Overhanging rock instability early warning method based on digital image correlation technology and gray features

Abstract

The invention belongs to the technical field of geological disaster monitoring, and particularly discloses a draping rock instability early warning method based on digital image correlation technologies and gray features. The method comprises the following steps: collecting a draping rock image and carrying out geometric correction, denoising and contrast enhancement; utilizing a DIC analysis image to obtain displacement field and strain field data, and storing the displacement field and strain field data as a strain field matrix; converting the strain field matrix into a gray value matrix; the gray value matrix generates a gray value histogram, calculates a gray average GM and a gray standard deviation GS, and draws a corresponding characteristic curve; constructing an early warning model of a first stage with a characteristic curve in a high-frequency oscillation trend, a second stage with a smooth descending trend and a third stage with a characteristic curve tending to be stable; and the early warning model reads the GM and GS characteristic curves, judges corresponding stages, and outputs early warning signals of'safety 'in the first stage, 'attention' in the second stage and'warning 'in the third stage according to judgment. The method has the characteristics of high timeliness, less human intervention and accurate judgment.

Description

Technical Field

[0001] The present invention relates to the field of geological disaster monitoring technology, and in particular to an overhang rock instability early warning method based on digital image correlation technology and grayscale features, which has high timeliness, less human intervention and accurate judgment. Background Art

[0002] Overhang rock is a rock mass with potential collapse risk formed on slopes or cliffs due to external forces such as weathering, rainfall, and earthquakes. The instability and collapse of overhang rock not only poses a threat to engineering structures, but may also cause major safety accidents. Therefore, accurately predicting the stability of overhang rock and conducting effective early warning have become important issues in the current prevention and control of geological disasters.

[0003] At present, traditional monitoring of overhanging rock stability mostly relies on manual inspection, survey, radar, sensor monitoring and other means, but manual inspection and survey have problems such as poor timeliness and reliance on human experience, while radar and sensor monitoring have the problem of limited data, which makes the accuracy of early warning low. In recent years, digital image correlation technology (DIC) is a non-contact, non-interference full-field deformation measurement method. It compares the digital images of the object before and after deformation, and uses the feature points or feature areas in the image to match the images before and after deformation, so as to calculate the displacement and deformation of the object surface. DIC is widely used in rock deformation monitoring due to its advantages such as extremely loose experimental environment requirements, strong anti-interference ability, and high measurement accuracy.

[0004] In the prior art, DIC is applied to overhang rock monitoring, which can accurately capture the slight changes in rock deformation by collecting displacement field or strain field images on the surface of the rock mass. However, the captured rock deformation information is generally displayed in the form of three-dimensional images, curve graphs, charts, etc., and then researchers intuitively analyze the deformation of the rock mass. However, the analysis based on images and charts by researchers is not only slow and labor-intensive, resulting in low timeliness, but also limited by the knowledge level and sense of responsibility of the researchers, making the accuracy and reliability of the analysis conclusions insufficient. Therefore, how to efficiently analyze and automatically judge the instability of overhang rocks is still a technical problem. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention provides an overhang rock instability early warning method based on digital image correlation technology and grayscale features, which has high timeliness, less human intervention and accurate judgment.

[0006] The present invention is implemented in this way: including data acquisition, image preprocessing, deformation field calculation, grayscale processing, grayscale index calculation, early warning model construction, monitoring and early warning steps, the specific contents are: A. Data collection: Regularly collect surface images of key structural parts of overhanging rocks under external forces; B. Image preprocessing: geometric correction, noise removal and contrast enhancement of the collected images; C. Deformation field calculation: The preprocessed image sequence is analyzed using DIC technology to obtain the displacement field and strain field data of all image sequences, and the strain field data is saved as a strain field matrix at different time points; D. Grayscale processing: convert the aforementioned strain field matrix into grayscale values ​​according to the grayscale standard to generate a standardized grayscale value matrix; E. Gray index calculation: Generate gray value histograms at different time points based on the gray value matrix. When there are no cracks on the surface of the overhanging rock, most of the grayscale of the gray value histogram is concentrated in the middle of the grayscale interval. After cracks appear on the surface of the overhanging rock, the overall low gray level percentage of the gray value histogram increases sharply, and as the cracks of the overhanging rock gradually expand, the gray level of the gray value histogram continues to move to a lower gray level; then calculate the gray mean GM and gray standard deviation GS of all pixel points at each moment in the gray value matrix, and then plot the changes of GM and GS at different moments over time into corresponding characteristic curves; F. Construction of early warning model: Based on the characteristic curves of GM and GS, early warning models are constructed for the first stage when the strain field distribution on the overhanging rock surface is relatively uniform, i.e., the characteristic curve shows a high-frequency oscillation trend; the second stage when high-strain localized areas gradually appear on the overhanging rock surface, i.e., the characteristic curve shows a smooth downward trend; and the third stage when the cracks on the overhanging rock surface enter the stable expansion stage, i.e., the characteristic curve tends to be stable; G. Monitoring and early warning: The early warning model reads the GM characteristic curve and the GS characteristic curve obtained in step E and determines the stages corresponding to the characteristic curves. According to the judgment, the early warning signals of "safety" for the first stage, "attention" for the second stage and "alert" for the third stage are output.

