear bands in dangerous rock masses mainly rely on on-site investigations and laboratory tests. On-site investigations require professionals to perform high-altitude operations, which poses certain safety risks; laboratory tests require sampling, which is a complex and time-consuming process. With the development of science and technology, shear band quality assessment methods based on machine vision and artificial intelligence have gradually gained attention. These methods use cameras to collect shear band images, and use image processing and machine learning techniques for analysis and evaluation, which not only improves work efficiency, but also ensures the objectivity and accuracy of monitoring results.

[0005] At present, some studies have attempted to apply machine vision technology to the monitoring of dangerous rock masses. By detecting and analyzing the cracks, displacements and other features on the surface of the rock mass, the automated monitoring of dangerous rock masses has been initially realized. However, most of these studies focus on the monitoring of the rock mass surface, and there are still certain limitations for the changes in the internal structure, especially the quality assessment of the shear zone. At the same time, due to the influence of external factors such as terrain and lighting, machine vision technology still faces many challenges in practical applications, such as unstable image quality and difficulty in feature extraction. Therefore, there is an urgent need for a more complete technical solution that can comprehensively utilize machine vision and artificial intelligence technology to conduct a comprehensive and accurate quality assessment of the shear zone of the dangerous rock mass. Summary of the invention

[0006] The present invention provides a method for evaluating the quality of shear zones in dangerous rock masses based on machine vision. The method collects images by arranging high-resolution cameras, performs image preprocessing by using technologies such as Gaussian filtering and histogram equalization, extracts features by using Canny edge detection and SIFT algorithm, and uses convolutional neural networks (CNN) for model training and cross-validation. The method performs image segmentation and structural analysis, realizes intelligent evaluation of shear zone quality and risk identification, monitors shear zone changes in real time and triggers early warning, displays analysis results through visualization tools such as heat maps and generates standardized reports, ensures the standardization and consistency of monitoring results, thereby improving the efficiency and accuracy of shear zone monitoring in dangerous rock masses and effectively preventing geological disaster risks.

[0007] In order to achieve the above object, the present invention provides the following technical solution: a method for evaluating the quality of shear zones of dangerous rock masses based on machine vision, comprising the following steps:

[0008] S1, arrange high-resolution cameras, adjust angles and focal lengths, and automatically collect and transmit image data to the server regularly;

[0009] S2, use filters to remove noise, adjust contrast and histogram equalization, and enhance image clarity;

[0010] S3, apply edge detection and feature point detection algorithms to extract key features and cover key areas;

[0011] S4. Label the data and select the training model, using cross-validation to improve classification and segmentation accuracy;

[0012] S5, segment the image, analyze the structural characteristics of the shear band, evaluate the quality, and identify potential risk areas;

[0013] S6. Collect and analyze data in real time, identify shear zone changes, establish an early warning system, and promptly warn of risks;

[0014] S7. Visualize the analysis results, generate standardized assessment reports regularly, and record the changes in shear bands.

[0015] Preferably, in step S1, a high-resolution camera is arranged in the shear zone of the dangerous rock mass, and the angle and focal length are adjusted to ensure a clear and complete image is obtained. Image data is collected regularly using automated acquisition software and transmitted to a central server via a network to ensure data continuity and integrity: i =T start +i·ΔT, where T i is the time of the ith collection, T start is the initial acquisition time, ΔT is the acquisition interval, and i is the i-th acquisition.

[0016] Preferably, in step S2, the collected image is denoised, a Gaussian filter is used to reduce noise, the filter parameters are adjusted to balance the denoising effect and image detail retention, and contrast adjustment and histogram equalization techniques are used to enhance image clarity to ensure that key features are clearly visible: Among them, G(x,y) is the Gaussian function value at the position (x,y), e is the standard deviation, which determines the width of the filter, and x and y are the coordinates of the pixel; the histogram equalization technology is specifically embodied as follows: Among them, H(v) is the equalization value corresponding to the pixel value v, L is the grayscale level, MN is the width and height of the image, and h(i) is the number of pixels at grayscale level i.

