Dam hidden danger three-dimensional imaging method and equipment based on two-dimensional ground penetrating radar and medium

Abstract

The invention relates to the field of geophysical exploration, and discloses a dam hidden danger three-dimensional imaging method and device based on a two-dimensional ground penetrating radar and a medium, and the method comprises the steps: selecting the two-dimensional ground penetrating radar with a proper frequency to be externally connected with an RTK, and determining a to-be-measured region; preprocessing the data acquired by the two-dimensional ground penetrating radar to obtain preprocessed data; it needs to be explained that for part of data with incomplete longitude and latitude data caused by RTK signal loss, longitude and latitude complementation is carried out through linear interpolation, then measured A scanning data of the part is mapped to a three-dimensional space based on complemented longitude and latitude, and preliminary three-dimensional data is obtained; adopting a multi-stage difference strategy for the preliminary three-dimensional data to obtain smooth three-dimensional data; and determining a dam potential safety hazard area by checking different slices of the smooth three-dimensional data. According to the method, the detection data quality is improved, the detection efficiency and the detection precision are improved, and the equipment use cost is reduced.

Description

Technical Field

[0001] The present invention relates to the field of geophysical exploration, and particularly to a three-dimensional imaging method, device and medium for hidden dangers of dikes based on two-dimensional ground penetrating radar. Background Art

[0002] Termite nests, badger holes, cavities, leakage channels, etc. existing on dikes endanger the safety of dikes. Due to the obvious difference in dielectric constant between hidden dangers such as termite nests, badger holes, cavities, leakage channels, etc. and the surrounding soil, ground penetrating radar can be used to detect and discover these safety hazards existing on dikes.

[0003] As a geophysical exploration technology, ground penetrating radar is mainly based on the principles of electromagnetic wave propagation, reflection and diffraction. When detecting a dike, electromagnetic waves are emitted. When the propagating electromagnetic waves encounter media with different electrical structures, reflection will occur. The waveform, amplitude and arrival time of the reflected signal will vary due to different surrounding rock media and geological targets. By analyzing the attribute differences of these reflected signals, the location and depth of the hidden danger can be determined, and then remedial measures such as excavation and grouting can be carried out.

[0004] In the detection of hidden dangers of dikes, the penetration ability of ground penetrating radar is crucial. Compared with three-dimensional radar, two-dimensional radar has a greater detection depth. Especially at the same working frequency, two-dimensional radar can obtain deeper underground information. However, since two-dimensional radar only provides a single cross-sectional image, in the face of complex geological environments and hidden dangers (such as cavities, leakage, termite nests, etc.), two-dimensional radar data often lacks intuitiveness and comprehensiveness, which is not conducive to the accurate identification of hidden dangers. In contrast, three-dimensional radar data can more intuitively display the underground structure, which helps to improve the ability to identify abnormal features.

[0005] However, current three-dimensional ground penetrating radar devices are usually bulky and expensive, which is not conducive to large-scale promotion. And at the same working frequency, their detection depth is shallower than that of two-dimensional radar, making it difficult to meet the demand for detecting deep hidden dangers in environments with higher humidity such as dikes. Therefore, how to combine the depth advantage of two-dimensional ground penetrating radar with the visualization advantage of three-dimensional data, and use two-dimensional radar data to generate high-precision three-dimensional data to improve the accuracy of hidden danger identification has become an urgent problem to be solved in the current technical field.

[0006] The existing three-dimensional imaging technologies for ground penetrating radar data have the following significant technical bottlenecks:

[0007] First, the spatial mapping method is not clear. For example, Patent CN117572410A does not clearly explain how to map the two-dimensional data (along the road direction) of the left, middle, and right wheel track bands' survey lines to the X, Y, and Z-axis coordinates in three-dimensional space. For example, the method does not define how the lateral spacing of the survey lines (such as the distances between the left and right wheel track bands and the road center line) is converted into the Y-axis coordinates of the three-dimensional data volume. Only based on the survey point order (trace number) and fixed grid division, a three-dimensional data model is generated without considering the curvature of the road or the uneven distribution of the survey lines, which may lead to deviations between the generated three-dimensional data volume and the real geographical space.

[0008] Second, the existing methods are highly dependent on survey lines, restricting their application in irregular scenarios. For example, Patent CN110764082A uses a method of collecting data at fixed intervals or strictly according to lane markings, making it difficult to adapt to unstructured paths (such as the arc-shaped detection route on the slope) in complex environments like dams. The fixed and standardized survey line layout method not only reduces the flexibility of data collection but also leads to a decrease in data collection efficiency.