[0007] Furthermore, the specific process of step A is as follows: A1. Select monitoring area: Select the overhanging rock including dangerous rock mass, main controlling joint surface and key structural parts of bedrock and the area containing natural cracks, joints or weathered layers as the monitoring area; A2. Equipment configuration: Install high-resolution camera equipment around the monitoring area of ​​the overhanging rock, and adjust the angle to optimize the contrast and clarity of the areas with weak natural texture on the overhanging rock surface; A3. Image acquisition: Using high-resolution camera equipment, based on the stability of the monitored overhang rock mass, regularly collect surface images of the overhang rock in the monitoring area under the action of external forces to ensure that the image resolution, light conditions and contrast meet the requirements of DIC analysis.

[0008] Furthermore, in the step B, the intrinsic and extrinsic parameters of the high-resolution camera device are obtained and applied, the tilt or stretching that may occur in the captured image is corrected using a perspective correction tool, and the lens distortion is removed using a distortion correction algorithm. Then, the image noise after geometric correction is removed by Gaussian filtering, median filtering and / or bilateral filtering, and finally, the contrast of the denoised image is enhanced by histogram equalization, adaptive histogram equalization and / or gamma correction.

[0009] Furthermore, the specific process of step C is as follows: C1. Displacement field calculation: import the preprocessed image sequence into Ncorr software. Ncorr software takes the first image or the specified image as the reference image and the remaining images as the deformed images, and compares them with the reference image one by one. Ncorr software calculates the displacement of each pixel by identifying the changes of natural speckles on the surface of the overhanging rock in the image, and then generates a two-dimensional displacement field reflecting the movement of the overhanging rock surface at different time points, and exports the displacement field into MATLAB format or other data formats; C2. Strain field calculation: Based on the above displacement field, Ncorr software obtains the strain field composed of the strain values ​​of each pixel point reflecting the deformation of the overhanging rock surface by analyzing the gradient change of the displacement; at the same time, Ncorr software visualizes the strain field in the form of a color map and saves the strain field data as a strain field matrix at different time points.

[0010] Furthermore, the specific process of step D is as follows: D1. Numerical conversion: The strain values ​​in the strain field matrix are mapped linearly according to the proportion of the minimum value corresponding to grayscale 0, the maximum value corresponding to grayscale 255, and the intermediate values. Each strain value in the strain field matrix is ​​mapped to a grayscale value between 0 and 255. The mapping formula is: , Where: G is the gray value of the corresponding pixel, E is the strain value of the corresponding pixel point, E min and E max are the minimum and maximum strain values ​​in the strain field data, respectively; D2, matrix generation: convert the image strain field matrix at different time points into the corresponding gray value matrix; D3. Multi-time grayscale matrix processing: directly compare the grayscale matrix at different time points and quantify the evolution trend of strain value over time, analyze the dynamic change of strain field and crack extension to determine the speed and scale of crack extension.

[0011] Furthermore, the specific process of step E is as follows: E1. Gray value histogram: The gray value matrix is ​​processed using the imhist function in MATLAB to generate a gray value histogram with the horizontal axis representing the gray value and the vertical axis representing the corresponding number of pixels; then, the gray value histograms at different time points are compared to observe the temporal evolution of the strain field: when there are no cracks on the surface of the overhanging rock, most of the grayscale of the gray value histogram is concentrated in the middle of the grayscale interval, and the graph forms a "short and fat" shape; after cracks appear on the surface of the overhanging rock, the overall low gray level percentage of the gray value histogram increases sharply, and the graph forms a "thin and tall" shape on the left; as the cracks of the overhanging rock gradually expand, the gray level of the gray value histogram moves from a low gray level to a high gray level, and the graph forms a "thin and tall" shape on the right; E2. Grayscale mean calculation: The grayscale mean GM represents the mean grayscale of all pixels at each moment, reflecting the overall deformation intensity of the rock surface. The corresponding expression is: , Where: k is the gray level, p r ( k ) is the gray level k The frequency of occurrence of p r ( k ) is calculated as follows: , Where: n k The gray level is k The number of pixels is obtained from the gray value histogram at the corresponding moment; n is the total number of pixels in the image; E3. Grayscale standard deviation calculation: Grayscale standard deviation GS reflects the local concentration of strain and can quantify the uniformity or local concentration of strain distribution on the overhanging rock surface. The corresponding expression is: ; E4. Curve generation: The changes of GM and GS at different moments with time are plotted into GM-time curve and GS-time curve.

[0012] Furthermore, the early warning model in step F is constructed by the following three stages: F1, initial stress stage: In the initial stage of overhanging rock subjected to external force, the strain field undergoes slight adjustments, resulting in the GM characteristic curve and the GS characteristic curve showing a high-frequency oscillation trend at this stage, indicating that the microcracks and microdefects inside the rock mass are gradually closing, but there is no obvious sign of crack extension. This stage is defined as the first stage; F2, crack initiation: As the external force increases, high strain localized areas gradually appear on the surface of the overhanging rock, and the risk of crack initiation in the main joint surface area increases, resulting in a smooth downward trend in the GM characteristic curve and the GS characteristic curve. This stage is defined as the second stage; F3, crack extension: The crack extension of overhanging rock is accelerated, and the localized strain area on the surface expands rapidly, resulting in a significant increase in the strain value in this area while other areas maintain lower strain values. The GM characteristic curve and the GS characteristic curve tend to be stable, and this stage is defined as the third stage.

[0013] Furthermore, the G step also includes precursor point identification: the early warning model identifies the dividing point between the second stage and the third stage of the GM characteristic curve and the GS characteristic curve in real time as the precursor point, and outputs prompt information for immediately taking protective measures including grouting and anchoring after the precursor point is identified.