[0017] Preferably, in step S3, the Canny edge detection algorithm is used to extract the edge features of the shear band in the image, and the detection threshold is adjusted to ensure accurate extraction; the SIFT algorithm is applied to extract the key feature points of the shear band, and the distribution and density of the feature points are evaluated to ensure coverage of the key area: the Canny edge detection formula is: E(x,y)=G(x,y)*I(x,y), where E(x,y) is the edge detection result, G(x,y) is the Gaussian filter, I(x,y) is the input image, and * is the convolution operation; the SIFT key point descriptor is: Among them, D is the feature point descriptor, |p i -q i | is the i-th component of the feature point vector, and n is the dimension of the feature point vector.

[0018] Preferably, in step S4, the image data is annotated to mark the cracks and rock blocks of the shear zone; a convolutional neural network (CNN) is selected for training, the performance of different models is evaluated, the optimal model is selected, the model is trained using the training data set, the model parameters are adjusted to improve the classification and segmentation accuracy, the model performance is evaluated using cross-validation and test sets, and the convolution operation is performed: Among them, O(i,j) is the value of the output image at position (i,j), I is the input image, K is the convolution kernel, and m, n are the coordinates of the convolution kernel.

[0019] Preferably, in step S5, the image is segmented using the trained model, different parts of the shear zone are identified, and the accuracy and completeness of the segmentation results are evaluated; the structural characteristics of the shear zone are analyzed, and its integrity and stability are evaluated in combination with geological knowledge and shear zone characteristics, the shear zone quality is comprehensively evaluated based on the structural analysis results, potential risk areas are identified, and standardized evaluation indicators are used to ensure the objectivity and consistency of the results; fully connected layer: y=Wx+b, wherein y is the output, W is the weight matrix, x is the input feature, and b is the bias term; shear zone quality score: Q=α·S+β·T+γ·U+δ·A+□·L+ζ·R+η·F, wherein Q is the shear zone quality score, α, β, γ, δ, □, ζ, η are weight coefficients, S is the structural stability score, T is the crack density score, U is the rock size score, A is the shear zone opening score, L is the shear zone length score, R is the shear zone interface undulation score, and F is the shear zone filling fragmentation score.

[0020] Preferably, in step S6, the image data of the shear band is collected in real time by a camera to ensure the timeliness and integrity of the data; the real-time data is analyzed using a trained model to identify changes in the shear band; an early warning system is established to issue an early warning signal based on the analysis results to remind monitoring personnel to pay attention to potential risks and monitor indicators in real time: Where M(t) is the monitoring index at time t, N is the number of data points, and f(x i ,t) is the characteristic value of the i-th data point at time t; the early warning signal formula is: Where W is the warning signal, M(t) is the monitoring index, and θ is the warning threshold. The specific operation method is as follows: first, real-time image acquisition: the system regularly (e.g., once per second) collects image data of the shear band through the camera and transmits it to the central server. Then, data analysis: use the trained machine learning model to analyze the real-time data and extract key features f(x i ,t), then calculate the monitoring indicators: use the formula Calculate the monitoring index at the current moment and finally trigger the warning: if the monitoring index M(t) exceeds the warning threshold θ, then W=1, the system triggers the warning signal, and reminds the monitoring personnel to pay attention to the potential risk; if M(t)≤θ, then W=0.

[0021] Preferably, in step S7, the analysis results are displayed through visualization tools, such as thermal maps, crack growth maps, etc., so that the monitoring personnel can intuitively understand the shear zone status; a quality assessment report is generated regularly (such as once a week) to record the changes in the shear zone and the assessment results. If significant changes are found, they are marked as key monitoring areas, and standardized report templates and formats are used to ensure the standardization and consistency of the report content; thermal map generation: Among them, H(i,j) is the value of the heat map at position (i,j), w k is the weight of the kth feature, V k (i,j) is the value of the kth feature at position (i,j), n is the number of features; report score summary: Among them, R is the total score, T is the number of time periods, and Q t : Quality score at time t.