[0009] Third, the accuracy of the interpolation algorithm has limitations. Most of the existing methods use a single and mechanical linear interpolation (CN110764082A), lacking a theoretical basis based on the electromagnetic wave propagation characteristics or geological continuity, which may introduce false signals or over-smooth real anomalies.

[0010] Fourth, the existing methods rely on three-dimensional radar equipment, increasing the system cost and the usage threshold. For example, Patent CN117630915A relies on a multi-channel antenna array (such as 16 channels) to simultaneously transmit and receive multiple electromagnetic wave signals to form a high-density three-dimensional data volume. However, the three-dimensional ground penetrating radar is bulky and not very suitable for detecting steep areas such as dams, and its detection depth is shallower compared to the two-dimensional ground penetrating radar of the same frequency. Summary of the Invention

[0011] The object of the present invention is to propose a three-dimensional imaging method, device, and medium for dam hidden danger based on a two-dimensional ground penetrating radar, to solve the technical problems of low accuracy and low precision existing in the current three-dimensional imaging of ground penetrating radar.

[0012] Specifically, a three-dimensional imaging method, device, and medium for dam hidden danger based on a two-dimensional ground penetrating radar provided by the present invention, the method includes the following steps:

[0013] S1. Select a two-dimensional ground penetrating radar with a suitable frequency and externally connect RTK to determine the area to be measured;

[0014] S2. Preprocess the data collected by the two-dimensional ground penetrating radar to obtain the preprocessed data;

[0015] S3. For the missing longitude and latitude data, linear interpolation is used for completion. Then, the A-scan is mapped to the three-dimensional space according to the completed longitude and latitude to obtain the preliminary three-dimensional data;

[0016] S4. A multi-stage interpolation strategy is adopted for the preliminary three-dimensional data to obtain the smooth three-dimensional data;

[0017] S5. By viewing different slices of the smooth three-dimensional data, the dam safety hazard area is determined.

[0018] A storage medium stores instructions and data for implementing a three-dimensional imaging method for dam hazards based on a two-dimensional ground penetrating radar.

[0019] A three-dimensional imaging device for dam hazards based on a two-dimensional ground penetrating radar includes: a processor and the storage medium; the processor loads and executes the instructions and data in the storage medium for implementing a three-dimensional imaging method for dam hazards based on a two-dimensional ground penetrating radar.

[0020] The beneficial effects provided by the present invention are as follows:

[0021] 1. By performing interpolation processing on the incomplete geographical coordinates, the problem of discontinuous original coordinate data is effectively solved, providing a complete geographical coordinate benchmark for subsequent three-dimensional space reconstruction;

[0022] 2. By means of a spatial expansion strategy, edge data loss is avoided. The spatial recombination of discrete measurement points is realized by using a regularized grid, and the exact correspondence between the original data and the geographical coordinates is maintained through nearest neighbor matching, providing standardized input data for subsequent three-dimensional geological model reconstruction;

[0023] 3. By fusing morphological operations and an adaptive interpolation algorithm, the interpolation strategy is dynamically adjusted to balance data fidelity and computational efficiency. The local processing range is optimized by using the minimum bounding rectangle, and the interpolation operation is ensured to be executed only within the effective geographical coverage area through a composite mask constraint, effectively improving the spatial continuity of the three-dimensional geological model.

[0024] The above contents work together to improve the quality of detection data, enhance the detection efficiency, detection accuracy, and reduce the equipment usage cost. Description of the Drawings

[0025] Figure 1 It is a simple flow schematic diagram of the method of the present invention;

[0026] Figure 2 It is a schematic diagram of demarcating the survey area and selecting equipment for measurement;

[0027] Figure 3 It is a schematic diagram of the original unprocessed two-dimensional data;

[0028] Figure 4It is a schematic diagram of two-dimensional data after preprocessing;

[0029] Figure 5 It is a schematic diagram of longitude and latitude coordinates before interpolation;

[0030] Figure 6 It is a schematic diagram of longitude and latitude coordinates after interpolation and completion;

[0031] Figure 7 It is a schematic diagram of mapping two-dimensional data to three-dimensional space according to longitude and latitude coordinates;

[0032] Figure 8 It is a schematic diagram of a three-dimensional data volume after interpolation using a multi-stage interpolation strategy;

[0033] Figure 9 It is a schematic diagram of a slice where an anomaly is found in the three-dimensional data volume;

[0034] Figure 10 It is a schematic diagram of an ant nest excavated according to the anomaly in the data;

[0035] Figure 11 It is a schematic diagram of the working of the hardware device in the embodiment of the present invention. Detailed implementation manners

[0036] To make the objectives, technical solutions and advantages of the present invention clearer, the embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0037] Before formally elaborating on the present invention, first a general description of the solution of the present invention is given for easy understanding.