[0014] Furthermore, the precursor point is obtained by respectively identifying the second-stage slopes of the GM characteristic curve and the GS characteristic curve. If the slopes corresponding to n consecutive time points of any curve are all 0, it indicates that the curve enters the third stage, and the first time point of these n consecutive time points is taken as the precursor point.

[0015] Beneficial effects of the present invention: 1. The present invention is based on digital image correlation technology (DIC), combined with the early warning model constructed by segmentation of grayscale characteristic curves. It can not only give full play to the advantages of DIC in terms of extremely loose requirements for experimental environment, strong anti-interference ability, and high measurement accuracy, but also automatically analyze the strain field characteristic curve generated by DIC through the early warning model, thereby predicting the stability of overhanging rocks in real time and generating multi-level early warning signals, providing a decision-making basis for geological disaster prevention, and effectively avoiding the problems of low timeliness, accuracy, and reliability of existing manual analysis of DIC charts.

[0016] 2. The present invention can analyze the distribution of grayscale values ​​of the strain field at different time points by graying the strain field matrix generated by DIC and calculating the grayscale index. This is not only convenient for understanding the trend of strain concentration in overhang rocks, but also can evaluate the localization of crack extension through the two characteristics of grayscale mean GM and grayscale standard deviation GS. Moreover, the generated characteristic curve can not only reflect the dynamic evolution of strain, but also provide a data basis for the judgment of the early warning model, thereby providing an important basis for predicting the instability of overhang rocks.

[0017] In summary, the present invention has the characteristics of high timeliness, less human intervention and accurate judgment. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 is a flow chart of the present invention; Figure 2 A schematic diagram of a typical overhanging rock of the present invention; Figure 3 An example diagram of the steps of calculating the deformation field to the grayscale index in the present invention; Figure 4 This is an example diagram of the gray value histogram evolution before crack expansion in the present invention; Figure 5 This is an example diagram of the gray value histogram evolution after crack expansion in the present invention; Figure 6 This is an example diagram of the early warning model in the present invention; In the figure: 1-bedrock, 2-dangerous rock mass, 3-free surface, 4-main controlling joint surface; Figure 3 a - calculated strain field, Figure 3 b-converted to strain field matrix, Figure 3 c- obtain the effective strain area (other areas excluding the main controlling joint surface, mainly including bedrock and dangerous rock mass), Figure 3 d- Grayscale processing of the strain field matrix, Figure 3 e-generates a gray value histogram, Figure 3 f- Calculate and draw characteristic curves; Figure 4 a-strain field I, Figure 4 b- Grayscale of strain field I, Figure 4 c- gray value histogram Figure I ; Figure 5 a-strain field II, Figure 5 b- Grayscale of strain field II, Figure 5 c- gray value histogram Figure II . DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0020] like Figures 1 to 6 As shown, the present invention includes data acquisition, image preprocessing, deformation field calculation, grayscale processing, grayscale index calculation, early warning model construction, monitoring and early warning steps, and the specific contents are: A. Data collection: Regularly collect surface images of key structural parts of overhanging rocks under external forces; B. Image preprocessing: geometric correction, noise removal and contrast enhancement of the collected images; C. Deformation field calculation: The preprocessed image sequence is analyzed using DIC technology to obtain the displacement field and strain field data of all image sequences, and the strain field data is saved as a strain field matrix at different time points; D. Grayscale processing: convert the aforementioned strain field matrix into grayscale values ​​according to the grayscale standard to generate a standardized grayscale value matrix; E. Gray index calculation: Generate gray value histograms at different time points based on the gray value matrix. When there are no cracks on the surface of the hanging rock, most of the grayscale of the gray value histogram is concentrated in the middle of the grayscale interval. After cracks appear on the surface of the hanging rock, the overall low grayscale percentage of the gray value histogram increases sharply, and as the cracks of the hanging rock gradually expand, the grayscale of the gray value histogram moves from a low grayscale to a high grayscale. Then calculate the grayscale mean GM and grayscale standard deviation GS of all pixels at each moment in the gray value matrix, and then plot the changes of GM and GS at different moments over time into corresponding characteristic curves. F. Construction of early warning model: Based on the characteristic curves of GM and GS, early warning models are constructed for the first stage when the strain field distribution on the overhanging rock surface is relatively uniform, i.e., the characteristic curve shows a high-frequency oscillation trend; the second stage when high-strain localized areas gradually appear on the overhanging rock surface, i.e., the characteristic curve shows a smooth downward trend; and the third stage when the cracks on the overhanging rock surface enter the stable expansion stage, i.e., the characteristic curve tends to be stable; G. Monitoring and early warning: The early warning model reads the GM characteristic curve and the GS characteristic curve obtained in step E and determines the stages corresponding to the characteristic curves. According to the judgment, the early warning signals of "safety" for the first stage, "attention" for the second stage and "alert" for the third stage are output.

[0021] The specific process of step A is as follows: A1. Select monitoring area: Select the overhanging rock including dangerous rock mass, main controlling joint surface and key structural parts of bedrock and the area containing natural cracks, joints or weathered layers as the monitoring area; A2. Equipment configuration: Install high-resolution camera equipment around the monitoring area of ​​the overhanging rock, and adjust the angle to optimize the contrast and clarity of the areas with weak natural texture on the overhanging rock surface; A3. Image acquisition: Using high-resolution camera equipment, based on the stability of the monitored overhang rock mass, regularly collect surface images of the overhang rock in the monitoring area under the action of external forces to ensure that the image resolution, light conditions and contrast meet the requirements of DIC analysis.