[0022] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0023] The present invention provides a method for evaluating the quality of shear zones in dangerous rock masses based on machine vision. The method collects images by arranging high-resolution cameras, performs image preprocessing by using technologies such as Gaussian filtering and histogram equalization, extracts features by using Canny edge detection and SIFT algorithm, and uses convolutional neural networks (CNN) for model training and cross-validation. The method performs image segmentation and structural analysis, realizes intelligent evaluation of shear zone quality and risk identification, monitors shear zone changes in real time and triggers early warning, displays analysis results through visualization tools such as heat maps and generates standardized reports, ensures the standardization and consistency of monitoring results, thereby improving the efficiency and accuracy of shear zone monitoring in dangerous rock masses and effectively preventing geological disaster risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0025] Figure 1 The module diagram of a method for evaluating shear zone quality of dangerous rock mass based on machine vision according to the present invention. DETAILED DESCRIPTION

[0026] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of the present disclosure will be more comprehensive and complete, and the concept of the example embodiments will be fully conveyed to those skilled in the art.

[0027] The present invention provides Figure 1 A method for evaluating the quality of shear zones of dangerous rock masses based on machine vision is shown, comprising the following steps: S1, arranging a high-resolution camera, adjusting the angle and focal length, and automatically collecting and transmitting image data to a server regularly;

[0028] S2, use filters to remove noise, adjust contrast and histogram equalization, and enhance image clarity;

[0029] S3, apply edge detection and feature point detection algorithms to extract key features and cover key areas;

[0030] S4. Label the data and select the training model, using cross-validation to improve classification and segmentation accuracy;

[0031] S5, segment the image, analyze the structural characteristics of the shear band, evaluate the quality, and identify potential risk areas;

[0032] S6. Collect and analyze data in real time, identify shear zone changes, establish an early warning system, and promptly warn of risks;

[0033] S7, visualize the analysis results, generate standardized assessment reports regularly, and record the changes in the shear zone; in step S1, arrange high-resolution cameras in the shear zone area of ​​the dangerous rock mass, adjust the angle and focal length to ensure a clear and complete image, use automated acquisition software to regularly collect image data, and transmit it to the central server through the network to ensure data continuity and integrity: T i =T start +i·ΔT, where T i is the time of the ith collection, T startis the initial acquisition time, ΔT is the acquisition interval time, and i is the i-th acquisition; a high-resolution camera is arranged in the shear zone area of ​​the dangerous rock mass to ensure that the angle and focal length of the camera are precisely adjusted to cover and clearly capture the image of the entire shear zone. The automated acquisition software collects image data regularly, usually in seconds, for example, once per second. Each collected image is transmitted to the central server through the network to ensure the continuity and integrity of the data. During the data transmission process, redundant transmission and data encryption technology are used to ensure that the data is not lost and is safe and reliable. The location of the camera should be selected taking into account the terrain conditions and the specific conditions of the shear zone to ensure that the monitoring area can be fully covered and blind spots can be avoided. At the same time, in order to cope with different weather and lighting conditions, the camera should have night vision and waterproof functions. In addition, to ensure the stability of data acquisition, an automatic correction function is set in the system to calibrate the camera regularly to ensure the consistency and accuracy of image acquisition. Through the above measures, the image acquisition process can efficiently and accurately obtain real-time image data of the shear zone, providing a reliable data basis for subsequent image preprocessing and analysis.