[0038] Please refer to Figure 1 , a three-dimensional imaging method for hidden dangers of dams based on two-dimensional ground penetrating radar provided by the present invention includes:

[0039] S1. Connect an RTK to a two-dimensional ground penetrating radar with a suitable frequency to determine the area to be measured;

[0040] It should be noted that in the present invention, an RTK is connected to a two-dimensional ground penetrating radar to provide accurate longitude and latitude coordinates for detecting the area to be measured. This solution does not need to strictly follow a straight path for measurement, which not only improves the data acquisition efficiency but also ensures the presentation effect of the measurement results.

[0041] Specifically, select a suitable ground penetrating radar and determine the area to be detected. For a dam structure composed of highly water-saturated clay, since the electromagnetic wave attenuation is relatively large and the detection depth of the ground penetrating radar is limited, it is advisable to select a medium-low frequency ground penetrating radar to improve the detection effect. The determination of the area to be detected can be flexibly adjusted according to the detection target. For the detection of large-scale defects such as dam leakage and cavities, large-area measurement can be carried out without presetting a specific detection area; for the detection of local defects such as termite nests and badger holes, a smaller area to be detected can be selected based on the activity traces of termites and badgers to improve the data acquisition efficiency and reduce unnecessary measurement workload.

[0042] S2. Preprocess the data collected by the two-dimensional ground penetrating radar to obtain the preprocessed data;

[0043] It should be noted that the preprocessing described in step S2 includes: direct wave removal, background removal, and gain processing.

[0044] As an example, the ground penetrating radar data collected are sequentially subjected to preprocessing operations such as direct wave removal, background removal, and gain processing to optimize the data quality. Among them, direct wave removal can reduce the direct coupling signal between the transmission and reception of the ground penetrating radar and improve the recognizability of shallow targets; background removal can eliminate the background noise caused by terrain undulation or equipment system errors and improve the contrast of abnormal features; gain processing can perform dynamic enhancement on signals at different depths, compensate for the electromagnetic wave attenuation, and improve the visualization effect of deep targets. Through the above data processing means, the useful signals in the data are effectively enhanced, the interference of environmental noise and irrelevant signals is reduced, and higher-quality input data is provided for the subsequent conversion of two-dimensional data to three-dimensional visualization.

[0045] S3. Use linear interpolation to complete the missing longitude and latitude data, and then map the A-scan to the three-dimensional space according to the completed longitude and latitude to obtain the preliminary three-dimensional data;

[0046] It should be noted that the linear interpolation in step S3 is as follows:

[0047] S31. Create an initial longitude array and latitude array based on the preprocessed data and establish a corresponding relationship with the corresponding A-scan data;

[0048] S32. Use the linear interpolation algorithm to supplement the missing values of the A-scan data with missing longitude and latitude, so that each A-scan data has a corresponding longitude and latitude coordinate.

[0049] As an example, first create an array containing the initial longitude array and latitude array, whose dimensions match the total number of A-scans. Read the longitude and latitude data of each A-scan in the coordinate file. Since RTK uses time triggering, the acquisition frequency of its longitude and latitude coordinates is lower than the frequency of the ground penetrating radar to acquire data for each A-scan. Therefore, there will be some A-scans without corresponding longitude and latitude coordinates. Finally, perform missing value filling on the A-scans with missing longitude and latitude respectively based on the linear interpolation algorithm to generate interpolated longitude and latitude information with complete geographic coordinate information. This method effectively solves the problem of discontinuous original coordinate data and provides a complete geographic coordinate benchmark for subsequent three-dimensional space reconstruction.

[0050] It should be noted that in step S3, the process of obtaining the preliminary three-dimensional data is as follows:

[0051] S33. Perform boundary extension processing on the A-scan data with complete geographic coordinate information to form a three-dimensional reconstruction spatial domain including a buffer area;

[0052] S34. Establish a regularized geographic coordinate system with uniform distribution in the three-dimensional reconstruction spatial domain according to the preset grid resolution, and construct a two-dimensional geographic grid matrix;

[0053] S35. For each A-scan data unit, perform spatial matching and positioning in the two-dimensional geographic grid matrix based on its original geographic coordinates to obtain the three-dimensional volume data with spatial continuity corresponding to each A-scan data unit, that is, the preliminary three-dimensional data.