[0022] In the step B, the intrinsic and extrinsic parameters of the high-resolution camera device are obtained and applied, the tilt or stretching that may appear in the captured image is corrected using a perspective correction tool, and the lens distortion is removed using a distortion correction algorithm. Then, the image noise after geometric correction is removed by Gaussian filtering, median filtering and / or bilateral filtering. Finally, histogram equalization, adaptive histogram equalization and / or gamma correction are used to enhance the contrast of the denoised image.

[0023] The specific process of step C is as follows: C1. Displacement field calculation: import the preprocessed image sequence into Ncorr software. Ncorr software takes the first image or the specified image as the reference image and the remaining images as the deformed images, and compares them with the reference image one by one. Ncorr software calculates the displacement of each pixel by identifying the changes of natural speckles on the surface of the overhanging rock in the image, and then generates a two-dimensional displacement field reflecting the movement of the overhanging rock surface at different time points, and exports the displacement field into MATLAB format or other data formats; C2. Strain field calculation: Based on the above displacement field, Ncorr software obtains the strain field composed of the strain values ​​of each pixel point reflecting the deformation of the overhanging rock surface by analyzing the gradient change of the displacement; at the same time, Ncorr software visualizes the strain field in the form of a color map and saves the strain field data as a strain field matrix at different time points.

[0024] The specific process of step D is as follows: D1. Numerical conversion: The strain values ​​in the strain field matrix are mapped linearly according to the proportion of the minimum value corresponding to grayscale 0, the maximum value corresponding to grayscale 255, and the intermediate values. Each strain value in the strain field matrix is ​​mapped to a grayscale value between 0 and 255. The mapping formula is: , Where: G is the gray value of the corresponding pixel, E is the strain value of the corresponding pixel point, E min and E max are the minimum and maximum strain values ​​in the strain field data, respectively; D2, matrix generation: convert the image strain field matrix at different time points into the corresponding gray value matrix; D3. Multi-time grayscale matrix processing: directly compare the grayscale matrix at different time points and quantify the evolution trend of strain value over time, analyze the dynamic change of strain field and crack extension to determine the speed and scale of crack extension.

[0025] The specific process of step E is as follows: E1. Gray value histogram: The gray value matrix is ​​used with the imhist function in MATLAB to generate a gray value histogram with the horizontal axis representing the gray value and the vertical axis representing the number of corresponding pixels; then the gray value histograms at different time points are compared to observe the evolution of the strain field over time: when there are no cracks on the overhanging rock surface (such as Figure 4 As shown in the figure, most of the grayscale values ​​of the grayscale value histogram are concentrated in the middle of the grayscale interval, and the graph forms a "short and fat" shape; when cracks appear on the surface of the overhanging rock (such as Figure 5As shown in the figure, the percentage of low gray levels in the gray value histogram increases sharply, and the figure forms a "thin and tall" shape on the left side. As the cracks of the overhanging rock gradually expand, the gray levels of the gray value histogram move from low gray levels to high gray levels, and the figure forms a "thin and tall" shape on the right side. E2. Grayscale mean calculation: Grayscale mean GM represents the mean grayscale of all pixels at each moment, reflecting the overall deformation intensity of the rock surface. The corresponding expression is: , Where: k is the gray level, p r ( k ) is the gray level k The frequency of occurrence of p r ( k ) is calculated as follows: , Where: n k The gray level is k The number of pixels is obtained from the gray value histogram at the corresponding moment; n is the total number of pixels in the image; E3. Grayscale standard deviation calculation: Grayscale standard deviation GS reflects the local concentration of strain and can quantify the uniformity or local concentration of strain distribution on the overhanging rock surface. The corresponding expression is: ; E4. Curve generation: The changes of GM and GS at different moments with time are plotted into GM-time curve and GS-time curve.

[0026] like Figure 6 As shown, the early warning model in step F is constructed by the following three stages: F1, initial stress stage: In the initial stage of overhanging rock subjected to external force, the strain field undergoes slight adjustments, resulting in the GM characteristic curve and the GS characteristic curve showing a high-frequency oscillation trend at this stage, indicating that the microcracks and microdefects inside the rock mass are gradually closing, but there is no obvious sign of crack extension. This stage is defined as the first stage; F2, crack initiation: As the external force increases, high strain localized areas gradually appear on the surface of the overhanging rock, and the risk of crack initiation in the main joint surface area increases, resulting in a smooth downward trend in the GM characteristic curve and the GS characteristic curve. This stage is defined as the second stage; F3, crack extension: The crack extension of overhanging rock is accelerated, and the localized strain area on the surface expands rapidly, resulting in a significant increase in the strain value in this area while other areas maintain lower strain values. The GM characteristic curve and the GS characteristic curve tend to be stable, and this stage is defined as the third stage.

[0027] The G step also includes precursor point identification: the early warning model identifies the dividing point between the second stage and the third stage of the GM characteristic curve and the GS characteristic curve as the precursor point in real time, and outputs prompt information for immediately taking protective measures including grouting and anchoring after the precursor point is identified.

[0028] The precursor point is obtained by respectively identifying the second-stage slopes of the GM characteristic curve and the GS characteristic curve. If it is identified that the slopes corresponding to n consecutive time points of any curve are all 0, it indicates that the curve enters the third stage, and the first time point of these n consecutive time points is taken as the precursor point.

[0029] Example 1 S100: Regularly collect surface images of key structural parts of overhanging rocks under external forces. The specific process is as follows: S110: Select the overhanging rock including dangerous rock mass, main controlling joint surface and key structural parts of bedrock and the area containing natural cracks, joints or weathered layers as the monitoring area.

