Tunnel surrounding rock structure deformation measuring system based on machine vision

Abstract

The invention relates to the technical field of tunnel surrounding rock structure deformation measurement, and provides a tunnel surrounding rock structure deformation measurement system based on machine vision, a visual recording module performs multi-angle image recording on the position of a tunnel surrounding rock structure, and a data preprocessing module performs gray processing on the image transmitted by the visual recording module; the structure deformation analysis module obtains tunnel surrounding rock structure images under the safe condition, and compares and measures the characteristic value difference value of the two images according to the images transmitted by the data preprocessing module. The cross validation module verifies the analysis result of the multi-angle recorded image in the structural deformation analysis module, the early warning module performs early warning according to the verification result of the cross validation module, and the standby measurement module monitors the operation state of an industrial camera in the visual recording module in real time. When the industrial camera in the visual recording module cannot work, access is carried out; according to the invention, the requirements of real-time monitoring and accurate early warning of surrounding rock deformation in the tunnel operation process are met.

Description

Technical Field

[0001] The invention relates to the technical field of tunnel deformation measurement, in particular to a tunnel surrounding rock structure deformation measurement system based on machine vision. Background Art

[0002] Tunnel surrounding rock deformation is an important parameter to characterize the mechanical response of rock mass under excavation disturbance. It reflects the overall stability characteristics and safety effects of the support and surrounding rock system during operation, and provides basic information for the inversion of rock mass parameters. Especially when laying cables in tunnels, surrounding rock deformation is closely related to the safe operation of cables. The deformation of surrounding rock may cause the cables to be squeezed, stretched or sheared, affecting the normal function of the cables, and may even damage the insulation layer of the cables, causing safety accidents such as leakage. Therefore, accurate acquisition and reliable prediction of surrounding rock deformation are the basic prerequisites for surrounding rock stability evaluation and support effectiveness inspection, and are important measures to ensure the reasonable design of tunnel engineering and safe operation.

[0003] After a tunnel is put into operation, the surrounding rock structure may continue to slowly deform over time and under the influence of external factors. If cables are laid in the tunnel, the deformation of the surrounding rock may cause the cables to be squeezed, stretched, and even damaged, causing safety risks such as leakage. Regular deformation measurement can timely detect potential safety hazards, such as lining cracks and water seepage, so that maintenance and reinforcement measures can be taken to ensure the safe use of the tunnel and the normal operation of the cables.

[0004] At the same time, natural disasters such as earthquakes and heavy rains may have an impact on the tunnel surrounding rock. Deformation measurement can quickly assess the damage to the tunnel after a disaster and provide a basis for emergency rescue and repair work. Through long-term deformation monitoring, the deformation laws and trends of the tunnel surrounding rock can be understood, providing a reference for formulating a reasonable maintenance plan.

[0005] The existing background technology still has the following shortcomings:

[0006] 1. System maintenance and continuity. Operating tunnels usually require uninterrupted services. Existing systems may require long downtime during maintenance and upgrades, and incomplete monitoring may lead to data loss or system instability, which brings risks to tunnel operations and affects the normal operation of the tunnel.

[0007] 2. Real-time and early warning accuracy. During the operation process, it is necessary to ensure real-time monitoring of surrounding rock deformation in order to detect problems in a timely manner, and to avoid false alarms and missed alarms. The system needs to more accurately judge whether the deformation reaches the early warning standard. At the same time, it is necessary to consider the impact of factors such as vehicle driving and temperature changes on the measurement results to improve the accuracy of the early warning. Summary of the invention

[0008] In order to solve the above technical problems, the present invention provides a tunnel surrounding rock structure deformation measurement system based on machine vision, which solves the problems in the prior art by performing multi-angle identification measurement and cross-verification on a unified tunnel surrounding rock structure area.

