A space-air-ground integrated underground facility detection method and system

Abstract

The application discloses a kind of underground facilities detection method and system based on space-air-ground integration, belong to facility risk detection technical field, and the ground surface abnormal deformation area of target detection area is obtained by SAR image analysis, and the reflection wave data and abnormal crack data are detected respectively by ground penetrating radar and unmanned aerial vehicle for the ground surface abnormal deformation area, avoid blind detection, improve efficiency, combine the deformation data of region Comprehensive evaluation underground facility damage risk of corresponding ground surface abnormal deformation area, realize space-air-ground multi-source data analysis, improve risk assessment accuracy, simultaneously in SAR image registration, the phase consistency value, gradient value and corresponding regional dispersion value of each pixel point are combined to carry out triple key point screening, improve registration accuracy and then improve the accuracy of analyzing ground surface abnormal deformation area, to improve the precision of underground facility key detection range.

Description

Technical Field

[0001] This invention relates to the field of facility risk detection technology, and in particular to a method and system for detecting underground facilities based on an integrated air-space-ground system. Background Technology

[0002] Underground facilities are closely related to human life today, such as various pipeline networks. The damage to underground facilities has a great impact on the quality of life. Therefore, it is particularly important to conduct damage risk detection of underground facilities in order to detect damage in time or make early predictions.

[0003] However, since underground facilities are buried underground, it is difficult and troublesome to directly inspect the facilities themselves. Therefore, most of the time, the risk of damage to underground facilities is assessed by inspecting the environment in which the facilities are located. In the current technology, ground-penetrating radar is often used to detect reflected waves for analysis. However, a single detection method can only obtain a limited data type of the environment in which the facilities are located, which makes the assessment of the risk of damage to underground facilities inaccurate. For example, abnormal deformation of the ground surface can also affect the damage to underground facilities.

[0004] Therefore, improving the accuracy of damage risk detection for underground facilities has become an urgent problem to be solved. Summary of the Invention

[0005] To address the technical problems existing in the prior art, this invention provides a method for detecting underground facilities based on an integrated air-space-ground system, comprising the following steps:

[0006] The latest SAR image and the initial SAR image of the target detection area are acquired. The latest SAR image and the initial SAR image are registered to obtain a registered image. Then, the surface anomaly deformation area and the corresponding deformation data of the target detection area are obtained by analyzing the registered image using InSAR technology.

[0007] Ground-penetrating radar is used to acquire reflected wave data of abnormal surface deformation areas; UAVs are used to perform image detection and identification of abnormal crack data in abnormal surface deformation areas.

[0008] The risk of damage to underground facilities in areas with abnormal surface deformation is assessed based on deformation data, reflected wave data, and abnormal crack data.

[0009] The process of registering the latest SAR image and the initial SAR image to obtain the registered image involves the following steps:

[0010] The first set of key features for the latest SAR image and the initial SAR image is obtained by using the phase consistency value of each pixel; the second set of key features for the latest SAR image and the initial SAR image is obtained by using the gradient value of each pixel; and the third set of key features for the latest SAR image and the initial SAR image is obtained by using the region discrete value corresponding to each pixel.

[0011] The first intersection is obtained by intersecting the first set of key features, the second set of key features, and the third set of key features of the latest SAR image. The key features in the first intersection are the first true key features of the latest SAR image. The second intersection is obtained by intersecting the first set of key features, the second set of key features, and the third set of key features of the initial SAR image. The key features in the second intersection are the second true key features of the initial SAR image.

[0012] Multiple matching point pairs are obtained by matching each first real key point with the second real key point. The transformation matrix is ​​obtained by using the matching point pairs. The latest SAR image and the initial SAR image are registered according to the transformation matrix to obtain the registered image.

[0013] Furthermore, the step of obtaining the first set of key feature points of the latest SAR image and the initial SAR image by using the phase consistency value of each pixel is specifically accomplished through the following steps:

[0014] By convolving a two-dimensional Log-Gabor filter with the corresponding SAR image, the frequency domain amplitude of each pixel at each scale n and each direction o is obtained. and phase components , where i represents the i-th pixel;

[0015] Calculate the phase consistency value of each pixel in different directions based on the frequency domain amplitude and phase components:

[0016] The minimum moment value of each pixel is obtained through moment analysis equation based on the phase consistency value:

[0017] Pixels with minimum moment values ​​greater than or equal to a preset threshold are selected as the first feature keypoints, thus forming the first feature keypoint set of the corresponding SAR image.

