A method for monitoring subgrade ground penetrating radar image contrast

Abstract

The application discloses a kind of roadbed ground penetrating radar image contrast monitoring methods, the method comprises the following steps: using ground penetrating radar to carry out road initial measurement;Synchronous recording of ground penetrating radar survey line position, establish block image filing index;Periodic rapid measurement is carried out using the same ground penetrating radar;According to survey line position retrieval corresponding filing image;Same position radar image is self-positioning calibration;Difference rate between periodic measurement image and initial measurement image is calculated;Depthwise offset between periodic measurement image and initial measurement image is calculated;Roadbed settlement risk assessment is carried out using difference rate and depthwise offset.The application realizes the accurate matching of road ground penetrating radar periodic measurement image and initial measurement image by positioning measurement initial calibration and radar image self-positioning calibration, and proposes image difference rate, depthwise offset, roadbed settlement risk value and other evaluation quantities, to provide an efficient method for large-scale road safety detection.

Description

Technical Field

[0001] This invention belongs to the field of geophysical exploration, specifically relating to a method for comparative monitoring of ground-penetrating radar images on a roadbed. Background Technology

[0002] Urban roads and highways are typical layered structures, consisting of an asphalt concrete surface layer, a crushed stone base (or concrete base), and an earthen subgrade from top to bottom. The internal media of each structural layer in a normal road are homogeneous, and the interlayer bonding is dense. Due to significant absorption and attenuation, the electromagnetic waves reflected from a normal road originate from the interfaces between the air and the surface layer, the surface layer and the base layer, and the base layer and the subgrade. These reflected waves are characterized by continuous phase axes, relatively uniform amplitude, and progressively weaker amplitude.

[0003] Ground-penetrating radar (GPR) is a device that uses electromagnetic waves to detect reflected signals from abnormal internal structures of non-metallic objects. It possesses high-resolution imaging capabilities and strong penetration performance, enabling non-destructive and rapid detection of electromagnetic wave signals reflected from roadbed anomalies such as cavities, cracks, and looseness. This technology plays a crucial role in road anomaly detection. For a long time, by identifying the polarity, intensity, and image shape of detected abnormal roadbed reflection signals and verifying them through drilling, a wealth of experience in GPR roadbed anomaly mapping has been accumulated. This experience has been used to select problematic areas from new radar images, providing important assurance for road safety. However, the internal conditions of roadbeds are extremely complex, and the aforementioned methods do not fully cover all anomalies, are relatively inefficient, and costly, making routine road surveys difficult.

[0004] On the other hand, roadbed anomalies such as cavities, cracks, and loosening are generally the result of long-term erosion accumulation, and the anomalous locations correspond to the areas with relatively strong reflections in ground-penetrating radar (GPR) images. By regularly monitoring changes in these anomalies (i.e., the areas with relatively strong reflections in the images), potential risks can be assessed or predicted, allowing for appropriate maintenance measures to ensure the safe use of the road. However, due to the relatively low efficiency of GPR measurements and positioning, no similar methods have yet been found that can be implemented in large-scale road surveys. Summary of the Invention

[0005] To address the shortcomings of the existing technologies, this invention provides a roadbed ground-penetrating radar image comparison and monitoring method. Through initial calibration of positioning measurement and self-positioning calibration of radar images, it achieves accurate matching between periodically measured images and initial measurement images of road ground-penetrating radar. Furthermore, it proposes evaluation metrics such as image difference rate, depth offset, and roadbed settlement risk value, providing an efficient method for large-scale road safety detection.

[0006] To achieve the above-mentioned objectives, the present invention provides the following technical solution:

[0007] A method for comparative monitoring of roadbed ground-penetrating radar images, the method comprising the following steps:

[0008] Step 1: Conduct initial road measurements using ground-penetrating radar: Record lane information of the road being measured during the measurement process, and simultaneously record the position of the ground-penetrating radar measurement line, keeping the lanes consistent in a single measurement;

[0009] Step 2: Based on the ground-penetrating radar survey line locations recorded synchronously in Step 1, establish a block image archiving index;

[0010] Step 3: Perform regular and rapid measurements using the same ground-penetrating radar as in Step 1: During the regular measurements, record the radar survey line position and lane information simultaneously, and after the data acquisition is completed, divide the regularly acquired radar images into blocks in the same way as in Step 2;

[0011] Step 4: Retrieve the corresponding archived image based on the survey line location: For the selected periodically measured radar image block, retrieve the archived image at the corresponding location based on its radar survey line location information using the query index table established in Step 2.