[0009] A tunnel surrounding rock structure deformation measurement system based on machine vision, comprising:

[0010] A visual recording module, which includes a plurality of industrial cameras and is capable of continuously recording the position of the surrounding rock structure of the same tunnel from multiple angles to obtain a color image of the position of the surrounding rock structure of the tunnel;

[0011] A data preprocessing module, which is data-connected to the visual recording module and performs grayscale processing on the color image transmitted by the visual recording module;

[0012] A structural deformation analysis module, which is data-connected to the data preprocessing module, obtains a tunnel surrounding rock structure image in a safe condition and extracts a characteristic value, extracts the tunnel surrounding rock structure image characteristic value based on the image recognition transmitted by the data preprocessing module, and compares and measures the two characteristic values;

[0013] A cross-validation module, which is data-connected with the structural deformation analysis module, cross-validates the analysis results of the images recorded at multiple angles at the same tunnel surrounding rock structure position in the structural deformation analysis module, and determines whether the tunnel surrounding rock structure has any abnormality;

[0014] An early warning module, which is data-connected to the cross-validation module and issues an early warning based on the verification result of the cross-validation module;

[0015] The backup measurement module is data-connected to the visual recording module, provides multiple backup industrial cameras, monitors the operating status of the industrial cameras in the visual recording module in real time, and accesses the industrial cameras in the visual recording module when they fail to work.

[0016] Preferably, the data preprocessing module specifically calculates the grayscale processing of the color image as follows: assuming that the color value of a pixel in the color image is (R, G, B), then its grayscale value is:

[0017] H = 0.299R + 0.587G + 0.114B;

[0018] Among them, (R, G, B) refers to the three color channels of each pixel, red R, green G, and blue B. Each channel is represented by 8 bits, and the value range is 0 to 255; H represents the grayscale value. Each pixel is represented by a grayscale value representing its brightness, ranging from 0 to 255, where 0 is black and 255 is white.

[0019] Preferably, the structural deformation analysis module includes a learning unit, an identification unit and a comparison measurement unit;

[0020] The learning unit obtains a structural grayscale image of a tunnel surrounding rock structure in a safe and undeformed state, calculates a grayscale co-occurrence matrix of the structural image, extracts texture feature values ​​including contrast, correlation, energy and homogeneity according to the grayscale co-occurrence matrix, calculates feature values ​​in multi-angle images of the same tunnel surrounding rock structure, and then calculates the average value thereof as the standard texture feature of the tunnel surrounding rock structure;

[0021] The recognition unit calculates the grayscale co-occurrence matrix according to the grayscale image data transmitted in real time by the data preprocessing module, and extracts texture feature values ​​including contrast, correlation, energy and homogeneity according to the grayscale co-occurrence matrix;

[0022] The comparison and measurement unit performs comparison and measurement based on the texture feature value extracted by the learning unit and the texture feature value extracted by the recognition unit.

[0023] Preferably, the process of extracting texture eigenvalues ​​based on the gray level co-occurrence matrix of the calculated image by the learning unit is as follows;

[0024] Let contrast be C, correlation be X, energy be N and homogeneity be T, perform statistical analysis on these feature values ​​and calculate the average value to obtain the reference texture feature set;

[0025] Let the gray-level co-occurrence matrix P(i,j,d,θ) represent the probability of the pixel with gray value i and the pixel with gray value j appearing at the same time in the direction θ and the distance d;

[0026] The eigenvalues ​​are extracted from the gray-level co-occurrence matrix, and the contrast is calculated as:

[0027]

[0028] Where L is the grayscale level. The grayscale level of the grayscale image is calculated by the bit depth. If its bit depth is ω, then the grayscale level is 2 ω ,The bit depth can be obtained by viewing the image properties;

[0029] The correlation calculation formula is:

[0030]

[0031] where μ i and μ j are the means of grayscale values ​​i and j, σ i and σ j are the standard deviations of grayscale values ​​i and j, respectively;

[0032] The energy calculation formula is:

[0033]

[0034] The formula for calculating homogeneity is:

[0035]

[0036] Assume that the multi-angle images of the surrounding rock structure of the same tunnel have a total of m image regions;

[0037] The average contrast ratio is:

[0038]

[0039] Among them C z is the contrast of the zth image, where z represents the image;

[0040] The average correlation is:

[0041]

[0042] The average energy is:

[0043]

[0044] The mean value of homogeneity is:

[0045]

[0046] Will As standard texture features.

[0047] Preferably, the process of extracting texture eigenvalues ​​based on the gray-level co-occurrence matrix of the calculated image by the recognition unit is consistent with that of the learning unit.