[0018] Furthermore, the step of obtaining the second feature keypoint set of the latest SAR image and the initial SAR image by using the gradient values ​​of each pixel is specifically accomplished through the following steps:

[0019] Calculate the gradient value of each pixel:

[0020] ;

[0021] in, This corresponds to the gradient value of the i-th pixel in the SAR image. This represents the gradient in the horizontal direction corresponding to the i-th pixel in the SAR image. This represents the gradient in the vertical direction corresponding to the i-th pixel in the SAR image. This represents the spatial derivative of the Gaussian function in the horizontal direction for the i-th pixel in the SAR image. is the spatial derivative of the i-th pixel in the corresponding SAR image in the vertical direction through the Gaussian function, * is the convolution operation, and S is the corresponding SAR image;

[0022] Pixels with gradient values ​​greater than or equal to a preset gradient threshold are used as second feature keypoints, thus forming the second feature keypoint set of the corresponding SAR image.

[0023] Furthermore, the step of obtaining the third feature keypoint set of the latest SAR image and the initial SAR image respectively through the region discrete values ​​corresponding to each pixel point is specifically as follows:

[0024] Based on the preset window size, blocks centered on each pixel will be obtained separately;

[0025] Calculate the whole image discrete value and the regional discrete value of each block of the corresponding SAR image. The whole image discrete value is the ratio of the standard deviation of the pixel value to the mean of the pixel value of the corresponding SAR image, and the regional discrete value is the ratio of the standard deviation of the pixel value to the mean of the pixel value of the corresponding block.

[0026] If the regional discrete value of a block is greater than the overall discrete value of the corresponding SAR image, then the center pixel of the block is taken as the third feature key point, thus forming the third feature key point set of the corresponding SAR image.

[0027] Furthermore, the step of matching each first real keypoint with the second real keypoint to obtain multiple matching point pairs specifically involves:

[0028] Construct feature descriptors for each first and second real keypoint;

[0029] Calculate the Euclidean distance between the feature descriptor of the first real keypoint and the feature descriptor of each second real keypoint, and select the second real keypoint with the smallest Euclidean distance to match the first real keypoint as the first matching point pair;

[0030] The RANSAC algorithm is used to remove mismatched point pairs from the first matching point pair, leaving the second matching point pair.

[0031] Furthermore, the process of obtaining the transformation matrix using matching point pairs is specifically achieved through:

[0032] Based on the analysis of the second matching point pair, the geometric distortion value G is: ; r is the number of the second matching point pairs, () represents the coordinates of the first real keypoint in the i-th second matching point pair. ) represents the coordinates of the second real keypoint in the i-th second matching point pair;

[0033] Based on the analysis of geometric distortion values, the compensation coefficient BC is: Gmax is the preset maximum threshold for geometric distortion.

[0034] The transformation matrix parameters are obtained by combining the compensation coefficients and solving the least squares objective function. The expression of the least squares objective function is as follows: ; To minimize the objective function value, , , , and All are transformation matrix parameters;

[0035] The transformation matrix is ​​constructed by transforming the transformation matrix parameters.

[0036] Furthermore, the step of using a drone to perform image detection and identification of abnormal crack data in areas of abnormal surface deformation includes the determination of abnormal cracks, the determination steps of which are as follows:

[0037] Multiple sub-images of corresponding abnormal surface deformation areas were obtained using drones;

[0038] The edge detection algorithm is used to obtain suspected cracks in the sub-image of the region, and then a pre-built deep learning model is used to identify the first crack from the suspected cracks;

[0039] The regional sub-images are stitched together to obtain a complete regional image. Based on the endpoint position of the first crack, the first crack connected to the endpoint of the current first crack is queried in the complete regional image and continuously stitched together to obtain the corresponding complete crack.

[0040] Obtain the actual crack width and its pixel percentage in the corresponding complete area image. If the actual crack width is greater than or equal to a preset width threshold and / or the pixel percentage is greater than or equal to a preset percentage threshold, then it is an abnormal crack.

[0041] Furthermore, the assessment of the risk of damage to underground facilities in areas of abnormal surface deformation based on deformation data, reflected wave data, and abnormal crack data specifically includes:

[0042] Obtain the pre-set BIM model of underground facilities in the target area, and obtain the facility parameters of underground facilities in the corresponding abnormal surface deformation area based on the BIM model;

[0043] The first risk coefficient F1 is estimated based on the deformation data and facility parameters using a pre-set first convolutional neural network.

[0044] The second risk coefficient F2 is estimated based on the reflected wave data and facility parameters using a pre-set second convolutional neural network.

[0045] The third risk coefficient F3 is estimated based on abnormal crack data and facility parameters using a pre-set third convolutional neural network.

[0046] Estimate the comprehensive damage risk value F of underground facilities in the corresponding abnormal surface deformation area, F=K1×F1+K2×F2+K3×F3, where K1, K2 and K3 are the first preset weight, the second preset weight and the third preset weight, respectively.