[0012] Step 5: Self-localization calibration of radar images at the same location: For the radar image blocks and archived images paired in Step 4, since there is an inevitable deviation between the two measurement and positioning, it is necessary to first perform overall translation calibration on the radar image blocks measured periodically, that is, set different overall offsets and calculate the correlation coefficient of the two images.

[0013] Step 6: Calculate the difference rate between periodically measured images and the initial measured images;

[0014] Step 7: Calculate the depth offset between the periodically measured images and the initial measured images;

[0015] Step 8: Use the difference rate and depth offset to conduct a roadbed settlement risk assessment.

[0016] Furthermore, in step 1, the location of the ground-penetrating radar survey line is recorded synchronously using a GPS module or odometer. When using an odometer, the starting point location of the survey line also needs to be recorded.

[0017] Furthermore, in step 2, the method for establishing the segmented image archiving index is as follows: for open, unobstructed roads, the radar images are segmented and archived using GPS latitude and longitude coordinates synchronously collected by the GPS module, denoted as I0(x i ,y i ,p),x i For the longitude information of the start or end point of this segmented radar image, y i For the corresponding latitude information, p represents the recorded lane number; then, the stored segmented radar image file name is matched with latitude, longitude, and lane position (x). i ,yi (p) Establish a query index table, and the corresponding radar image can be retrieved based on the location information in the future;

[0018] For roads with weak GPS signals, the radar image is divided into blocks using the relative mileage positions synchronously collected by the odometer, denoted as I0(S). i ,p),S i This provides relative mileage and location information for segmented radar images; and it also includes the stored segmented radar image filenames and their corresponding mileage and lane positions (S). i (p) Establish a query index table, and the corresponding radar image can be retrieved based on the location information in the future;

[0019] For different radar survey line location recording methods, the initial measurement archived block radar images will be uniformly recorded as I. 0,i,p The j-th echo is denoted as I. 0,i,p (j).

[0020] Furthermore, in step 3, when using an odometer to record the radar survey line position, the same starting position must be maintained, or the offset dS0 of the starting position relative to the initial measurement must be recorded; if the starting position is the same, dS0 can be recorded as 0; after data acquisition is completed, the periodically acquired radar images are divided into blocks as in step 2, and the position is recorded as I when using GPS to record the position. n (x i ,y i When using an odometer to record position, it is denoted as I. n (L i -dS0,p), where n is the number of periodic measurements, starting from 1; subsequently, it is uniformly denoted as I. n,i,p The j-th echo is denoted as I. n,i,p (j).

[0021] Further, in step 5, the selected periodically measured radar image block is denoted as I. n,i,p The archived image retrieved at the corresponding location is denoted as I. 0,i,p If the overall offset is set to Δj0, then the correlation coefficient between the periodically measured radar image blocks and the archived images at the corresponding locations is:

[0022]

[0023] Each echo is processed as a vector, I n,i,p (j+Δj0)·I 0,i,p (j), I n,i,p (j+Δj0)·I n,i,p (j+Δj0), I 0,i,p (j)·I 0,i,p (j) are all vector dot products; and record c n,i,pThe overall offset when (Δj0) reaches its maximum value is denoted as Δj. 0max Based on this, radar image I n Perform local translation calibration, that is, set different local offsets Δj for the echoes in columns j1 to j2 of the image, and calculate the local correlation coefficient between the two images:

[0024]

[0025]

[0026] In the formula, the local offset Δj is expressed as Δj 0max To reduce computation time, a range of values ​​is selected centered on c, and c is recorded. n,i,p The local offset when (j, Δj) reaches its maximum value is denoted as Δj. max (j), where Based on this, the overall correlation coefficient between the two images is calculated again:

[0027]

[0028] Further, in step 6, the method for calculating the difference rate between periodically measured images and the initial measured image is as follows:

[0029] First, denote the sequence number of the acquisition point in each echo column of the radar image as k, i.e., I n,i,p (j,k) or I 0,i,p (j,k) represents the value of the kth sampling point of the jth echo in the corresponding image. Then, for the k1 to k2th sampling points of the j1 to j2th echoes, calculate their correlation coefficient:

[0030]

[0031] And and The larger value is denoted as a. n,i,p (j,k) represents the amplitude of this portion of the image;

[0032] Then calculate its difference rate:

[0033]

[0034] Among them, H(a) n,i,p (j,k) and The transition function can be a step function, a linear function, or a quadratic curve function, where H(a) n,i,p (j,k) is used to reduce the deviation caused by small amplitude in local images. Used to exclude evaluation failures caused by low overall image similarity;

[0035] For the function H(a), setting its independent variable threshold a0, when taking the step function,

[0036]

[0037] When taking a linear transition function,

[0038]

[0039] When choosing the transition function of a quadratic curve,

[0040]

[0041] Further, in step 7, the method for calculating the depth offset between the periodically measured image and the initial measured image is as follows: for the k1 to k2 sampling points of the j1 to j2 column echoes of the image, different depth offsets Δk are selected, and the corresponding correlation coefficients are first calculated:

[0042]

[0043] And record c n,i,p The depth offset when (j,k,Δk) reaches its maximum value is denoted as Δk. max (j,k); and will and The larger value is denoted as b. n,i,p (j,k), then calculate the depth offset between the periodically measured image and the initial measured image:

[0044] δ n,i,p (j,k)=Δk max (j,k)×H(b n,i,p (j,k))×H(c n,i,p (j,k,Δk max )),

[0045] In the formula, as in step 6, H(b) n,i,p (j,k)) or H(c) n,i,p (j,k,Δk max The transition function can be a step function, a linear function, or a quadratic curve function, where H(b) is the transition function. n,i,p (j,k)) is used to reduce the deviation caused by small local image amplitude, H(c n,i,p (j,k,Δk max This is used to exclude failures caused by low correlation coefficients in local images.

[0046] Furthermore, in step 8, the method for assessing roadbed settlement risk using difference rate and depth offset includes the following steps:

[0047] Step 81: Set a weighting value α0 for the image difference rate;

[0048] Step 82: To eliminate the effects of changes in road dielectric constant and overall radar system drift, the entire image δ can be subtracted first. n,i,p The mean of (j,k) is then used to classify and weight the depth offset of the image column by column: when the entire column δ n,i,p When the value tends towards a positive value, it can be marked as Category 1, indicating that the road surface has local depressions relative to the initial measurement, and the weighting value is set as α1; when the entire δ n,i,p When the value tends towards negative, it can be marked as Class 2, indicating that there is a local bulge in the road surface relative to the initial measurement, and the weighting value is set as α2; when δ near a certain depth n,i,p When δ approaches a negative value, the δ values ​​at other depths n,i,p When it approaches zero, it can be marked as Category 3, indicating that the roadbed is loose or voiding is developing towards the pavement, and the weighting value is set to α3;

[0049] Step 83: After classification and weighting, the formula for calculating the roadbed settlement risk based on ground-penetrating radar image comparison monitoring is as follows:

[0050]

[0051] In the formula, α q (q = 1, 2, 3, ...) represents different classification weights; For the entire image δ n,i,p The mean of (j,k); δ0 is the set depth offset reference value. When the risk value is greater than 0.7, it indicates a high risk. When the risk value is less than 0.3, it indicates a low risk.

[0052] Based on the above technical solution, the present invention has the following advantages compared with the prior art:

[0053] 1. A method for large-scale roadbed anomaly comparison and monitoring based on changes in the strong reflective portion of ground-penetrating radar images is proposed. This method does not require prior experience in accumulating anomaly radar image spectra and can significantly improve the efficiency and coverage of roadbed safety detection. 2. A ground-penetrating radar image matching method based on a combination of positioning measurement and self-positioning calibration is proposed, which reduces the requirements for high-precision positioning measurement of ground-penetrating radar.