[0048] Preferably, the comparison measurement unit calculates the difference between the texture feature value of the image z extracted by the current recognition unit and the reference texture feature value;

[0049] Contrast Difference:

[0050] Among them C z is the contrast of the current image region z;

[0051] Correlation Difference:

[0052] Energy Difference:

[0053] Homogeneity Difference:

[0054] Combining multiple texture feature differences to construct deformation index;

[0055] Using the weighted summation method, the deformation index Y is:

[0056] Y=αΔC+βΔX+γΔN+δΔT

[0057] Among them, α, β, γ, and δ are weight coefficients, which can be adjusted according to actual conditions to reflect the importance of different texture features in deformation measurement.

[0058] Preferably, the cross-validation module cross-validates the results of comparing images at each angle of the same area, selects two image sets at different angles for comparison, and for each eigenvalue, calculates the absolute value of the difference between the eigenvalues ​​of the two angle sets to verify whether the recognition calculation results are consistent. If not, the visual recording module is notified to re-record the verification calculation.

[0059] Preferably, the early warning module determines the deformation degree of the tunnel surrounding rock structure according to the calculated value of the deformation index Y, and performs corresponding alarm levels by setting different threshold ranges, and the threshold range is confirmed by comparing historical records.

[0060] Compared with the prior art, the present invention has the following beneficial effects:

[0061] 1. The present invention uses multiple industrial cameras in the visual recording module to continuously record the position of the surrounding rock structure of the same tunnel from multiple angles, and combines the grayscale processing image of the data preprocessing module to reduce the amount of information for calculation, making the calculation faster. The coordinated work of the structural deformation analysis module and the cross-validation module realizes all-round real-time monitoring of the tunnel surrounding rock structure, achieves the effect of more accurately judging the deformation of the tunnel surrounding rock structure, effectively avoids false alarms and missed alarms, improves the accuracy of early warning, and meets the needs of real-time monitoring and accurate early warning of surrounding rock deformation during operation.

[0062] 2. The present invention provides multiple backup industrial cameras through a backup measurement module, and monitors the operating status of the industrial cameras in the visual recording module in real time. When the industrial cameras in the visual recording module fail to work, they are accessed, thereby realizing seamless switching of the system and achieving the effect of equipment maintenance and upgrading without interrupting monitoring. It effectively solves the problems of long downtime, incomplete monitoring, data loss and system instability that may occur in the existing system during maintenance and upgrading, thereby ensuring the safety and stability of tunnel operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 It is a work flow chart of the present invention;

[0064] Figure 2 It is the working flow chart of the structural deformation analysis module of the present invention. DETAILED DESCRIPTION

[0065] The following embodiments of the present invention are described in further detail in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0066] The present invention provides a tunnel surrounding rock structure deformation measurement system based on machine vision, comprising:

[0067] A visual recording module, which includes a plurality of industrial cameras and is capable of continuously recording the position of the surrounding rock structure of the same tunnel from multiple angles to obtain a color image of the position of the surrounding rock structure of the tunnel;

[0068] A data preprocessing module, which is data-connected to the visual recording module and performs grayscale processing on the color image transmitted by the visual recording module;

[0069] A structural deformation analysis module, which is data-connected to the data preprocessing module, obtains a tunnel surrounding rock structure image in a safe condition and extracts a characteristic value, extracts the tunnel surrounding rock structure image characteristic value based on the image recognition transmitted by the data preprocessing module, and compares and measures the two characteristic values;

[0070] A cross-validation module, which is data-connected with the structural deformation analysis module, cross-validates the analysis results of the images recorded at multiple angles at the same tunnel surrounding rock structure position in the structural deformation analysis module, and determines whether the tunnel surrounding rock structure has any abnormality;

[0071] An early warning module, which is data-connected to the cross-validation module and issues an early warning based on the verification result of the cross-validation module;

[0072] The backup measurement module is data-connected to the visual recording module, provides multiple backup industrial cameras, monitors the operating status of the industrial cameras in the visual recording module in real time, and accesses the industrial cameras in the visual recording module when they fail to work.