[0047] This invention also provides an underground facility detection system based on an integrated air-space-ground system, which, when applied to any of the aforementioned underground facility detection methods based on an integrated air-space-ground system, includes:

[0048] The first data acquisition module acquires the latest SAR image and the initial SAR image of the target detection area, registers the latest SAR image and the initial SAR image to obtain a registered image, and then uses InSAR technology to analyze the registered image to obtain the surface anomaly deformation area of ​​the target detection area and the corresponding deformation data.

[0049] The process of registering the latest SAR image and the initial SAR image to obtain the registered image involves the following steps:

[0050] The first set of key features for the latest SAR image and the initial SAR image is obtained by using the phase consistency value of each pixel; the second set of key features for the latest SAR image and the initial SAR image is obtained by using the gradient value of each pixel; and the third set of key features for the latest SAR image and the initial SAR image is obtained by using the region discrete value corresponding to each pixel.

[0051] The first intersection is obtained by intersecting the first set of key features, the second set of key features, and the third set of key features of the latest SAR image. The key features in the first intersection are the first true key features of the latest SAR image. The second intersection is obtained by intersecting the first set of key features, the second set of key features, and the third set of key features of the initial SAR image. The key features in the second intersection are the second true key features of the initial SAR image.

[0052] Multiple matching point pairs are obtained by matching each first real key point with the second real key point. The transformation matrix is ​​obtained by using the matching point pairs. The latest SAR image and the initial SAR image are registered according to the transformation matrix to obtain the registered image.

[0053] The second data acquisition module acquires reflected wave data from areas of abnormal surface deformation using ground penetrating radar.

[0054] The third data acquisition module uses drones to perform image detection and identification of abnormal crack data in abnormally deformed areas of the ground surface.

[0055] The risk assessment module evaluates the risk of damage to underground facilities in areas with abnormal surface deformation based on deformation data, reflected wave data, and abnormal crack data.

[0056] Furthermore, the step of matching each first real keypoint with the second real keypoint to obtain multiple matching point pairs specifically involves:

[0057] Construct feature descriptors for each first and second real keypoint;

[0058] Calculate the Euclidean distance between the feature descriptor of the first real keypoint and the feature descriptor of each second real keypoint, and select the second real keypoint with the smallest Euclidean distance to match the first real keypoint as the first matching point pair;

[0059] The RANSAC algorithm is used to remove mismatched point pairs from the first matching point pair, leaving the second matching point pair.

[0060] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0061] This invention identifies anomalous surface deformation areas within the target detection region through SAR image analysis. For these areas, ground-penetrating radar and unmanned aerial vehicles (UAVs) are used to detect reflected wave data and anomalous crack data, respectively, avoiding blind detection and improving efficiency. The deformation data of the region is combined to comprehensively assess the risk of damage to underground facilities in the corresponding anomalous surface deformation areas, achieving multi-source data analysis from air, space, and ground to improve the accuracy of risk assessment. Simultaneously, in SAR image registration, the phase consistency value, gradient value, and corresponding regional discrete value of each pixel are combined for triple key point screening, improving registration accuracy and thus enhancing the accuracy of analyzing anomalous surface deformation areas, thereby increasing the precision of the key detection range for underground facilities. Attached Figure Description

[0062] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0063] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0064] Figure 1 This is a flowchart of an underground facility detection method based on the integrated air-space-ground system of the present invention;

[0065] Figure 2 This is a flowchart of step S1 of the method of the present invention, which involves registering the latest SAR image and the initial SAR image to obtain a registered image.

[0066] Figure 3 This is a structural block diagram of an underground facility detection system based on the integration of air, space, and ground systems according to the present invention. Detailed Implementation

[0067] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0068] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indication will also change accordingly.

[0069] Furthermore, the use of terms such as "first" and "second" in this invention is for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, features defined with "first" and "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions of the various embodiments can be combined with each other, but only on the basis of being achievable by those skilled in the art. When the combination of technical solutions is contradictory or impossible to implement, such a combination of technical solutions should be considered non-existent and not within the scope of protection claimed by this invention.

[0070] Example 1

[0071] See Figure 1 and Figure 2 As shown, the present invention provides a method for detecting underground facilities based on an integrated air-space-ground system, which specifically includes the following steps:

[0072] S1. Acquire the latest SAR image and the initial SAR image of the target detection area, register the latest SAR image and the initial SAR image to obtain the registered image, and then use InSAR technology to analyze the registered image to obtain the surface anomaly deformation area of ​​the target detection area and the corresponding deformation data.

[0073] SAR imagery is a high-resolution image generated by transmitting electromagnetic waves and receiving reflected signals from ground objects using SAR (Synthetic Aperture Radar). InSAR (Interferometric Synthetic Aperture Radar) is a remote sensing measurement technique based on synthetic aperture radar that obtains surface elevation or deformation data by processing the phase interferometric information of two complex numerical radar images of the same area.