[0054] 3. A method for calculating ground-penetrating radar image difference rate, depth offset, and roadbed settlement risk is proposed, which can provide a reliable quantitative approach for roadbed safety assessment. Attached Figure Description

[0055] Figure 1 A flowchart of a method for comparative monitoring of roadbed ground-penetrating radar images.

[0056] Figure 2The initial measurement archived image of a road at a certain location in Example 1.

[0057] Figure 3 The image shows a road at the same location measured periodically in Example 1.

[0058] Figure 4 The local offset curve of self-positioning calibration in Example 1.

[0059] Figure 5 The difference rate distribution between the periodic and initial measurement images in Example 1.

[0060] Figure 6 The depth offset distribution map between the periodic and initial measurement images in Example 1.

[0061] Figure 7 The distribution map of roadbed settlement risk evaluated in Example 1.

[0062] Figure 8 The initial measurement archived image of a road at a certain location in Example 2.

[0063] Figure 9 The image shows a road at the same location measured periodically in Example 2.

[0064] Figure 10 The distribution map of roadbed settlement risk evaluated in Example 2. Detailed Implementation

[0065] To make the objectives, technical solutions, and advantages of the present invention clearer, specific embodiments are shown below to describe the present invention. However, it should be understood that these descriptions are merely exemplary and not intended to limit the scope of the present invention. Furthermore, in the following description, descriptions of well-known structures and technologies are omitted to avoid unnecessarily obscuring the concept of the present invention.

[0066] The terminology used in this disclosure is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. The singular forms “a,” “the,” and “the” as used in this disclosure and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any and all possible combinations of one or more of the associated listed items.

[0067] It should be understood that although the terms first, second, third, etc., may be used in this disclosure to describe various information, such information should not be limited to these terms and should not be construed as indicating or implying relative importance. These terms are used only to distinguish information of the same type from one another. For example, without departing from the scope of this disclosure, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to determination."

[0068] In the description of this invention, unless otherwise specified and limited, it should be noted that the terms "installation", "connection" and "linking" should be interpreted broadly. For example, they can refer to mechanical connection or internal connection between two components. They can be direct connection or indirect connection through an intermediate medium. Those skilled in the art can understand the specific meaning of the above terms according to the specific circumstances.

[0069] To better understand the technical solution of the present invention, the following detailed description is provided in conjunction with specific embodiments. The embodiments described below are only for explaining the present invention and are not intended to limit the present invention. Specific experimental methods not mentioned in the embodiments are generally performed according to conventional experimental methods. Unless otherwise specified, all equipment used is commercially available conventional product.

[0070] like Figure 1 As shown, a method for comparative monitoring of roadbed ground-penetrating radar images includes the following steps:

[0071] Step 1: Conduct initial road measurements using ground-penetrating radar: Record lane information of the road being measured during the measurement process, and simultaneously record the position of the ground-penetrating radar measurement line, keeping the lanes consistent in a single measurement;

[0072] Step 2: Based on the ground-penetrating radar survey line locations recorded synchronously in Step 1, establish a block image archiving index;

[0073] Step 3: Perform regular and rapid measurements using the same ground-penetrating radar as in Step 1: During the regular measurements, record the radar survey line position and lane information simultaneously, and after the data acquisition is completed, divide the regularly acquired radar images into blocks in the same way as in Step 2;

[0074] Step 4: Retrieve the corresponding archived image based on the survey line location: For the selected periodically measured radar image block, retrieve the archived image at the corresponding location based on its radar survey line location information using the query index table established in Step 2.

[0075] Step 5: Self-localization calibration of radar images at the same location: For the radar image blocks and archived images paired in Step 4, since there is an inevitable deviation between the two measurement and positioning, it is necessary to first perform overall translation calibration on the radar image blocks measured periodically, that is, set different overall offsets and calculate the correlation coefficient of the two images.

[0076] Step 6: Calculate the difference rate between periodically measured images and the initial measured images;

[0077] Step 7: Calculate the depth offset between the periodically measured images and the initial measured images;

[0078] Step 8: Use the difference rate and depth offset to conduct a roadbed settlement risk assessment.

[0079] Furthermore, in step 1, the location of the ground-penetrating radar survey line is recorded synchronously using a GPS module or odometer. When using an odometer, the starting point location of the survey line also needs to be recorded.