[0030] like Figure 2 As shown in Figure 1, the rock mass is the main part of the overhanging rock, which is often located above the bedrock and hanging on the slope or cliff. The rock mass is therefore the area where strain concentration occurs and is therefore the focus of monitoring.

[0031] The main controlling joint surface is a weak structure between the dangerous rock mass and the bedrock, and the rock mass usually slides or cracks along this surface. Therefore, monitoring the stress concentration and crack propagation in the main controlling joint surface area is crucial for early warning of instability.

[0032] Although the bedrock is relatively stable, monitoring its tiny deformation can help determine the overall stress state of the rock mass and the instability potential of the dangerous rock mass.

[0033] When selecting monitoring areas, priority should be given to key structural parts with rich surface textures, especially areas with natural cracks, joints or weathering layers. The textures in these areas provide good natural speckle patterns, which are convenient for subsequent DIC (digital image correlation) analysis. Compared with artificial speckle, natural speckle does not require additional spraying materials, so it is more suitable for monitoring in field environments.

[0034] S120: Install industrial cameras with at least 50 million pixels around the monitoring area of ​​the overhanging rock. While ensuring that the area to be monitored is fully covered, the angle should be adjusted to optimize the contrast and clarity of the area with weak natural texture on the surface of the overhanging rock. If necessary, artificial light sources such as adjustable LED fill lights can be configured near the industrial camera to ensure clear images under different weather or time conditions.

[0035] S130: Based on the stability of the monitored overhanging rock mass and other actual conditions, the above industrial camera is used to regularly collect surface images of the overhanging rock in the monitoring area under the action of external forces (e.g., for overhanging rocks with poor stability, images are collected once an hour; for relatively stable rock masses, the collection frequency can be set to once a day), and ensure that the image resolution, light conditions and contrast meet the requirements of DIC analysis. If it is found that some images are of unqualified quality, they should be adjusted and re-collected immediately.

[0036] S200: Obtain and apply the intrinsic and extrinsic parameters of the aforementioned industrial camera (usually obtained through camera calibration), use the perspective correction tool to correct the tilt or stretch that may appear in the captured image, and apply the distortion correction algorithm to remove lens distortion, then use Gaussian filtering, median filtering and / or bilateral filtering to remove the image noise after geometric correction, and finally use histogram equalization, adaptive histogram equalization and / or Gamma correction to enhance the contrast of the denoised image, so as to ensure the accuracy of subsequent strain field calculation. Contrast enhancement processing can improve the recognizability of speckle in the image and ensure that the DIC algorithm can accurately capture the changes in details in the image.

[0037] Perspective distortion correction: To correct perspective distortion, a common method is to define a reference plane in the image and project the image back to the front view using a projection transformation. This embodiment can implement this operation through the getPerspectiveTransform() and warpPerspective() functions provided by OpenCV.

[0038] Lens distortion correction: The intrinsic parameters of the lens (such as focal length, distortion coefficient, etc.) are obtained through camera calibration, and then these parameters are used to correct the image. In this embodiment, the undistort() function in OpenCV is used to perform image correction based on known camera parameters to ensure the consistency of the speckle image with the actual physical surface.

[0039] Gaussian filtering: This embodiment implements Gaussian filtering through the GaussianBlur() function in OpenCV.

[0040] Median filtering: This embodiment implements median filtering through the medianBlur() function in OpenCV.

[0041] Bilateral filtering: This embodiment implements bilateral filtering through the bilateralFilter() function in OpenCV, which is particularly suitable for denoising while retaining the edge features of natural speckles.

[0042] Histogram equalization: It can make more pixel values ​​in the image distributed in the full grayscale range of 0 to 255, equalize the bright and dark parts in the image, and make the speckle pattern clearer. In this embodiment, histogram equalization is implemented through the equalizeHist() function in OpenCV.

[0043] Adaptive Histogram Equalization (CLAHE): For natural speckle images with local brightness differences, CLAHE can better highlight local details. This embodiment implements CLAHE through the createCLAHE() function in OpenCV.

[0044] Gamma correction: When the brightness distribution of the natural speckle image is uneven, gamma correction can be used to adjust the bright and dark parts of the image to an ideal state. In this embodiment, the pow() function in OpenCV is combined with different gamma values ​​to adjust the overall brightness of the image, thereby optimizing the contrast.

[0045] S300: Analyze the pre-processed image sequence using DIC technology to obtain the displacement field and strain field data of all image sequences, and save the strain field data as strain field matrices at different time points. The specific process is as follows: S310: Import the preprocessed image sequence into Ncorr software. Ncorr software uses the first image or the designated image as the reference image and the remaining images as the deformed images, and compares them with the reference image one by one. Ncorr software identifies the changes in the natural speckle on the surface of the overhanging rock in the image, calculates the displacement of each pixel, and then generates a two-dimensional displacement field reflecting the movement of the overhanging rock surface at different time points, and exports the displacement field into MATLAB format or other data formats.

[0046] S320: Based on the above displacement field, the Ncorr software obtains the strain field composed of the strain values ​​of each pixel point reflecting the deformation of the overhanging rock surface by analyzing the gradient change of the displacement (such as Figure 3 a. Figure 4 a and Figure 5 a); at the same time, the Ncorr software visualizes the strain field in the form of a color map, so that high strain areas can be identified by color changes, and possible crack initiation points and expansion trends can be determined; finally, the strain field data is saved as a strain field matrix at different time points (such as Figure 3 b and 3c), in the strain field matrix, each element represents the strain value at that position.