[0034] In step S2, the collected image is denoised by using a Gaussian filter to reduce noise, adjusting the filter parameters to balance the denoising effect and image detail retention, and using contrast adjustment and histogram equalization techniques to enhance image clarity and ensure that key features are clearly visible: Among them, G(x,y) is the Gaussian function value at the position (x,y), e is the standard deviation, which determines the width of the filter, and x and y are the coordinates of the pixel; the histogram equalization technology is specifically embodied as follows: Among them, H(v) is the equalization value corresponding to the pixel value v, L is the grayscale level, MN is the width and height of the image, and h(i) is the number of pixels at grayscale level i; the collected image is preprocessed, and denoising is first performed. Common denoising methods include Gaussian filtering and mean filtering. Gaussian filtering can effectively remove high-frequency noise in the image, while mean filtering reduces the impact of noise by smoothing the image. It is crucial to select appropriate filter parameters, and the filter parameters need to strike a balance between denoising effect and image detail retention. Next, the image is contrast adjusted and histogram equalized to enhance the clarity and recognizability of the image. Contrast adjustment can make the brightness difference of the image more obvious, and histogram equalization improves the overall contrast by redistributing the grayscale values ​​of the image. When performing histogram equalization, it is necessary to ensure that the adjusted image is not distorted and the key features are more prominent. In the last step of image preprocessing, image enhancement techniques such as sharpening and edge enhancement are used to further improve the quality of the image and the visibility of key features. The goal of all these steps is to generate high-quality preprocessed images to ensure that key information in the image is not obscured by noise and low contrast during feature extraction and analysis, providing clear and accurate base images for subsequent steps.

[0035] In step S3, the Canny edge detection algorithm is used to extract the edge features of the shear band in the image, and the detection threshold is adjusted to ensure accurate extraction; the SIFT algorithm is applied to extract the key feature points of the shear band, and the distribution and density of the feature points are evaluated to ensure coverage of the key area: the Canny edge detection formula is: E(x,y)=G(x,y)*I(x,y), where E(x,y) is the edge detection result, G(x,y) is the Gaussian filter, I(x,y) is the input image, and * is the convolution operation; the SIFT key point descriptor is: Among them, D is the feature point descriptor, |p i -q i | is the i-th component of the feature point vector, n is the dimension of the feature point vector, and the Canny edge detection algorithm is used to extract the edge features of the shear band in the preprocessed image. This algorithm can effectively detect the edges in the image, including the important features such as the contour and cracks of the shear band. Canny edge detection extracts the edge of the image by calculating the gradient of the image to identify the area with drastic changes. When applying this algorithm, it is necessary to set an appropriate threshold to ensure that the extracted edges are both complete and accurate. Subsequently, the SIFT algorithm is applied to extract the key feature points of the image. The SIFT algorithm is an algorithm based on scale-invariant feature transformation, which can extract stable feature points at different scales and rotation angles. The extracted feature points include key points in the shear band, such as the starting point and end point of the crack. The distribution and density of the feature points are evaluated to ensure that they can cover all the key areas of the shear band. In the process of feature extraction, other image processing techniques, such as image segmentation and morphological operations, are also combined to further enhance the effect of feature extraction. Through the comprehensive application of multiple algorithms, it is ensured that the key features of the shear band can be accurately and comprehensively extracted, providing rich and accurate data support for the subsequent machine learning model training.

[0036] In step S4, the image data is annotated to mark the cracks and rock blocks of the shear zone; a convolutional neural network (CNN) is selected for training, the performance of different models is evaluated, the optimal model is selected, the model is trained using the training data set, the model parameters are adjusted to improve the classification and segmentation accuracy, and the model performance is evaluated using cross-validation and test sets. The convolution operation: O(i,j)=(I*K)(i,j)=∑ m ∑ nI(im,jn)K(m,n), where O(i,j) is the value of the output image at position (i,j), I is the input image, K is the convolution kernel, and m,n are the coordinates of the convolution kernel; after the feature extraction is completed, the image data is annotated to mark the key areas such as cracks and rock blocks in the shear zone. These annotated data will be input into the machine learning model as a training set. Convolutional neural network (CNN) is selected as the main model. CNN has significant advantages in processing image data. First, the network structure is designed, including input layer, convolution layer, pooling layer and fully connected layer. The design of the network structure needs to be adjusted according to the specific characteristics of the shear zone image to ensure that the model can fully extract and understand the key information in the image. Next, the annotated training data is input into the network for model training. During the training process, the parameters of the network, such as learning rate, convolution kernel size, etc., are adjusted to improve the accuracy and convergence speed of the model. The model is evaluated by cross-validation method. Through multiple training and testing, the model parameters are continuously optimized to ensure the generalization ability of the model. In the final stage of model training, an independent test set is used to evaluate the model and verify its performance on unseen data. After multiple iterations and parameter adjustments, a high-performance convolutional neural network model is finally obtained, which can accurately classify and segment shear band images, providing strong support for structural analysis and quality assessment.