[0054] As an example, first read the pre-stored longitude and latitude coordinate matrix and the radar A-scan data set, extract the longitude and latitude parameters of each measurement point, and determine its spatial distribution range. To optimize the spatial mapping effect, perform boundary extension processing on the original longitude and latitude range, and extend it outward using an offset of 5% of the original extreme value to form a three-dimensional reconstruction spatial domain including a buffer area. Subsequently, according to the preset grid resolution (such as a 100×100 longitude and latitude grid), establish a regularized geographic coordinate system with uniform distribution in the extended spatial domain, generate an array of equally spaced latitude lines and an array of longitude lines, and construct a two-dimensional geographic grid matrix through Cartesian product operation.

[0055] For each A-scan data unit, perform spatial matching and positioning based on its original geographic coordinates.

[0056] The nearest neighbor interpolation algorithm is used to determine the belonging position of the measurement point in the regular grid: calculate the absolute deviation of the latitude value of the point from the latitude line array respectively, and select the latitude index corresponding to the minimum deviation; similarly, obtain the longitude index, so as to determine the spatial grid coordinates corresponding to the measurement point in the three-dimensional data volume. Map the time-domain waveform data of each A-scan completely to the corresponding position of the three-dimensional array, and finally construct a three-dimensional volume data with spatial continuity, and its three dimensions respectively represent geographical latitude, geographical longitude and radar detection depth.

[0057] The feature of this method is to avoid edge data loss through a spatial expansion strategy, realize the spatial reorganization of discrete measurement points by using a regular grid, and maintain the exact corresponding relationship between the original data and geographical coordinates through nearest neighbor matching, providing standardized input data for subsequent three-dimensional geological model reconstruction.

[0058] S4. Adopt a multi-stage interpolation strategy for the preliminary three-dimensional data to obtain smooth three-dimensional data;

[0059] It should be noted that step S4 is specifically as follows:

[0060] S41. Perform binary mask processing on the preliminary three-dimensional data to identify all non-zero regions and form an initial valid region;

[0061] S42. Perform morphological processing on the initial valid region to obtain a locally processed region;

[0062] S43. Dynamically select an interpolation algorithm for dynamic interpolation according to the number of valid data points in the locally processed region to eliminate data holes;

[0063] S44. Generate smooth three-dimensional data according to the region after dynamic interpolation.

[0064] As an embodiment, first perform a binary mask extraction operation on the input three-dimensional data slice to identify all non-zero data regions and form an initial valid region.

[0065] To improve the connectivity of discrete non-zero regions, a circular structural kernel with a preset radius (such as 5 pixels) is used to perform binary morphological closing operation to bridge the gaps between adjacent non-zero regions and generate an optimized global mask. Subsequently, the minimum bounding rectangle containing all non-zero regions is determined through row and column projection analysis, and the sub-matrix within this rectangle is extracted as the locally processed region to reduce computational redundancy.

[0066] For the valid data points within the sub - matrix, extract their coordinates and corresponding values as the interpolation reference point set. Dynamically select the interpolation algorithm according to the number of reference points: when the number of valid points is lower than the cubic interpolation threshold, degrade to linear interpolation; when it does not meet the linear interpolation condition, use nearest - neighbor interpolation to ensure adaptability in different data density scenarios. Use the grid interpolation algorithm to generate a continuous surface throughout the sub - matrix, and preferentially use the selected interpolation method to cover the regular grid nodes. For the remaining invalid value areas after interpolation, further apply nearest - neighbor interpolation for global filling to eliminate data holes.

[0067] Through the composite mask control mechanism, only allow data replacement operations to be performed within the coverage range of the original non - zero area, filling the zero - value area while retaining the original values of the valid data. After completing local interpolation, write the processed sub - matrix back to the corresponding spatial position in the original three - dimensional data. Iteratively perform the above operations on each layer slice of the three - dimensional data to finally generate three - dimensional volume data with spatial continuity, eliminating zero - value gaps caused by uneven distribution of measurement points or data loss.

[0068] The innovative feature of this method lies in the integration of morphological operations and adaptive interpolation algorithms, dynamically adjusting the interpolation strategy to balance data fidelity and computational efficiency, using the minimum bounding rectangle to optimize the local processing range, and ensuring that the interpolation operation is only performed within the valid geographical coverage area through composite mask constraints, effectively improving the spatial continuity of the three - dimensional geological model.