[0047] S400: The strain field matrix is ​​numerically converted according to the grayscale standard to generate a standardized grayscale value matrix. The specific process is as follows: S410: The strain values ​​in the strain field matrix are linearly mapped in proportion to grayscale values ​​between 0 and 255, with the minimum value corresponding to grayscale 0 and the maximum value corresponding to grayscale 255, and the intermediate values ​​are linearly mapped in proportion, so that each strain value in the strain field matrix is ​​mapped to a grayscale value between 0 and 255. The mapping formula is: , Where: G is the gray value of the corresponding pixel, E is the strain value of the corresponding pixel point, E min and E max are the minimum and maximum strain values ​​in the strain field data, respectively; after this mapping, all strain values ​​are normalized to grayscale values ​​from 0 to 255.

[0048] S420: Convert the image strain field matrix at different time points into a corresponding gray value matrix (such as Figure 3 d. Figure 4 b and Figure 5 b).

[0049] S430: By directly comparing the gray value matrices at different time points, the dynamic changes of the strain field and the expansion of the crack can be analyzed; at the same time, the evolution trend of the strain value over time can be quantified (such as calculating the rate of change of the gray value of a certain area), so as to determine the speed and scale of crack expansion.

[0050] S500: Generate gray value histograms at different time points according to the gray value matrix (such as Figure 3 e), when there is no crack on the overhanging rock surface, most of the grayscale values ​​in the grayscale value histogram are concentrated in the middle of the grayscale interval (such as Figure 4 , Figure 4 c), and after cracks appear on the surface of the overhanging rock, the overall low gray level percentage of the gray value histogram increases sharply, and as the cracks of the overhanging rock gradually expand, the gray level of the gray value histogram moves from low gray level to high gray level (as shown in Figure 5 , Figure 5 c); then calculate the grayscale mean GM and grayscale standard deviation GS of all pixels at each moment in the grayscale value matrix, and then plot the changes of GM and GS at different moments over time into corresponding characteristic curves (such as Figure 3 f). The specific process is as follows: S510: The gray value matrix is ​​used with the imhist function in MATLAB to generate a gray value histogram with the horizontal axis representing the gray value and the vertical axis representing the corresponding number of pixels; then the gray value histograms at different time points are compared to observe the evolution of the strain field over time (with the expansion of the crack, the histogram will change significantly): When there are no cracks on the surface of the overhanging rock, the strain field is evenly distributed, the overall strain value is small, most of the grayscale of the grayscale value histogram is concentrated in the middle of the grayscale interval, and the graph forms a "short and fat" shape; after cracks appear on the surface of the overhanging rock, the increase in the strain difference between the strain concentration area and other areas causes the overall low grayscale level (i.e., darker area) percentage of the grayscale value histogram to increase sharply, and the graph forms a "thin and tall" shape on the left; as the cracks in the overhanging rock gradually expand, the strain value in the strain localization area continues to increase gradually, and the difference in strain values ​​with other areas also gradually increases. The grayscale of the grayscale value histogram continues to move toward the low grayscale level (i.e., darker area), and the graph forms a "thin and tall" shape on the left, and the peak frequency generally shows a trend of gradual increase.

[0051] S520: Grayscale mean GM represents the mean grayscale of all pixels at each moment, reflecting the overall deformation intensity of the rock surface. Its corresponding expression is: , Where: k is the gray level, p r ( k ) is the gray level k The frequency of occurrence of p r ( k ) is calculated as follows: , Where: n k The gray level is k The number of pixels is obtained from the gray value histogram at the corresponding moment; n is the total number of pixels in the image.

[0052] S530: Grayscale standard deviation GS reflects the local concentration of strain and can quantify the uniformity or local concentration of strain distribution on the overhanging rock surface. The corresponding expression is: .

[0053] S540: The changes of GM and GS at different moments with time are plotted into GM-time curve and GS-time curve to show the strain evolution process of the overhanging rock mass. By observing the above curves, the deformation behavior at different stages can be identified, and the instability risk of the rock mass can be inferred.

[0054] S600: Figure 6 As shown, according to the characteristic curves of GM and GS, the early warning model is constructed by the following three stages: S610: In the early stage of overhanging rock subjected to external force, the strain field distribution is relatively uniform and slight adjustments occur inside, resulting in the GM characteristic curve and the GS characteristic curve showing a high-frequency oscillation trend at this stage, reflecting that the strain value is constantly changing within a small amplitude range, representing that the microcracks and microdefects inside the rock body are gradually closing, but no obvious crack extension traces have yet appeared. This stage is defined as the first stage. The curve defining the overhanging rock is the first stage, which ends at the time point corresponding to the curve where the turning point first appears in the first stage of the GM and GS characteristic curves. The gray value histogram of this stage shows a concentrated distribution in the middle area, and the graph forms a typical "short and fat" shape, that is, it presents a relatively symmetrical and compact form.

[0055] S620: As the external force increases, high strain localized areas gradually appear on the surface of the overhanging rock, and the risk of crack initiation in the main control joint surface area increases, resulting in a smooth downward trend in the GM characteristic curve and the GS characteristic curve. This stage is defined as the second stage. The second stage of the curve defining the overhanging rock starts from the time point corresponding to the curve where the turning point first appears in the first stage of the GM and GS characteristic curves. The gray value histogram of this stage gradually moves to the low gray level (i.e., the darker area), reflecting that the strain value in the high strain area increases and the brightness decreases. The shape of the gray value histogram gradually changes from "short and fat" to "thin and tall", that is, the distribution range of the gray level widens and the frequency of the dark area increases.