[0037] In step S5, the trained model is used to segment the image, identify different parts of the shear zone, and evaluate the accuracy and completeness of the segmentation results; the structural characteristics of the shear zone are analyzed, and its integrity and stability are evaluated by combining geological knowledge and shear zone characteristics. The quality of the shear zone is comprehensively evaluated based on the structural analysis results, and potential risk areas are identified. Standardized evaluation indicators are used to ensure the objectivity and consistency of the results; the fully connected layer: y = Wx + b, where y is the output, W is the weight matrix, x is the input feature, and b is the bias term; the quality of the shear zone Quantity score: Q = α·S + β·T + γ·U + δ·A + □·L + ζ·R + η·F, where Q is the shear band quality score, α, β, γ, δ, □, ζ, η are weight coefficients, S is the structural stability score, T is the crack density score, U is the rock block size score, A is the shear band opening score, L is the shear band length score, R is the shear band interface undulation score, and F is the shear band filling fragmentation score. After the feature extraction is completed, the image data is annotated to mark the key areas such as cracks and rock blocks in the shear band. These annotated data will be input into the machine learning model as a training set. Convolutional neural network (CNN) is selected as the main model. CNN has significant advantages in processing image data. First, the network structure is designed, including input layer, convolution layer, pooling layer and fully connected layer. The design of the network structure needs to be adjusted according to the specific characteristics of the shear band image to ensure that the model can fully extract and understand the key information in the image. Next, the annotated training data is input into the network for model training. During the training process, the parameters of the network, such as learning rate and convolution kernel size, are adjusted to improve the accuracy and convergence speed of the model. The model is evaluated by cross-validation. Through multiple training and testing, the model parameters are continuously optimized to ensure the generalization ability of the model. In the final stage of model training, the model is evaluated using an independent test set to verify its performance on unseen data. After multiple iterations and parameter adjustments, a high-performance convolutional neural network model is finally obtained, which can accurately classify and segment shear zone images, providing strong support for structural analysis and quality assessment; after that, the structural characteristics of the shear zone are deeply analyzed, and its integrity and stability are evaluated by combining geological knowledge and the specific conditions of the shear zone. The following factors will be focused on in the analysis: Shear zone opening: The shear zone opening refers to the degree of opening of the cracks in the shear zone. The larger the opening, the lower the strength of the shear zone and the worse the stability. Therefore, accurate measurement and analysis of the shear zone opening is crucial to evaluating the stability of the shear zone; Shear zone length: The length of the shear zone is also an important indicator for evaluating its stability. Longer shear zones may affect the rock structure in a larger range and increase the risk of instability. We will use image segmentation and feature extraction technology to accurately measure the length of the shear band and incorporate it into the evaluation model; Shear band interface undulation: The undulation of the shear band interface reflects the complexity of the internal structure of the shear band.High undulation usually means that there are more irregularities and weaknesses inside the shear band, which reduces its overall stability. We will extract and quantify the undulation of the shear band interface through edge detection and morphological processing techniques; Shear band filling thickness: Filling refers to the material that fills the cracks in the shear band. The thickness of the filling has an important impact on the strength of the shear band. The greater the thickness, the more material deposition process the shear band has undergone, which may reduce its stability. We will use image processing technology to measure and analyze the thickness of the filling; Shear band filling fragmentation: The fragmentation of the filling reflects the physical state and stability of the filling. Fillings with high fragmentation are usually loose and difficult to provide sufficient support, which in turn affects the stability of the shear band. Through image detail enhancement and feature analysis techniques, we will evaluate the fragmentation of the filling.