[0080] Furthermore, in step 2, the method for establishing the segmented image archiving index is as follows: for open, unobstructed roads, the radar images are segmented and archived using GPS latitude and longitude coordinates synchronously collected by the GPS module, denoted as I0(x i ,y i ,p),x i For the longitude information of the start or end point of this segmented radar image, y i For the corresponding latitude information, p represents the recorded lane number; then, the stored segmented radar image file name is matched with latitude, longitude, and lane position (x). i ,y i (p) Establish a query index table, and the corresponding radar image can be retrieved based on the location information in the future;

[0081] For roads with weak GPS signals, the radar image is divided into blocks using the relative mileage positions synchronously collected by the odometer, denoted as I0(S). i ,p),S i This provides relative mileage and location information for segmented radar images; and it also includes the stored segmented radar image filenames and their corresponding mileage and lane positions (S). i (p) Establish a query index table, and the corresponding radar image can be retrieved based on the location information in the future;

[0082] For different radar survey line location recording methods, the initial measurement archived block radar images will be uniformly recorded as I. 0,i,p The j-th echo is denoted as I. 0,i,p (j).

[0083] Furthermore, in step 3, when using an odometer to record the radar survey line position, the same starting position must be maintained, or the offset dS0 of the starting position relative to the initial measurement must be recorded; if the starting position is the same, dS0 can be recorded as 0; after data acquisition is completed, the periodically acquired radar images are divided into blocks as in step 2, and the position is recorded as I when using GPS to record the position. n (x i ,y i When using an odometer to record position, it is denoted as I. n (L i -dS0,p), where n is the number of periodic measurements, starting from 1; subsequently, it is uniformly denoted as I. n,i,p The j-th echo is denoted as I. n,i,p (j).

[0084] Further, in step 5, the selected periodically measured radar image block is denoted as I. n,i,p The archived image retrieved at the corresponding location is denoted as I. 0,i,p If the overall offset is set to Δj0, then the correlation coefficient between the periodically measured radar image blocks and the archived images at the corresponding locations is:

[0085]

[0086] Each echo is processed as a vector, I n,i,p (j+Δj0)·I 0,i,p (j), I n,i,p (j+Δj0)·I n,i,p (j+Δj0), I 0,i,p (j)·I 0,i,p (j) are all vector dot products; and record c n,i,p The overall offset when (Δj0) reaches its maximum value is denoted as Δj. 0max ;

[0087] Based on this, radar image I n Perform local translation calibration, that is, set different local offsets Δj for the echoes in columns j1 to j2 of the image, and calculate the local correlation coefficient between the two images:

[0088]

[0089] In the formula, the local offset Δj is expressed as Δj 0max To reduce computation time, a range of values ​​is selected centered on c, and c is recorded. n,i,p The local offset when (j, Δj) reaches its maximum value is denoted as Δj. max (j), where

[0090] Based on this, the overall correlation coefficient between the two images is calculated again:

[0091]

[0092] Further, in step 6, the method for calculating the difference rate between periodically measured images and the initial measured image is as follows:

[0093] First, denote the sequence number of the acquisition point in each echo column of the radar image as k, i.e., I n,i,p (j,k) or I 0,i,p (j,k) represents the value of the kth sampling point of the jth echo in the corresponding image. Then, for the k1 to k2th sampling points of the j1 to j2th echoes, calculate their correlation coefficient:

[0094]

[0095] And and The larger value is denoted as a. n,i,p (j,k) represents the amplitude of this portion of the image;

[0096] Then calculate its difference rate:

[0097]

[0098] Among them, H(a) n,i,p (j,k) and The transition function can be a step function, a linear function, or a quadratic curve function, where H(a) n,i,p (j,k) is used to reduce the deviation caused by small amplitude in local images. Used to exclude evaluation failures caused by low overall image similarity;

[0099] For the function H(a), setting its independent variable threshold a0, when taking the step function,

[0100]

[0101] When taking a linear transition function,

[0102]

[0103] When choosing the transition function of a quadratic curve,

[0104]

[0105] Further, in step 7, the method for calculating the depth offset between the periodically measured image and the initial measured image is as follows: for the k1 to k2 sampling points of the j1 to j2 column echoes of the image, different depth offsets Δk are selected, and the corresponding correlation coefficients are first calculated:

[0106]

[0107] And record c n,i,p The depth offset when (j,k,Δk) reaches its maximum value is denoted as Δk. max (j,k); and will and The larger value is denoted as b. n,i,p (j,k), then calculate the depth offset between the periodically measured image and the initial measured image:

[0108] δ n,i,p (j,k)=Δk max (j,k)×H(b n,i,p (j,k))×H(c n,i,p (j,k,Δk max )),

[0109] In the formula, as in step 6, H(b) n,i,p (j,k)) or H(c) n,i,p (j,k,Δk max The transition function can be a step function, a linear function, or a quadratic curve function, where H(b) is the transition function. n,i,p (j,k)) is used to reduce the deviation caused by small local image amplitude, H(c n,i,p (j,k,Δk max This is used to exclude failures caused by low correlation coefficients in local images.

[0110] Furthermore, in step 8, the method for assessing roadbed settlement risk using difference rate and depth offset includes the following steps:

[0111] Step 81: Set a weighting value α0 for the image difference rate;

[0112] Step 82: To eliminate the effects of changes in road dielectric constant and overall radar system drift, the entire image δ can be subtracted first. n,i,p The mean of (j,k) is then used to classify and weight the depth offset of the image column by column: when the entire column δ n,i,p When the value tends towards a positive value, it can be marked as Category 1, indicating that the road surface has local depressions relative to the initial measurement, and the weighting value is set as α1; when the entire δ n,i,p When the value tends towards negative, it can be marked as Class 2, indicating that there is a local bulge in the road surface relative to the initial measurement, and the weighting value is set as α2; when δ near a certain depth n,i,p When δ approaches a negative value, the δ values ​​at other depths n,i,p When it approaches zero, it can be marked as Category 3, indicating that the roadbed is loose or voiding is developing towards the pavement, and the weighting value is set to α3;

[0113] Step 83: After classification and weighting, the formula for calculating the roadbed settlement risk based on ground-penetrating radar image comparison monitoring is as follows:

[0114]

[0115] In the formula, αp (q = 1, 2, 3, ...) represents different classification weights; For the entire image δn ,i,p The mean of (j,k); δ0 is the set depth offset reference value. When the risk value is greater than 0.7, it indicates a high risk. When the risk value is less than 0.3, it indicates a low risk.

[0116] Example 1

[0117] A certain air-coupled ground-penetrating radar was used to conduct roadbed perspective measurement of the highway, and GPS was used for measurement and positioning. The radar detection center frequency was about 300MHz, the air-coupled detection distance from the bottom of the radar antenna to the road surface was about 15cm, the distance between the centers of the transmitting antenna and the receiving antenna was 65cm, it had a large bistatic angle, the total survey line length exceeded 100km, and the measurement speed was about 90km / h.

[0118] The measured ground-penetrating radar images were divided into blocks of 0.001° according to GPS latitude and longitude, with a 0.1m interval between the survey lines in each data column. The initial ground-penetrating radar image with the endpoint at 121.2850°E and 31.0518°N is shown below. Figure 2 As shown, the radar image measured again one month later is as follows. Figure 3 As shown, the overall horizontal offset Δj of the two images 0max =-7, local offset Δj max (j) such as Figure 4 As shown, the overall correlation coefficient of the two images after matching is 93%.

[0119] The calculated distribution of the difference rate between periodic and initial measurement images is shown in the figure below. Figure 5 As shown; the calculated depth offset distribution between the periodic and initial measurement images is shown in the figure. Figure 6 As shown.

[0120] Set the settlement risk assessment weighting values ​​as follows: α0 = 0.4, α1 = 0.15, α2 = 0.05, α3 = 0.4; and take the depth offset reference value as δ0 = 10; the distribution of roadbed settlement risk according to this weighted assessment is as follows. Figure 7 As shown, its maximum is approximately 0.12, which is considered a relatively low-risk situation.

[0121] Example 2

[0122] A certain air-coupled ground-penetrating radar was used to conduct roadbed perspective measurement of the highway, and GPS was used for measurement and positioning. The radar detection center frequency was about 350MHz, the air-coupled detection distance from the bottom of the radar antenna to the road surface was about 15cm, the distance between the centers of the transmitting antenna and the receiving antenna was 13cm, the total measurement line length exceeded 100km, and the measurement speed was about 90km / h.