[0074] S2. Obtain reflected wave data of abnormal surface deformation areas through ground penetrating radar; use UAVs to perform image detection and identification of abnormal crack data in abnormal surface deformation areas.

[0075] S3. Assess the risk of damage to underground facilities in areas with abnormal surface deformation based on deformation data, reflected wave data, and abnormal crack data.

[0076] The process of registering the latest SAR image and the initial SAR image to obtain the registered image involves the following steps:

[0077] S11. Obtain the first set of key features of the latest SAR image and the initial SAR image respectively by using the phase consistency value of each pixel; obtain the second set of key features of the latest SAR image and the initial SAR image respectively by using the gradient value of each pixel; obtain the third set of key features of the latest SAR image and the initial SAR image respectively by using the region discrete value corresponding to each pixel.

[0078] S12. The first intersection of the first set of key features, the second set of key features, and the third set of key features of the latest SAR image is obtained. The key features in the first intersection are the first true key features of the latest SAR image. The second intersection of the first set of key features, the second set of key features, and the third set of key features of the initial SAR image is obtained. The key features in the second intersection are the second true key features of the initial SAR image.

[0079] S13. Match each first real key point with the second real key point to obtain multiple matching point pairs. Use the matching point pairs to obtain the transformation matrix. Register the latest SAR image and the initial SAR image according to the transformation matrix to obtain the registered image.

[0080] In step S11, the first feature keypoint sets of the latest SAR image and the initial SAR image are obtained respectively through the phase consistency value of each pixel. Specifically, this is achieved through the following steps:

[0081] Sa1. Convolve the corresponding SAR image with a two-dimensional Log-Gabor filter to obtain the frequency domain amplitude of each pixel at each scale n and each direction o. and phase components , where i represents the i-th pixel;

[0082] Sa2. Calculate the phase consistency value of each pixel in different directions based on the frequency domain amplitude and phase components:

[0083] ;

[0084] This represents the phase consistency value of the i-th pixel in the SAR image along direction o. Let $\frac{i}{i}$ be the average phase value of the i-th pixel in the SAR image along direction $o$. For noise compensation in direction o, Here, C is the weighting factor in direction o, and C is a preset constant. This indicates a truncation operation with a lower bound of 0;

[0085] Sa3. Obtain the minimum moment value of each pixel using the moment analysis equation based on the phase consistency value:

[0086] ;

[0087] This represents the minimum moment value corresponding to the i-th pixel in the SAR image. This corresponds to the first moment component of the i-th pixel in the SAR image. This corresponds to the second moment component of the i-th pixel in the SAR image. This corresponds to the third moment component of the i-th pixel in the SAR image;

[0088] Sa4. Pixels with minimum moment values ​​greater than or equal to a preset threshold are taken as first feature keypoints, thereby forming the first feature keypoint set of the corresponding SAR image.

[0089] In step S11, the second feature keypoint set of the latest SAR image and the initial SAR image is obtained by using the gradient value of each pixel, specifically through the following steps:

[0090] Sb1, Calculate the gradient value of each pixel:

[0091] ;

[0092] in, This corresponds to the gradient value of the i-th pixel in the SAR image. This represents the gradient in the horizontal direction corresponding to the i-th pixel in the SAR image. This represents the gradient in the vertical direction corresponding to the i-th pixel in the SAR image. This represents the spatial derivative of the Gaussian function in the horizontal direction for the i-th pixel in the SAR image. is the spatial derivative of the i-th pixel in the corresponding SAR image in the vertical direction through the Gaussian function, * is the convolution operation, and S is the corresponding SAR image;

[0093] Sb2. Pixels with gradient values ​​greater than or equal to a preset gradient threshold are used as second feature keypoints to form the second feature keypoint set of the corresponding SAR image.

[0094] After obtaining key points based on phase consistency values, there are often many falsely detected key points around the real key points. These key points are similar to the feature descriptors of the real key points, which can lead to incorrect matching later. This embodiment adds a key point overlap screening and retention based on pixel gradient, which detects key points with obvious gradient changes. These points are usually located at the edges and corners of the image, thus retaining more real key points and improving the accuracy of matching.

[0095] In step S11, the sets of third feature key points of the latest SAR image and the initial SAR image are obtained respectively through the discrete values ​​of the regions corresponding to each pixel point, specifically:

[0096] Sc1: Based on the preset window size, each pixel-centered block will be obtained separately.

[0097] Sc2. Calculate the whole image discrete value and the regional discrete value of each block of the corresponding SAR image. The whole image discrete value is the ratio of the standard deviation of the pixel value to the mean of the pixel value of the corresponding SAR image. The regional discrete value is the ratio of the standard deviation of the pixel value to the mean of the pixel value of the corresponding block.