[0056] S630: The crack propagation of overhanging rock accelerates, and the localized strain area of ​​the surface expands rapidly, resulting in a significant increase in the strain value of this area while other areas maintain lower strain values, making the localized strain difference more obvious. The GM characteristic curve and the GS characteristic curve tend to be stable, and this stage is defined as the third stage. The curve defining the overhanging rock enters the third stage, starting from the time point corresponding to the curve of the first turning point in the second stage of the GM and GS characteristic curves. The gray value histogram of this stage is concentrated in the low gray level, and the shape becomes more "thin and tall", indicating that the strain values ​​corresponding to most pixels tend to lower gray values.

[0057] S700: The early warning model reads the GM characteristic curve and the GS characteristic curve obtained in step S500, determines the stage corresponding to the characteristic curve, and outputs the early warning signal of "safety" for the first stage, "caution" for the second stage, and "alert" for the third stage according to the corresponding judgment. Among them, the overhanging rock in the first stage is still in a stable state, and the risk of cracks is low, so it can be considered that the overhanging rock mass is in a "safe" state; the overhanging rock in the second stage enters an unstable stage, and the trend of crack initiation gradually emerges, so it can be considered that the overhanging rock is in a "caution" state; the cracks in the overhanging rock in the third stage have gradually formed and developed, and the further expansion of the cracks in the overhanging rock may lead to instability, so "early warning" and measures are needed.

[0058] The boundary point between the second and third stages of the characteristic curve evolution is considered to be the precursor point of the initiation of cracks in the overhanging rock mass (by identifying the second stage slopes of the GM characteristic curve and the GS characteristic curve respectively, if the slopes corresponding to n consecutive time points of any curve are all 0, it indicates that the curve has entered the third stage, and the first time point of these n consecutive time points is taken as the precursor point), marking the transition of the crack from the initial initiation stage to the stable expansion stage. Therefore, in the early warning model detection process, the precursor points of the GM characteristic curve and the GS characteristic curve are identified in real time, and after the precursor point is identified, prompt information is output to immediately take protective measures including grouting and anchoring, so as to delay or prevent the instability process of the rock mass and ensure the safety of the structure.

[0059] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by any technician familiar with the technical field within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for early warning of overhanging rock instability based on digital image correlation technology and grayscale features, characterized by: It includes data acquisition, image preprocessing, deformation field calculation, grayscale processing, grayscale index calculation, early warning model construction, monitoring and early warning steps. The specific contents are as follows: A. Data collection: Regularly collect surface images of key structural parts of overhanging rocks under external forces; B. Image preprocessing: geometric correction, noise removal and contrast enhancement of the collected images; C. Deformation field calculation: The preprocessed image sequence is analyzed using DIC technology to obtain the displacement field and strain field data of all image sequences, and the strain field data is saved as a strain field matrix at different time points; D. Grayscale processing: convert the aforementioned strain field matrix into grayscale values ​​according to the grayscale standard to generate a standardized grayscale value matrix; E. Gray index calculation: Generate gray value histograms at different time points based on the gray value matrix. When there are no cracks on the surface of the overhanging rock, most of the grayscale of the gray value histogram is concentrated in the middle of the grayscale interval. After cracks appear on the surface of the overhanging rock, the overall low grayscale percentage of the gray value histogram increases sharply. As the cracks of the overhanging rock gradually expand, the grayscale of the gray value histogram continues to move to a lower grayscale level. Then, based on the gray value histogram, calculate the grayscale mean GM and grayscale standard deviation GS of all pixel points at each moment in the gray value matrix, and then plot the changes of GM and GS at different moments over time into corresponding characteristic curves. F. Construction of early warning model: Based on the characteristic curves of GM and GS, early warning models are constructed for the first stage when the strain field distribution on the overhanging rock surface is relatively uniform, i.e., the characteristic curve shows a high-frequency oscillation trend; the second stage when high-strain localized areas gradually appear on the overhanging rock surface, i.e., the characteristic curve shows a smooth downward trend; and the third stage when the cracks on the overhanging rock surface enter the stable expansion stage, i.e., the characteristic curve tends to be stable; G. Monitoring and early warning: The early warning model reads the GM characteristic curve and the GS characteristic curve obtained in step E and determines the stages corresponding to the characteristic curves. According to the judgment, the early warning signals of "safety" for the first stage, "attention" for the second stage and "alert" for the third stage are output.

2. The overhang rock instability early warning method based on digital image correlation technology and grayscale features according to claim 1 is characterized by: The specific process of step A is as follows: A1. Select monitoring area: Select the overhanging rock including dangerous rock mass, main controlling joint surface and key structural parts of bedrock and the area containing natural cracks, joints or weathered layers as the monitoring area; A2. Equipment configuration: Install high-resolution camera equipment around the monitoring area of ​​the overhanging rock, and adjust the angle to optimize the contrast and clarity of the areas with weak natural texture on the overhanging rock surface; A3. Image acquisition: Using high-resolution camera equipment, based on the stability of the monitored overhang rock mass, regularly collect surface images of the overhang rock in the monitoring area under the action of external forces to ensure that the image resolution, light conditions and contrast meet the requirements of DIC analysis.

3. The overhang rock instability early warning method based on digital image correlation technology and grayscale features according to claim 2 is characterized by: In the step B, the intrinsic and extrinsic parameters of the high-resolution camera device are obtained and applied, the tilt or stretching that may appear in the captured image is corrected using a perspective correction tool, and the lens distortion is removed using a distortion correction algorithm. Then, the image noise after geometric correction is removed by Gaussian filtering, median filtering and / or bilateral filtering. Finally, histogram equalization, adaptive histogram equalization and / or gamma correction are used to enhance the contrast of the denoised image.