[0038] In step S6, the image data of the shear band is collected in real time by the camera to ensure the timeliness and integrity of the data; the real-time data is analyzed using the trained model to identify the changes in the shear band; an early warning system is established to issue early warning signals based on the analysis results to remind monitoring personnel to pay attention to potential risks and monitor indicators in real time: Where M(t) is the monitoring index at time t, N is the number of data points, and f(x i ,t) is the characteristic value of the i-th data point at time t; the early warning signal formula is: Where W is the warning signal, M(t) is the monitoring index, and θ is the warning threshold. The specific operation method is as follows: first, real-time image acquisition: the system regularly (e.g., once per second) collects image data of the shear band through the camera and transmits it to the central server. Then, data analysis: use the trained machine learning model to analyze the real-time data and extract key features f(x i ,t), then calculate the monitoring indicators: use the formula The monitoring index at the current moment is calculated and finally triggered to warn: if the monitoring index M(t) exceeds the warning threshold θ, then W=1, the system triggers the warning signal, and reminds the monitoring personnel to pay attention to the potential risk; if M(t)≤θ, then W=0, and the image data of the shear zone is collected in real time through the camera to ensure the timeliness and integrity of the data. The real-time data is transmitted to the central server through the network, and the real-time data is analyzed by the trained convolutional neural network model to extract key features and calculate the monitoring index at the current moment. The monitoring index is calculated by analyzing the changes in cracks in the image, the movement of rock blocks and other information to reflect the real-time status of the shear zone. If the monitoring index exceeds the warning threshold, the system automatically triggers the warning signal to remind the monitoring personnel to pay attention to the potential risk. The warning signal is conveyed to the monitoring personnel through various channels, including text messages, emails, and alarm prompts on the monitoring platform. A multi-level warning mechanism is set up in the system, and different levels of warning signals are issued according to different situations of the monitoring indicators. For example, when the monitoring index slightly exceeds the threshold, a first-level warning is issued to remind the monitoring personnel to pay attention; when the monitoring index greatly exceeds the threshold, a second-level warning is issued, requiring immediate preventive measures. Through real-time monitoring and early warning mechanisms, abnormal changes in shear zones can be discovered in a timely manner, rapid responses can be made, and effective measures can be taken to prevent potential geological disaster risks and improve the timeliness and reliability of dangerous rock mass monitoring and management.

[0039] In step S7, the analysis results are displayed through visualization tools, such as thermal maps, crack growth maps, etc., so that the monitoring personnel can intuitively understand the shear zone status; a quality assessment report is generated regularly (such as once a week) to record the changes in the shear zone and the assessment results. If significant changes are found, they are marked as key monitoring areas, and standardized report templates and formats are used to ensure the standardization and consistency of the report content; Thermal map generation: Among them, H(i,j) is the value of the heat map at position (i,j), w k is the weight of the kth feature, V k (i,j) is the value of the kth feature at position (i,j), n is the number of features; report score summary: Among them, R is the total score, T is the number of time periods, and Q t:Quality score at time t; Display the analysis results through visualization tools, such as heat map, crack growth map, etc., so that monitoring personnel can intuitively understand the status of the shear zone. The heat map shows the high-risk areas in the image, and the risk level is indicated by color changes. Red indicates high-risk areas and blue indicates low-risk areas. The crack growth map records the changes in cracks, including the starting point, end point and growth rate of the cracks. Generate a quality assessment report regularly (such as once a week) to record the changes in the shear zone and the evaluation results. The report includes data sources, analysis methods, result presentation, key monitoring areas, etc. to ensure the standardization and consistency of the report. If significant changes are found, it will be marked as a key monitoring area in the report, and it is recommended to increase the monitoring frequency and analysis intensity of the area. The report uses a standardized template to facilitate comparative analysis of reports in different time periods. Through the generation and presentation of the report, monitoring personnel can fully understand the status of the shear zone, discover potential problems in a timely manner, and take corresponding measures. In addition, the data and analysis results in the report can also provide reference for subsequent geological research and engineering decision-making, and improve the scientificity and effectiveness of dangerous rock mass monitoring and management.