[0098] Sc3. If the regional discrete value of a block is greater than the overall discrete value of the corresponding SAR image, then the center pixel of the block is taken as the third feature key point, thus forming the third feature key point set of the corresponding SAR image.

[0099] The preset window size is the same as the preset sliding window size.

[0100] In SAR images, some areas with high scattering intensity, such as buildings, may have significant noise. These areas are often falsely detected key points during surface deformation detection. Therefore, in this embodiment, by dividing the area into blocks and calculating the discrete values ​​of the corresponding areas, the center pixel is retained as the key point based on its size. This greatly reduces the number of falsely detected key points and improves the matching accuracy of subsequent key point matching pairs.

[0101] In step S13, matching points are performed between each first real keypoint and the second real keypoint to obtain multiple matching point pairs, specifically:

[0102] Sa131, construct feature descriptors for each first and second real keypoint;

[0103] Sa132. Calculate the Euclidean distance between the feature descriptor of the first real key point and the feature descriptor of each second real key point, and select the second real key point with the smallest Euclidean distance to match the first real key point as the first matching point pair.

[0104] Sa133: Remove mismatched point pairs from the first matching point pairs using the RANSAC algorithm, leaving the second matching point pairs.

[0105] In step S13, the transformation matrix is ​​obtained using the matching point pairs, specifically through:

[0106] Sb131, Analyze the geometric distortion value G based on the second matching point pair:

[0107] ;

[0108] r is the number of the second matching point pairs. () represents the coordinates of the first real keypoint in the i-th second matching point pair. ) represents the coordinates of the second real keypoint in the i-th second matching point pair;

[0109] Sb132, Analysis of compensation coefficient BC based on geometric distortion value:

[0110] ;

[0111] Gmax is the preset maximum threshold for geometric distortion values;

[0112] Sb133. The transformation matrix parameters are obtained by combining the compensation coefficients and solving the least squares objective function. The expression of the least squares objective function is as follows:

[0113] ;

[0114] To minimize the objective function value, , , , and All are transformation matrix parameters;

[0115] Sb134. Construct the transformation matrix using the transformation matrix parameters.

[0116] In step S2, the UAV is used to perform image detection and identification of abnormal crack data in the abnormal deformation area of ​​the ground surface, including the determination of abnormal cracks. The determination steps are as follows:

[0117] S21. Use drones to acquire multiple sub-images of the corresponding abnormal surface deformation areas;

[0118] S22. Use an edge detection algorithm to obtain suspected cracks in the sub-image of the region, and then use a pre-built deep learning model to identify the first crack from the suspected cracks.

[0119] S23. The regional sub-images are stitched together to obtain a complete regional image. Based on the endpoint position of the first crack, the first crack connected to the endpoint of the current first crack is queried in the complete regional image and continuously stitched together to obtain the corresponding complete crack.

[0120] S24. Obtain the actual crack width of the complete crack and its pixel proportion in the corresponding complete area image. If the actual crack width is greater than or equal to a preset width threshold and / or the pixel proportion is greater than or equal to a preset proportion threshold, then it is an abnormal crack.

[0121] The edge detection algorithm can be the Canny edge detection algorithm. The Canny algorithm is designed based on a single-channel grayscale image. In order to reduce noise in the image and maintain the integrity of the edges, a Gaussian filter is used to smooth the grayscale image, which can effectively remove high-frequency noise in the image. Strong edge points are detected by a high threshold and weak edge points are detected by a low threshold. When the edges are connected, it is checked whether each weak edge point is adjacent to any strong edge point. If so, the weak edge point is retained. Therefore, the Canny algorithm can ensure the continuity and integrity of edge points, thereby producing clear and continuous edge contours.

[0122] The deep learning model can be either a U-Net model or a convolutional neural network model, and is trained using a large number of surface crack images as training samples.

[0123] In some embodiments, in step S3, the risk of damage to underground facilities in the corresponding abnormal surface deformation area is assessed based on deformation data, reflected wave data, and abnormal crack data. Specifically:

[0124] Obtain the pre-set BIM model of underground facilities in the target area, and obtain the facility parameters of underground facilities in the corresponding abnormal surface deformation area based on the BIM model;

[0125] The first risk coefficient F1 is estimated based on the deformation data and facility parameters using a pre-set first convolutional neural network.

[0126] The second risk coefficient F2 is estimated based on the reflected wave data and facility parameters using a pre-set second convolutional neural network.

[0127] The third risk coefficient F3 is estimated based on abnormal crack data and facility parameters using a pre-set third convolutional neural network.

[0128] Estimate the comprehensive damage risk value F of underground facilities in the corresponding abnormal surface deformation area, F=K1×F1+K2×F2+K3×F3, where K1, K2 and K3 are the first preset weight, the second preset weight and the third preset weight, respectively.