[0069] S5. Determine the dam safety hazard area by viewing different slices of the smoothed three - dimensional data.

[0070] The present invention checks the data of different depth layers one by one from top to bottom by viewing the generated three - dimensional data slices layer by layer. When a continuous shadow or abnormal signal is found at a certain position in several consecutive depth layers, it can be determined that there is a hazard at that position. This method can effectively locate potential hazard areas through inter - layer comparison and signal anomaly recognition.

[0071] As an embodiment, the following is a specific case of the application of this method in termite control of dams.

[0072] Traces of termite activities were found in a certain section of the Yangtze River dam in Wuhan. Based on this, a 5m×5m area was demarcated for detection, and a 170M, 600M dual - frequency two - dimensional ground - penetrating radar was selected for data collection, as shown below. Figure 2 By connecting the radar with RTK, accurate longitude and latitude coordinates can be provided for each A - scan. During the detection process, drag the ground - penetrating radar to walk along the entire area to be measured to ensure complete coverage of the area.

[0073] For the collected two-dimensional ground penetrating radar (GPR) raw data, preprocessing operations such as direct wave removal, background removal, and gain processing are first performed to improve the signal-to-noise ratio of the data and prepare for subsequent three-dimensional high-quality imaging. The two-dimensional data before and after processing is as shown in Figure 3 , Figure 4 . Then, the longitude and latitude coordinates corresponding to all A-scan data are read, and interpolation processing is performed on the missing coordinates. The effects before and after longitude and latitude interpolation are as shown in Figure 5 , Figure 6 . All A-scan data is projected into three-dimensional space based on its longitude and latitude coordinates. The mapped data is as shown in Figure 7 , and interpolation processing is performed on the three-dimensional data through a multi-stage interpolation strategy to improve the continuity of the data. The three-dimensional data after multi-stage strategy interpolation is as shown in Figure 8 .

[0074] When viewing all slices, an anomaly is found in the top-down slice of the 150th layer sampling points of the three-dimensional GPR data, as shown in Figure 9 . Combining the experiment, the dielectric constant of the dam soil at this location is obtained, and the depth of the anomaly location is estimated to be about 1.5 m. According to the coordinates of this location in the slice, excavation is carried out at the corresponding location, and a main termite nest with a diameter of about 50 cm is successfully discovered, as shown in Figure 10 .

[0075] In some other embodiments, the technical solution in the present invention is essentially to splice two-dimensional GPR data into three-dimensional data. Therefore, the present invention can be applied not only to the above-mentioned dam hidden danger imaging detection field, but also to other detection fields including road detection, with a wide range of applications.

[0076] Please refer to Figure 11 , Figure 11 , which is a schematic diagram of the operation of the hardware device in the embodiment of the present invention. The hardware device specifically includes: a three-dimensional imaging device 401 for dam hidden dangers based on a two-dimensional GPR, a processor 402, and a storage medium 403.

[0077] A three-dimensional imaging device 401 for dam hidden dangers based on a two-dimensional GPR: The three-dimensional imaging device 401 for dam hidden dangers based on a two-dimensional GPR implements the three-dimensional imaging method for dam hidden dangers based on a two-dimensional GPR.

[0078] Processor 402: The processor 402 loads and executes the instructions and data in the storage medium 403 to implement the three-dimensional imaging method for dam hidden dangers based on a two-dimensional GPR.

[0079] Storage medium 403: The storage medium 403 stores instructions and data; the storage medium 403 is used to implement the three-dimensional imaging method for dam hidden dangers based on a two-dimensional GPR.

[0080] The beneficial effects of the present invention are as follows:

[0081] 1. By interpolating incomplete geographical coordinates, the problem of discontinuous original coordinate data is effectively solved, providing a complete geographical coordinate benchmark for subsequent three-dimensional space reconstruction;

[0082] 2. By means of a spatial expansion strategy, edge data loss is avoided. The spatial recombination of discrete measurement points is realized by using a regularized grid, and the exact correspondence between the original data and geographical coordinates is maintained through nearest neighbor matching, providing standardized input data for subsequent three-dimensional geological model reconstruction;

[0083] 3. By integrating morphological operations and an adaptive interpolation algorithm, the interpolation strategy is dynamically adjusted to balance data fidelity and computational efficiency. The local processing range is optimized using the minimum bounding rectangle, and the interpolation operation is ensured to be performed only within the effective geographical coverage area through a composite mask constraint, effectively improving the spatial continuity of the three-dimensional geological model.