4. The overhang rock instability early warning method based on digital image correlation technology and grayscale features according to claim 1 is characterized by: The specific process of step C is as follows: C1. Displacement field calculation: import the preprocessed image sequence into Ncorr software. Ncorr software takes the first image or the specified image as the reference image and the remaining images as the deformed images, and compares them with the reference image one by one. Ncorr software calculates the displacement of each pixel by identifying the changes of natural speckles on the surface of the overhanging rock in the image, and then generates a two-dimensional displacement field reflecting the movement of the overhanging rock surface at different time points, and exports the displacement field into MATLAB format or other data formats; C2. Strain field calculation: Based on the above displacement field, Ncorr software obtains the strain field composed of the strain values ​​of each pixel point reflecting the deformation of the overhanging rock surface by analyzing the gradient change of the displacement; at the same time, Ncorr software visualizes the strain field in the form of a color map and saves the strain field data as a strain field matrix at different time points.

5. The overhang rock instability early warning method based on digital image correlation technology and grayscale features according to claim 1 is characterized by: The specific process of step D is as follows: D1. Numerical conversion: The strain values ​​in the strain field matrix are mapped linearly according to the proportion of the minimum value corresponding to grayscale 0, the maximum value corresponding to grayscale 255, and the intermediate values. Each strain value in the strain field matrix is ​​mapped to a grayscale value between 0 and 255. The mapping formula is: ， Where: G is the gray value of the corresponding pixel, E is the strain value of the corresponding pixel point, E min and E max are the minimum and maximum strain values ​​in the strain field data, respectively; D2, matrix generation: convert the image strain field matrix at different time points into the corresponding gray value matrix; D3. Multi-time grayscale matrix processing: directly compare the grayscale matrix at different time points and quantify the evolution trend of strain value over time, analyze the dynamic change of strain field and crack extension to determine the speed and scale of crack extension.

6. The overhang rock instability early warning method based on digital image correlation technology and grayscale features according to claim 1 is characterized by: The specific process of step E is as follows: E1. Gray value histogram: The gray value matrix is ​​processed using the imhist function in MATLAB to generate a gray value histogram with the horizontal axis representing the gray value and the vertical axis representing the number of corresponding pixels; then, the gray value histograms at different time points are compared to observe the temporal evolution of the strain field: when there are no cracks on the surface of the overhanging rock, most of the grayscale of the gray value histogram is concentrated in the middle of the grayscale interval, and the graph forms a "short and fat" shape; after cracks appear on the surface of the overhanging rock, the overall low gray level percentage of the gray value histogram increases sharply, and the graph forms a "thin and tall" shape on the left; as the cracks of the overhanging rock gradually expand, the gray level of the gray value histogram moves from a low gray level to a high gray level, and the graph forms a "thin and tall" shape on the right; E2. Grayscale mean calculation: Grayscale mean GM represents the mean grayscale of all pixels at each moment, reflecting the overall deformation intensity of the rock surface. The corresponding expression is: ， Where: k is the gray level, p r ( k ) is the gray level k The frequency of occurrence of p r ( k ) is calculated as follows: ， Where: n k The gray level is k The number of pixels is obtained from the gray value histogram at the corresponding moment; n is the total number of pixels in the image; E3. Grayscale standard deviation calculation: Grayscale standard deviation GS reflects the local concentration of strain and can quantify the uniformity or local concentration of strain distribution on the overhanging rock surface. The corresponding expression is: ； E4. Curve generation: The changes of GM and GS at different moments with time are plotted into GM-time curve and GS-time curve.

7. The overhang rock instability early warning method based on digital image correlation technology and grayscale features according to claim 1 is characterized by: The early warning model in step F is constructed by the following three stages: F1, initial stress stage: In the initial stage of overhanging rock subjected to external force, the strain field undergoes slight adjustments, resulting in the GM characteristic curve and the GS characteristic curve showing a high-frequency oscillation trend at this stage, indicating that the microcracks and microdefects inside the rock mass are gradually closing, but there is no obvious sign of crack extension. This stage is defined as the first stage; F2, crack initiation: As the external force increases, high strain localized areas gradually appear on the surface of the overhanging rock, and the risk of crack initiation in the main joint surface area increases, resulting in a smooth downward trend in the GM characteristic curve and the GS characteristic curve. This stage is defined as the second stage; F3, crack extension: The crack extension of overhanging rock is accelerated, and the localized strain area on the surface expands rapidly, resulting in a significant increase in the strain value in this area while other areas maintain lower strain values. The GM characteristic curve and the GS characteristic curve tend to be stable, and this stage is defined as the third stage.

8. The overhang rock instability early warning method based on digital image correlation technology and grayscale features according to claim 1 is characterized by: The G step also includes precursor point identification: the early warning model identifies the dividing point between the second stage and the third stage of the GM characteristic curve and the GS characteristic curve as the precursor point in real time, and outputs prompt information for immediately taking protective measures including grouting and anchoring after the precursor point is identified.

9. The overhang rock instability early warning method based on digital image correlation technology and grayscale features according to claim 8 is characterized by: The precursor point is obtained by respectively identifying the second-stage slopes of the GM characteristic curve and the GS characteristic curve. If it is identified that the slopes corresponding to n consecutive time points of any curve are all 0, it indicates that the curve enters the third stage, and the first time point of these n consecutive time points is taken as the precursor point.

Images