[0129] In another embodiment, the risk of damage to underground facilities in areas of abnormal surface deformation can be directly assessed manually by professionals based on deformation data, reflected wave data, and abnormal crack data.

[0130] The facility parameters include, but are not limited to, facility dimensions, structure, burial depth, and material properties.

[0131] The first convolutional network described above is trained using a large number of different facility parameters, historical deformation data, and corresponding assessed historical risk coefficients as training samples. Other convolutional networks are trained in the same way.

[0132] Example 2

[0133] See Figure 3 As shown, the present invention also provides an underground facility detection system based on an integrated air-space-ground system, which applies the underground facility detection method based on an integrated air-space-ground system as described in Embodiment 1, specifically including:

[0134] The central processing unit, and a first data acquisition module, a second data acquisition module, a third data acquisition module, and a risk assessment module that are communicatively connected to the central processing unit.

[0135] The first data acquisition module acquires the latest SAR image and the initial SAR image of the target detection area, registers the latest SAR image and the initial SAR image to obtain a registered image, and then uses InSAR technology to analyze the registered image to obtain the surface anomaly deformation area of ​​the target detection area and the corresponding deformation data.

[0136] The process of registering the latest SAR image and the initial SAR image to obtain the registered image involves the following steps:

[0137] The first set of key features for the latest SAR image and the initial SAR image is obtained by using the phase consistency value of each pixel; the second set of key features for the latest SAR image and the initial SAR image is obtained by using the gradient value of each pixel; and the third set of key features for the latest SAR image and the initial SAR image is obtained by using the region discrete value corresponding to each pixel.

[0138] The first intersection is obtained by intersecting the first set of key features, the second set of key features, and the third set of key features of the latest SAR image. The key features in the first intersection are the first true key features of the latest SAR image. The second intersection is obtained by intersecting the first set of key features, the second set of key features, and the third set of key features of the initial SAR image. The key features in the second intersection are the second true key features of the initial SAR image.

[0139] Multiple matching point pairs are obtained by matching each first real key point with the second real key point. The transformation matrix is ​​obtained by using the matching point pairs. The latest SAR image and the initial SAR image are registered according to the transformation matrix to obtain the registered image.

[0140] The second data acquisition module acquires reflected wave data from areas of abnormal surface deformation using ground penetrating radar.

[0141] The third data acquisition module uses drones to perform image detection and identification of abnormal crack data in abnormally deformed areas of the ground surface.

[0142] The risk assessment module evaluates the risk of damage to underground facilities in areas with abnormal surface deformation based on deformation data, reflected wave data, and abnormal crack data.

[0143] Example 3

[0144] The present invention also provides an electronic device, including: a processor, a transmitting device, an input device, an output device, and a memory. The processor may be implemented using a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit, or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application. The memory may be implemented using a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM), and is used to store computer program code. The computer program code includes computer instructions. When the processor executes the computer instructions, the electronic device executes a method as described in any of the above possible implementation methods.

[0145] Example 4

[0146] The present invention also provides a computer-readable storage medium storing a computer program, the computer program including program instructions, which, when executed by a processor of an electronic device, cause the processor to perform a method as described in any of the above possible implementations.

[0147] The beneficial effects of this invention are as follows:

[0148] This invention identifies anomalous surface deformation areas within the target detection region through SAR image analysis. For these areas, ground-penetrating radar and unmanned aerial vehicles (UAVs) are used to detect reflected wave data and anomalous crack data, respectively, avoiding blind detection and improving efficiency. The deformation data of the region is combined to comprehensively assess the risk of damage to underground facilities in the corresponding anomalous surface deformation areas, achieving multi-source data analysis from air, space, and ground to improve the accuracy of risk assessment. Simultaneously, in SAR image registration, the phase consistency value, gradient value, and corresponding regional discrete value of each pixel are combined for triple key point screening, improving registration accuracy and thus enhancing the accuracy of analyzing anomalous surface deformation areas, thereby increasing the precision of the key detection range for underground facilities.