[0084] The above-mentioned contents work together to improve the quality of detection data, enhance the detection efficiency and accuracy, and reduce the equipment usage cost.

[0085] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A three-dimensional imaging method for hidden dangers of dikes based on two-dimensional ground penetrating radar, characterized in that: It includes the following steps: S1. Connect an RTK to a two-dimensional ground penetrating radar with a suitable frequency and determine the area to be measured; S2. Preprocess the data collected by the two-dimensional ground penetrating radar to obtain preprocessed data; S3. Complement the missing longitude and latitude data by linear interpolation, and then map the A-scan to the three-dimensional space according to the complemented longitude and latitude to obtain preliminary three-dimensional data; S4. Adopt a multi-stage interpolation strategy for the preliminary three-dimensional data to obtain smooth three-dimensional data; S5. Determine the dam safety hazard area by viewing different slices of the smooth three-dimensional data.

2. The three-dimensional imaging method for hidden dangers of dikes based on two-dimensional ground penetrating radar according to claim 1, wherein: The preprocessing described in step S2 includes: direct wave removal, background removal, and gain processing.

3. A three-dimensional imaging method for hidden dangers of dikes based on two-dimensional ground penetrating radar according to claim 1, characterized in that: The specific linear interpolation in step S3 is as follows: S31. Create an initial longitude array and a latitude array according to the preprocessed data, and establish a corresponding relationship with the corresponding A-scan data; S32. Use a linear interpolation algorithm to supplement the missing values of the A-scan data with missing longitude and latitude, so that each A-scan data has corresponding longitude and latitude coordinates.

4. The three-dimensional imaging method for hidden dangers of dikes based on a two-dimensional ground penetrating radar according to claim 3, wherein: In step S3, the process of obtaining the preliminary three-dimensional data is as follows: S33. Perform boundary expansion processing on the A-scan data with complete geographic coordinate information to form a three-dimensional reconstruction spatial domain including a buffer area; S34. Establish a regular geographic coordinate system with uniform distribution in the three-dimensional reconstruction spatial domain according to the preset grid resolution, and construct a two-dimensional geographic grid matrix; S35. For each A-scan data unit, perform spatial matching and positioning in the two-dimensional geographic grid matrix based on its original geographic coordinates to obtain the three-dimensional volume data with spatial continuity corresponding to each A-scan data unit, that is, the preliminary three-dimensional data.

5. A three-dimensional imaging method for hidden dangers of dikes based on two-dimensional ground penetrating radar according to claim 1, characterized in that: Step S4 is specifically as follows: S41. Perform binary masking on the preliminary three-dimensional data to identify all non-zero regions and form an initial effective region; S42. Perform morphological processing on the initial effective region to obtain a locally processed region; S43. Dynamically select an interpolation algorithm for dynamic interpolation according to the number of valid data points in the locally processed region to eliminate data holes; S44. Generate smooth three-dimensional data according to the region after dynamic interpolation.

6. The three-dimensional imaging method for hidden dangers of dikes based on two-dimensional ground penetrating radar according to claim 5, wherein: The process of dynamically selecting an interpolation algorithm in step S43 is: When the number of valid data points is lower than the cubic interpolation threshold, it degrades to linear interpolation. When it does not meet the linear interpolation condition, nearest neighbor interpolation is used; the grid interpolation algorithm is used to generate a continuous surface throughout the locally processed region, and the selected interpolation method is preferentially used to cover the regular grid nodes; for the remaining invalid value regions after interpolation, nearest neighbor interpolation is further applied for global filling.

7. The three-dimensional imaging method for hidden dangers of dikes based on two-dimensional ground penetrating radar according to claim 6, characterized in that: The method can also be applied to other target detection fields including road defect detection.

8. A storage medium, characterized in that: The storage medium stores instructions and data for implementing a three-dimensional imaging method for dam hidden dangers based on a two-dimensional ground penetrating radar according to any one of claims 1 to 7.

9. A three-dimensional imaging device for dam hidden dangers based on a two-dimensional ground penetrating radar, characterized in that: It includes: A processor and a storage medium; the processor loads and executes the instructions and data in the storage medium for implementing a three-dimensional imaging method for dam hidden dangers based on a two-dimensional ground penetrating radar according to any one of claims 1 to 7.

Images