[0123] The measured ground-penetrating radar images were divided into blocks of 0.001° according to GPS latitude and longitude, with each data column having a survey line length interval of 0.1m. The initial ground-penetrating radar image with the endpoint at 121.2830°E and 31.0505°N is shown below. Figure 8 As shown, the radar image measured again one month later is as follows. Figure 9 As shown, the overall horizontal offset Δj of the two is... 0max =-1, the overall correlation coefficient after matching the two images is 94%.

[0124] Set the settlement risk assessment weighting values ​​as follows: α0 = 0.4, α1 = 0.15, α2 = 0.05, α3 = 0.4; and take the depth offset reference value as δ0 = 10; the distribution of roadbed settlement risk according to this weighted assessment is as follows. Figure 10 As shown, its maximum is approximately 0.13, which is considered a relatively low-risk situation.

[0125] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them; although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications can still be made to the specific implementation of the invention or equivalent substitutions can be made to some technical features without departing from the spirit of the technical solutions of the present invention, and all such modifications and substitutions should be covered within the scope of the technical solutions claimed in the present invention.

Claims

1. A method for comparative monitoring of roadbed ground-penetrating radar images, characterized in that, The method includes the following steps: Step 1: Conduct initial road measurements using ground-penetrating radar: Record lane information of the road being measured during the measurement process, and simultaneously record the position of the ground-penetrating radar measurement line, keeping the lanes consistent in a single measurement; Step 2: Based on the ground-penetrating radar survey line locations recorded synchronously in Step 1, establish a block image archiving index; Step 3: Perform regular and rapid measurements using the same ground-penetrating radar as in Step 1: During the regular measurements, record the radar survey line position and lane information simultaneously, and after the data acquisition is completed, divide the regularly acquired radar images into blocks in the same way as in Step 2; Step 4: Retrieve the corresponding archived image based on the survey line location: For the selected periodically measured radar image block, retrieve the archived image at the corresponding location based on its radar survey line location information using the query index table established in Step 2. Step 5: Self-localization calibration of radar images at the same location: For the radar image blocks and archived images paired in Step 4, since there is an inevitable deviation between the two measurement and positioning, it is necessary to first perform overall translation calibration on the radar image blocks measured periodically, that is, set different overall offsets and calculate the correlation coefficient of the two images. Step 6: Calculate the difference rate between periodically measured images and the initial measured images; Step 7: Calculate the depth offset between the periodically measured images and the initial measured images; Step 8: Use the difference rate and depth offset to conduct a roadbed settlement risk assessment.

2. The method for roadbed ground-penetrating radar image comparison and monitoring according to claim 1, characterized in that, In step 1, the location of the ground-penetrating radar survey line is recorded synchronously using a GPS module or odometer. When using an odometer, the starting point location of the survey line also needs to be recorded.

3. The method for roadbed ground-penetrating radar image comparison and monitoring according to claim 2, characterized in that, In step 2, the method for establishing the segmented image archiving index is as follows: for open, unobstructed roads, the radar images are segmented and archived using GPS latitude and longitude coordinates synchronously collected by the GPS module, denoted as I0(x i ,y i ,p),x i For the longitude information of the start or end point of this segmented radar image, y i For the corresponding latitude information, p represents the recorded lane number; then, the stored segmented radar image file name is matched with latitude, longitude, and lane position (x). i ,y i (p) Create a query index table, and then query the corresponding radar image based on the location information; For roads with weak GPS signals, the radar image is divided into blocks using the relative mileage positions synchronously collected by the odometer, denoted as I0(S). i ,p),S i This provides relative mileage and location information for segmented radar images; And the stored segmented radar image file names are associated with mileage and lane position (S i (p) Create a query index table, and then query the corresponding radar image based on the location information; For different radar survey line location recording methods, the initial measurement archived segmented radar images will be uniformly recorded as follows: The j-th echo is denoted as .