[0040] The above embodiments are implemented in whole or in part by software, hardware, firmware or any other combination. When implemented using software, the above embodiments are implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer is a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions are stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions are transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired (e.g., infrared, wireless, microwave, etc.) mode. The computer-readable storage medium is any available medium that a computer can access or a data storage device such as a server or data center containing one or more available media sets. The available medium is a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium is a solid-state hard disk.

[0041] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0042] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0043] The units described as separate components are or are not physically separated, and the components shown as units are or are not physical units, that is, located in one place or distributed on multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0044] In addition, each functional unit in each embodiment of the present application is integrated into one processing unit, each unit exists physically separately, or two or more units are integrated into one unit.

[0045] If the functions are implemented in the form of software functional units and sold or used as independent products, they are stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or partly in other words, the part that contributes to the prior art or part of the technical solution is embodied in the form of a software product, and the computer software product is stored in a storage medium, including several instructions for enabling a computer device (such as a personal computer, a server, or a network device) to perform all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media for storing program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM), random access memories (RAM), magnetic disks or optical disks.

[0046] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A method for evaluating the quality of shear zones in dangerous rock mass based on machine vision, characterized in that: The following steps are involved: S1, arrange high-resolution cameras, adjust angles and focal lengths, and automatically collect and transmit image data to the server regularly; S2, use filters to remove noise, adjust contrast and histogram equalization, and enhance image clarity; S3, apply edge detection and feature point detection algorithms to extract key features and cover key areas; S4. Label the data and select the training model, using cross-validation to improve classification and segmentation accuracy; S5, segment the image, analyze the structural characteristics of the shear band, evaluate the quality, and identify potential risk areas; S6. Collect and analyze data in real time, identify shear zone changes, establish an early warning system, and promptly warn of risks; S7. Visualize the analysis results, generate standardized assessment reports regularly, and record the changes in shear bands.

2. The method for evaluating the quality of shear zones in dangerous rock masses based on machine vision according to claim 1, characterized in that: In step S1, a high-resolution camera is arranged in the shear zone of the dangerous rock mass, and the angle and focal length are adjusted to ensure a clear and complete image. Image data is collected regularly using automated acquisition software and transmitted to a central server via the network to ensure data continuity and integrity: T i =T start +i·ΔT, where T i is the time of the ith collection, T start is the initial acquisition time, ΔT is the acquisition interval, and i is the i-th acquisition.

3. The method for evaluating the quality of shear zones in dangerous rock masses based on machine vision according to claim 1, characterized in that: In step S2, the collected image is denoised by using a Gaussian filter to reduce noise, adjusting the filter parameters to balance the denoising effect and image detail retention, and using contrast adjustment and histogram equalization techniques to enhance image clarity and ensure that key features are clearly visible: Among them, G(x,y) is the Gaussian function value at the position (x,y), e is the standard deviation, which determines the width of the filter, and x and y are the coordinates of the pixel; the histogram equalization technology is specifically embodied as follows: Among them, H(v) is the equalization value corresponding to the pixel value v, L is the grayscale level, MN is the width and height of the image, and h(i) is the number of pixels at grayscale level i.

4. The method for evaluating the quality of shear zones in dangerous rock masses based on machine vision according to claim 1, characterized in that: In step S3, the Canny edge detection algorithm is used to extract the edge features of the shear band in the image, and the detection threshold is adjusted to ensure accurate extraction; the SIFT algorithm is applied to extract the key feature points of the shear band, and the distribution and density of the feature points are evaluated to ensure coverage of the key area: the Canny edge detection formula is: E(x,y)=G(x,y)*I(x,y), where E(x,y) is the edge detection result, G(x,y) is the Gaussian filter, I(x,y) is the input image, and * is the convolution operation; the SIFT key point descriptor is: Among them, D is the feature point descriptor, |p i -q i | is the i-th component of the feature point vector, and n is the dimension of the feature point vector.