[0149] In the description of this specification, the references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0150] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0151] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A method for detecting underground facilities based on space-air-ground integration, characterized in that, Includes the following steps: The latest SAR image and the initial SAR image of the target detection area are acquired. The latest SAR image and the initial SAR image are registered to obtain a registered image. Then, the surface anomaly deformation area and the corresponding deformation data of the target detection area are obtained by analyzing the registered image using InSAR technology. Ground-penetrating radar is used to acquire reflected wave data of abnormal surface deformation areas; UAVs are used to perform image detection and identification of abnormal crack data in abnormal surface deformation areas. The risk of damage to underground facilities in areas with abnormal surface deformation is assessed based on deformation data, reflected wave data, and abnormal crack data. The process of registering the latest SAR image and the initial SAR image to obtain the registered image involves the following steps: The first set of key features for the latest SAR image and the initial SAR image is obtained by using the phase consistency value of each pixel; the second set of key features for the latest SAR image and the initial SAR image is obtained by using the gradient value of each pixel; and the third set of key features for the latest SAR image and the initial SAR image is obtained by using the region discrete value corresponding to each pixel. The first intersection is obtained by intersecting the first set of key features, the second set of key features, and the third set of key features of the latest SAR image. The key features in the first intersection are the first true key features of the latest SAR image. The second intersection is obtained by intersecting the first set of key features, the second set of key features, and the third set of key features of the initial SAR image. The key features in the second intersection are the second true key features of the initial SAR image. Multiple matching point pairs are obtained by matching each first real key point with the second real key point. The transformation matrix is ​​obtained by using the matching point pairs. The latest SAR image and the initial SAR image are registered according to the transformation matrix to obtain the registered image. The process involves matching each first real keypoint with a second real keypoint to obtain multiple matching point pairs, specifically: Construct feature descriptors for each first and second real keypoint; Calculate the Euclidean distance between the feature descriptor of the first real keypoint and the feature descriptor of each second real keypoint, and select the second real keypoint with the smallest Euclidean distance to match the first real keypoint as the first matching point pair; The RANSAC algorithm is used to remove mismatched point pairs from the first matching point pairs, leaving the second matching point pairs; The process of obtaining the transformation matrix using matching point pairs is specifically achieved through: Based on the analysis of the second matching point pair, the geometric distortion value G is: ; r is the number of the second matching point pairs, () represents the coordinates of the first real keypoint in the i-th second matching point pair. () represents the coordinates of the second real keypoint in the i-th second matching point pair; Based on the analysis of geometric distortion values, the compensation coefficient BC is: Gmax is the preset maximum threshold for geometric distortion. The transformation matrix parameters are obtained by combining the compensation coefficients and solving the least squares objective function. The expression of the least squares objective function is as follows: ; To minimize the objective function value, , , , and All are transformation matrix parameters; The transformation matrix is ​​constructed by transforming the transformation matrix parameters.

2. The underground facility detection method based on air-space-ground integration according to claim 1, characterized in that, The process of obtaining the first set of key feature points for the latest SAR image and the initial SAR image by using the phase consistency value of each pixel is specifically carried out through the following steps: By convolving a two-dimensional Log-Gabor filter with the corresponding SAR image, the frequency domain amplitude of each pixel at each scale n and each direction o is obtained. and phase components , where i represents the i-th pixel; Calculate the phase consistency value of each pixel in different directions based on the frequency domain amplitude and phase components: The minimum moment value of each pixel is obtained through moment analysis equation based on the phase consistency value: Pixels with minimum moment values ​​greater than or equal to a preset threshold are selected as the first feature keypoints, thus forming the first feature keypoint set of the corresponding SAR image.

3. The underground facility detection method based on integrated air-space-ground communication as described in claim 1, characterized in that, The process of obtaining the second feature keypoint set of the latest SAR image and the initial SAR image by using the gradient values ​​of each pixel is specifically carried out through the following steps: Calculate the gradient value of each pixel: ； in, This corresponds to the gradient value of the i-th pixel in the SAR image. This represents the gradient in the horizontal direction corresponding to the i-th pixel in the SAR image. This represents the gradient in the vertical direction corresponding to the i-th pixel in the SAR image. This represents the spatial derivative of the Gaussian function in the horizontal direction for the i-th pixel in the SAR image. is the spatial derivative of the i-th pixel in the corresponding SAR image in the vertical direction through the Gaussian function, * is the convolution operation, and S is the corresponding SAR image; Pixels with gradient values ​​greater than or equal to a preset gradient threshold are used as second feature keypoints, thus forming the second feature keypoint set of the corresponding SAR image.

4. The method for detecting underground facilities based on integrated air-space-ground technology according to claim 1, characterized in that, The process of obtaining the third feature keypoint set of the latest SAR image and the initial SAR image by using the region discrete values ​​corresponding to each pixel is as follows: Based on the preset window size, blocks centered on each pixel will be obtained separately; Calculate the whole image discrete value and the regional discrete value of each block of the corresponding SAR image. The whole image discrete value is the ratio of the standard deviation of the pixel value to the mean of the pixel value of the corresponding SAR image, and the regional discrete value is the ratio of the standard deviation of the pixel value to the mean of the pixel value of the corresponding block. If the regional discrete value of a block is greater than the overall discrete value of the corresponding SAR image, then the center pixel of the block is taken as the third feature key point, thus forming the third feature key point set of the corresponding SAR image.