[0073] Example:

[0074] like Figure 1-2 As shown, in this embodiment, a certain important traffic tunnel has a large traffic volume, and it is necessary to ensure the safety of the tunnel structure to ensure the continuous smooth flow of traffic. In addition, the tunnel is long and the geological conditions are complex. The stability of the surrounding rock structure is crucial to the safety of the tunnel, and it is necessary to perform real-time deformation measurement of the tunnel surrounding rock structure.

[0075] The visual recording module configures multiple industrial cameras at key locations in the tunnel to ensure that the surrounding rock structure of the same tunnel can be continuously recorded from multiple angles. For example, a group of industrial cameras is installed at certain intervals on the arch and side walls of the tunnel. Each group of cameras includes three cameras at different angles, which are aimed at different parts of the surrounding rock structure.

[0076] The data preprocessing module is connected to the visual recording module through a high-speed data transmission line. When the visual recording module collects a color image, the data preprocessing module immediately performs grayscale processing on the image.

[0077] If the color value of a pixel in a color image is (R, G, B), then its grayscale value is:

[0078] H = 0.299R + 0.587G + 0.114B;

[0079] Among them, (R, G, B) refers to the three color channels of each pixel, red R, green G, and blue B. Each channel is represented by 8 bits, and the value range is 0 to 255; H represents the grayscale value. Each pixel is represented by a grayscale value representing its brightness, ranging from 0 to 255, where 0 is black and 255 is white.

[0080] Grayscale processing of images can effectively reduce the amount of calculation when processing image calculations, making calculations faster. In real-time scenarios such as tunnel operation, it is more conducive to real-time measurement and makes deformation monitoring more timely.

[0081] The structural deformation analysis module includes a learning unit, an identification unit and a comparative measurement unit;

[0082] The learning unit obtains a structural grayscale image of a tunnel surrounding rock structure in a safe and undeformed state, calculates a grayscale co-occurrence matrix of the structural image, extracts texture feature values ​​including contrast, correlation, energy and homogeneity according to the grayscale co-occurrence matrix, calculates feature values ​​in multi-angle images of the same tunnel surrounding rock structure, and then calculates the average value thereof as the standard texture feature of the tunnel surrounding rock structure;

[0083] First, let contrast be C, correlation be X, energy be N and homogeneity be T, perform statistical analysis on these feature values ​​and calculate the average value to obtain the reference texture feature set;

[0084] Let the gray-level co-occurrence matrix P(i,j,d,θ) represent the probability of the pixel with gray value i and the pixel with gray value j appearing at the same time in the direction θ and the distance d;

[0085] The eigenvalues ​​are extracted from the gray-level co-occurrence matrix, and the contrast is calculated as:

[0086]

[0087] Where L is the grayscale level. The grayscale level of the grayscale image is calculated by the bit depth. If its bit depth is ω, then the grayscale level is 2 ω ,The bit depth can be obtained by viewing the image properties;

[0088] The correlation calculation formula is:

[0089]

[0090] where μ i and μ j are the means of grayscale values ​​i and j, σ i and v j are the standard deviations of grayscale values ​​i and j, respectively;

[0091] The energy calculation formula is:

[0092]

[0093] The formula for calculating homogeneity is:

[0094]

[0095] Assume that the multi-angle images of the surrounding rock structure of the same tunnel have a total of m image regions;

[0096] The average contrast ratio is:

[0097]

[0098] Among them C z is the contrast of the zth image, where z represents the image;

[0099] The average correlation is:

[0100]

[0101] The average energy is:

[0102]

[0103] The mean value of homogeneity is:

[0104]

[0105] Will As standard texture features.

[0106] The recognition unit calculates the grayscale co-occurrence matrix based on the grayscale image data transmitted in real time by the data preprocessing module, and extracts the texture feature values ​​including contrast, correlation, energy and homogeneity based on the grayscale co-occurrence matrix. The calculation process is consistent with that of the learning unit.

[0107] The contrast measurement unit performs contrast measurement based on the texture feature value extracted by the learning unit and the texture feature value extracted by the recognition unit, and first calculates the difference between the texture feature value of the image z extracted by the current recognition unit and the reference texture feature value;

[0108] Contrast Difference:

[0109] Among them C z is the contrast of the current image region z;

[0110] Correlation Difference:

[0111] Energy Difference:

[0112] Homogeneity Difference:

[0113] Combining multiple texture feature differences to construct deformation index;

[0114] Using the weighted summation method, the deformation index Y is:

[0115] Y=αΔC+βΔX+γΔN+δΔT

[0116] Among them, α, β, γ, and δ are weight coefficients, which can be adjusted according to actual conditions to reflect the importance of different texture features in deformation measurement.