5. The method for evaluating the quality of shear zones in dangerous rock masses based on machine vision according to claim 1, characterized in that: In step S4, the image data is annotated and characterized, the cracks and rock blocks of the shear zone are marked, and the length and spacing of the internal cracks in the shear zone, the undulation of the shear zone interface, and the size of the internal filling rock blocks are obtained; a convolutional neural network is selected for training, the performance of different models is evaluated, the optimal model is selected, the model is trained using the training data set, the model parameters are adjusted to improve the classification and segmentation accuracy, and the model performance is evaluated using cross-validation and test sets. Convolution operation: O(i,j)=(I*K)(i,j)=∑ m ∑ n I(im,jn)K(m,n), where O(i,j) is the value of the output image at position (i,j), I is the input image, K is the convolution kernel, and m,n are the coordinates of the convolution kernel.

6. The method for evaluating the quality of shear zones in dangerous rock masses based on machine vision according to claim 1, characterized in that: In step S5, the trained model is used to segment the image, identify different parts of the shear band, and evaluate the accuracy and completeness of the segmentation results; Analyze the structural characteristics of the shear zone, evaluate its integrity and stability by combining geological knowledge and shear zone characteristics, comprehensively evaluate the quality of the shear zone based on the structural analysis results, identify potential risk areas, and use standardized evaluation indicators to ensure the objectivity and consistency of the results; Fully connected layer: y=Wx+b, where y is the output, W is the weight matrix, x is the input feature, and b is the bias term; shear band quality score: Q=α·S+β·T+γ·U+δ·A+□·L+ζ·R+η·F, where Q is the shear band quality score, α, β, γ, δ, □, ζ, η are weight coefficients, S is the structural stability score, T is the crack density score, U is the rock size score, A is the shear band opening score, L is the shear band length score, R is the shear band interface undulation score, and F is the shear band filling fragmentation score.

7. The method for evaluating the quality of shear zones in dangerous rock masses based on machine vision according to claim 1, characterized in that: In step S6, the image data of the shear band is collected in real time by the camera to ensure the timeliness and integrity of the data; the real-time data is analyzed using the trained model to identify the changes in the shear band; an early warning system is established to issue early warning signals based on the analysis results to remind monitoring personnel to pay attention to potential risks and monitor indicators in real time: Where M(t) is the monitoring index at time t, N is the number of data points, and f(x i ,t) is the characteristic value of the i-th data point at time t; the early warning signal formula is: Where W is the warning signal, M(t) is the monitoring index, and θ is the warning threshold. The specific operation method is as follows: first, real-time image acquisition: the system regularly (e.g., once per second) collects image data of the shear band through the camera and transmits it to the central server. Then, data analysis: use the trained machine learning model to analyze the real-time data and extract key features f(x i ,t), then calculate the monitoring indicators: use the formula Calculate the monitoring index at the current moment and finally trigger the warning: if the monitoring index M(t) exceeds the warning threshold θ, then W=1, the system triggers the warning signal, and reminds the monitoring personnel to pay attention to the potential risk; if M(t)≤θ, then W=0.

8. The method for evaluating the quality of shear zones in dangerous rock masses based on machine vision according to claim 1, characterized in that: In step S7, the analysis results are displayed through visualization tools, such as thermal maps, crack growth maps, etc., so that the monitoring personnel can intuitively understand the shear zone status; a quality assessment report is generated regularly (such as once a week) to record the changes in the shear zone and the assessment results. If significant changes are found, they are marked as key monitoring areas, and standardized report templates and formats are used to ensure the standardization and consistency of the report content; Thermal map generation: Among them, H(i,j) is the value of the heat map at position (i,j), w k is the weight of the kth feature, V k (i,j) is the value of the kth feature at position (i,j), n is the number of features; report score summary: Among them, R is the total score, T is the number of time periods, and Q t : Quality score at time t.

Images