5. The method for detecting underground facilities based on integrated air-space-ground technology according to claim 1, characterized in that, The method involves using drones to detect and identify abnormal cracks in areas of abnormal surface deformation, including the determination of abnormal cracks. The determination steps are as follows: Multiple sub-images of corresponding abnormal surface deformation areas were obtained using drones; The edge detection algorithm is used to obtain suspected cracks in the sub-image of the region, and then a pre-built deep learning model is used to identify the first crack from the suspected cracks; The regional sub-images are stitched together to obtain a complete regional image. Based on the endpoint position of the first crack, the first crack connected to the endpoint of the current first crack is queried in the complete regional image and continuously stitched together to obtain the corresponding complete crack. Obtain the actual crack width and its pixel percentage in the corresponding complete area image. If the actual crack width is greater than or equal to a preset width threshold and / or the pixel percentage is greater than or equal to a preset percentage threshold, then it is an abnormal crack.

6. The method for detecting underground facilities based on integrated air-space-ground technology according to claim 1, characterized in that, The assessment of the risk of damage to underground facilities in areas with abnormal surface deformation based on deformation data, reflected wave data, and abnormal crack data specifically includes: Obtain the pre-set BIM model of underground facilities in the target area, and obtain the facility parameters of underground facilities in the corresponding abnormal surface deformation area based on the BIM model; The first risk coefficient F1 is estimated based on the deformation data and facility parameters using a pre-set first convolutional neural network. The second risk coefficient F2 is estimated based on the reflected wave data and facility parameters using a pre-set second convolutional neural network. The third risk coefficient F3 is estimated based on abnormal crack data and facility parameters using a pre-set third convolutional neural network. Estimate the comprehensive damage risk value F of underground facilities in the corresponding abnormal surface deformation area, F=K1×F1+K2×F2+K3×F3, where K1, K2 and K3 are the first preset weight, the second preset weight and the third preset weight, respectively.

7. A space-air-ground integrated underground facility detection system, employing the space-air-ground integrated underground facility detection method as described in any one of claims 1 to 6, characterized in that, include: The first data acquisition module acquires the latest SAR image and the initial SAR image of the target detection area, registers the latest SAR image and the initial SAR image to obtain a registered image, and then uses InSAR technology to analyze the registered image to obtain the surface anomaly deformation area of ​​the target detection area and the corresponding deformation data. The process of registering the latest SAR image and the initial SAR image to obtain the registered image involves the following steps: The first set of key features for the latest SAR image and the initial SAR image is obtained by using the phase consistency value of each pixel; the second set of key features for the latest SAR image and the initial SAR image is obtained by using the gradient value of each pixel; and the third set of key features for the latest SAR image and the initial SAR image is obtained by using the region discrete value corresponding to each pixel. The first intersection is obtained by intersecting the first set of key features, the second set of key features, and the third set of key features of the latest SAR image. The key features in the first intersection are the first true key features of the latest SAR image. The second intersection is obtained by intersecting the first set of key features, the second set of key features, and the third set of key features of the initial SAR image. The key features in the second intersection are the second true key features of the initial SAR image. Multiple matching point pairs are obtained by matching each first real key point with the second real key point. The transformation matrix is ​​obtained by using the matching point pairs. The latest SAR image and the initial SAR image are registered according to the transformation matrix to obtain the registered image. The second data acquisition module acquires reflected wave data from areas of abnormal surface deformation using ground penetrating radar. The third data acquisition module uses drones to perform image detection and identification of abnormal crack data in abnormally deformed areas of the ground surface. The risk assessment module assesses the risk of damage to underground facilities in areas with abnormal surface deformation based on deformation data, reflected wave data, and abnormal crack data. The process involves matching each first real keypoint with a second real keypoint to obtain multiple matching point pairs, specifically: Construct feature descriptors for each first and second real keypoint; Calculate the Euclidean distance between the feature descriptor of the first real keypoint and the feature descriptor of each second real keypoint, and select the second real keypoint with the smallest Euclidean distance to match the first real keypoint as the first matching point pair; The RANSAC algorithm is used to remove mismatched point pairs from the first matching point pairs, leaving the second matching point pairs; The process of obtaining the transformation matrix using matching point pairs is specifically achieved through: Based on the analysis of the second matching point pair, the geometric distortion value G is: ; r is the number of the second matching point pairs, () represents the coordinates of the first real keypoint in the i-th second matching point pair. () represents the coordinates of the second real keypoint in the i-th second matching point pair; Based on the analysis of geometric distortion values, the compensation coefficient BC is: Gmax is the preset maximum threshold for geometric distortion. The transformation matrix parameters are obtained by combining the compensation coefficients and solving the least squares objective function. The expression of the least squares objective function is as follows: ; To minimize the objective function value, , , , and All are transformation matrix parameters; The transformation matrix is ​​constructed by transforming the transformation matrix parameters.

Images