[0117] The structural deformation analysis module obtains the characteristic values ​​under safe conditions as the standard through the learning unit, and the recognition unit extracts the characteristic values ​​and compares the measurements in real time. Combined with the cross-validation of the multi-angle image analysis results by the cross-validation module, it achieves the effect of more accurately judging the deformation of the tunnel surrounding rock structure and effectively avoiding false alarms and omissions.

[0118] The cross-validation module cross-validates the results of the image comparison at each angle in the same area, selects two sets of images at different angles for comparison, and for each eigenvalue, calculates the absolute value of the difference between the eigenvalues ​​of the two angle sets to verify whether the recognition calculation results are consistent. If not, the visual recording module is notified to re-record and verify the calculation. For example, for a surrounding rock structure at a specific location, images taken from different angles by multiple cameras are processed by the structural deformation analysis module to obtain their respective deformation results. The cross-validation module compares these results, and if the difference is large, the visual recording module is notified to re-record and verify the calculation.

[0119] The collaborative work of the structural deformation analysis module and the cross-validation module enables all-round real-time monitoring of the tunnel surrounding rock structure.

[0120] The early warning module determines the deformation degree of the tunnel surrounding rock structure according to the calculated value of the deformation index Y, and sets the corresponding alarm level by setting different threshold ranges. The threshold range is confirmed by comparing historical records.

[0121] After the system was put into use in the tunnel, uninterrupted monitoring of the tunnel's surrounding rock structure was achieved. Through multi-angle image acquisition and precise analysis and calculation, it is possible to detect small deformations of the surrounding rock structure in a timely manner, effectively avoiding safety accidents caused by failure to detect deformation in a timely manner. At the same time, the backup measurement module ensures the continuous operation of the system in the event of equipment failure, improving the reliability and stability of the system. The cross-validation module further improves the accuracy of the measurement results and reduces false alarms and missed alarms. The early warning module issues corresponding alarms according to different degrees of deformation, providing timely and effective decision-making basis for the tunnel management department and ensuring the safe operation of the tunnel.

[0122] The embodiments of the present invention are provided for the purpose of illustration and description. Although the embodiments of the present invention have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations of the present invention. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A tunnel surrounding rock structure deformation measurement system based on machine vision, characterized in that: include: A visual recording module, which includes a plurality of industrial cameras and is capable of continuously recording the position of the surrounding rock structure of the same tunnel from multiple angles to obtain a color image of the position of the surrounding rock structure of the tunnel; A data preprocessing module, which is data-connected to the visual recording module and performs grayscale processing on the color image transmitted by the visual recording module; A structural deformation analysis module, which is data-connected to the data preprocessing module, obtains a tunnel surrounding rock structure image in a safe condition and extracts a characteristic value, extracts the tunnel surrounding rock structure image characteristic value based on the image recognition transmitted by the data preprocessing module, and compares and measures the two characteristic values; A cross-validation module, which is data-connected with the structural deformation analysis module, cross-validates the analysis results of the images recorded at multiple angles at the same tunnel surrounding rock structure position in the structural deformation analysis module, and determines whether the tunnel surrounding rock structure has any abnormality; An early warning module, which is data-connected to the cross-validation module and issues an early warning based on the verification result of the cross-validation module; The backup measurement module is data-connected to the visual recording module, provides multiple backup industrial cameras, monitors the operating status of the industrial cameras in the visual recording module in real time, and accesses the industrial cameras in the visual recording module when they fail to work.

2. The machine vision-based tunnel surrounding rock structure deformation measurement system according to claim 1, characterized in that: The specific calculation of the data preprocessing module for grayscale processing of color images is as follows: If the color value of a pixel in a color image is (R, G, B), then its grayscale value is: H = 0.299R + 0.587G + 0.114B; Among them, (R, G, B) refers to the three color channels of each pixel, red R, green G, and blue B. Each channel is represented by 8 bits, and the value range is 0 to 255; H represents the grayscale value. Each pixel is represented by a grayscale value representing its brightness, ranging from 0 to 255, where 0 is black and 255 is white.