4. The method for roadbed ground-penetrating radar image comparison and monitoring according to claim 3, characterized in that, In step 3, when using an odometer to record the radar survey line position, the same starting position must be maintained, or the offset dS0 of the starting position relative to the initial measurement must be recorded; if the starting positions are the same, dS0 is recorded as 0; after data acquisition is completed, the periodically acquired radar images are divided into blocks as in step 2, and the position is recorded as I when using GPS to record the position. n (x i ,y i When using an odometer to record position, it is denoted as I. n (L i -dS0,p), where n is the number of periodic measurements, starting from 1; subsequently, it will be uniformly denoted as... The j-th echo is denoted as .

5. The method for roadbed ground-penetrating radar image comparison and monitoring according to claim 4, characterized in that, In step 5, the selected periodically measured radar image blocks are denoted as... The archived image at the corresponding location is retrieved and denoted as... The overall offset is set to The correlation coefficient between periodically measured radar image patches and archived images at corresponding locations is: ， Each echo is processed as a vector. , , All are vector dot products; and are recorded. The overall offset when the maximum value is taken is denoted as ; Based on this, radar image I n Perform local translation calibration, that is, perform local translation calibration on the first part of the image. to Column echo, set different local offsets Calculate the local correlation coefficient between the two images: ， ， ， In the formula, the local offset by To reduce computation time, a range of values ​​is selected around the center, and the results are recorded. The local offset when the maximum value is obtained is denoted as ,in ; Based on this, the overall correlation coefficient between the two images is calculated again: 。 6. The method for roadbed ground-penetrating radar image comparison and monitoring according to claim 5, characterized in that, In step 6, the method for calculating the difference rate between periodically measured images and the initial measured images is as follows: First, record the serial number of the acquisition point in each echo column of the radar image as... k ,Right now or Let represent the value of the k-th sampling point of the j-th column echo in the corresponding image. Then, for the... to The first echo to For each sampling point, calculate its correlation coefficient: ， ， ， And and The larger value is denoted as , indicating the amplitude of that portion of the image; Then calculate its difference rate: ， in, and Choose a step function, a linear or quadratic transition function, where Used to reduce deviations caused by small amplitude in local images. Used to exclude evaluation failures caused by low overall image similarity; For functions Set its independent variable threshold. When taking the step function, ， When taking a linear transition function, ， When choosing the transition function of a quadratic curve, 。 7. The method for roadbed ground-penetrating radar image comparison and monitoring according to claim 6, characterized in that, In step 7, the method for calculating the depth offset between the periodically measured image and the initial measured image is as follows: For the image... to The first echo to Each sampling point is selected with different depth offsets. First, calculate the corresponding correlation coefficient: ， ， And record The depth offset at the maximum value is denoted as . ; and will and The larger value is denoted as Then calculate the depth offset between the periodically measured images and the initial measured images: ， In the formula, the same as step 6, or Choose a step function, a linear or quadratic transition function, where Used to reduce deviations caused by small amplitude in local images. This is used to eliminate failures caused by low correlation coefficients in local images.

8. The method for roadbed ground-penetrating radar image comparison and monitoring according to claim 7, characterized in that, In step 8, the method for assessing roadbed settlement risk using difference rate and depth offset includes the following steps: Step 81: Set weighting values ​​for image difference rates. ; Step 82: To eliminate the effects of changes in road dielectric constant and overall radar system drift, the entire image is first subtracted. The mean is then used to classify and weight the depth offset of the image column by column: when the entire column... When the value approaches positive, it is marked as Category 1, indicating that the road surface has local depressions relative to the initial measurement, and the weighting value is set to... ;When the entire column When the value tends towards negative, it is marked as Category 2, indicating that there is a local bulge in the road surface relative to the initial measurement, and the weighting value is set to... When a certain depth is near When it tends to be negative, other depths When the value approaches zero, it is marked as Category 3, indicating that the roadbed is becoming loose or voiding and is developing towards the pavement. The weighted value is set to... ; Step 83: After classification and weighting, the formula for calculating the roadbed settlement risk based on ground-penetrating radar image comparison monitoring is as follows: ， In the formula, , This represents the weighted values ​​for different categories; For the whole image The mean; The set depth offset reference value indicates a high risk when the risk value is greater than 0.7 and a low risk when the risk value is less than 0.3.

Images