3. The machine vision-based tunnel surrounding rock structure deformation measurement system according to claim 1, characterized in that: The structural deformation analysis module includes a learning unit, an identification unit and a comparison measurement unit; The learning unit obtains a structural grayscale image of a tunnel surrounding rock structure in a safe and undeformed state, calculates a grayscale co-occurrence matrix of the structural image, extracts texture feature values ​​including contrast, correlation, energy and homogeneity according to the grayscale co-occurrence matrix, calculates feature values ​​in multi-angle images of the same tunnel surrounding rock structure, and then calculates the average value thereof as the standard texture feature of the tunnel surrounding rock structure; The recognition unit calculates the grayscale co-occurrence matrix according to the grayscale image data transmitted in real time by the data preprocessing module, and extracts texture feature values ​​including contrast, correlation, energy and homogeneity according to the grayscale co-occurrence matrix; The comparison and measurement unit performs comparison and measurement based on the texture feature value extracted by the learning unit and the texture feature value extracted by the recognition unit.

4. The machine vision-based tunnel surrounding rock structure deformation measurement system according to claim 3, characterized in that: The process of extracting texture feature values ​​based on the gray level co-occurrence matrix of the calculated image by the learning unit is as follows; Let contrast be C, correlation be X, energy be N and homogeneity be T, perform statistical analysis on these feature values ​​and calculate the average value to obtain the reference texture feature set; Let the gray-level co-occurrence matrix P(i,j,d,θ) represent the probability of the pixel with gray value i and the pixel with gray value j appearing at the same time in the direction θ and the distance d; The eigenvalues ​​are extracted from the gray-level co-occurrence matrix, and the contrast is calculated as: Where L is the grayscale level. The grayscale level of the grayscale image is calculated by the bit depth. If its bit depth is ω, then the grayscale level is 2 ω ,The bit depth can be obtained by viewing the image properties; The correlation calculation formula is: where μ i and μ j are the means of grayscale values ​​i and j, σ i and σ j are the standard deviations of grayscale values ​​i and j, respectively; The energy calculation formula is: The formula for calculating homogeneity is: Assume that the multi-angle images of the surrounding rock structure of the same tunnel have a total of m image regions; The average contrast ratio is: Among them C $ is the contrast of the zth image, where z represents the image; The average correlation is: The average energy is: The mean value of homogeneity is: Will As standard texture features.

5. The machine vision-based tunnel surrounding rock structure deformation measurement system according to claim 4, characterized in that: The process of extracting texture feature values ​​based on the gray level co-occurrence matrix of the calculated image by the recognition unit is consistent with that of the learning unit.

6. The machine vision-based tunnel surrounding rock structure deformation measurement system according to claim 3, characterized in that: The comparison measurement unit calculates the difference between the texture feature value of the image z extracted by the current recognition unit and the reference texture feature value respectively; Contrast Difference: Among them C $ is the contrast of the current image region z; Correlation Difference: Energy Difference: Homogeneity Difference: Combining multiple texture feature differences to construct deformation index; Using the weighted summation method, the deformation index Y is: Y=αΔC+βΔX+γΔN+δΔT Among them, α, β, γ, and δ are weight coefficients, which can be adjusted according to actual conditions to reflect the importance of different texture features in deformation measurement.

7. The machine vision-based tunnel surrounding rock structure deformation measurement system according to claim 1, characterized in that: The cross-validation module cross-validates the results of comparing images at each angle of the same area, selects two image sets at different angles for comparison, and for each eigenvalue, calculates the absolute value of the difference between the eigenvalues ​​of the two angle sets to verify whether the recognition calculation results are consistent. If not, the visual recording module is notified to re-record the verification calculation.

8. The machine vision-based tunnel surrounding rock structure deformation measurement system according to claim 6, characterized in that: The early warning module determines the deformation degree of the tunnel surrounding rock structure according to the calculated value of the deformation index Y, and performs corresponding alarm levels by setting different threshold ranges, and the threshold range is confirmed by comparing historical records.

Images